JP4588324B2 - Tissue measurement probe - Google Patents

Description

  This invention was supported in whole or in part by grants P41-RR-02594 and R01-HL-64675 from the National Institute of Health. The US government has certain rights in this invention.

  This application is a continuation-in-part of US Ser. No. 10 / 178,062, filed Jun. 21, 2002, which claims the benefit of US Provisional Application No. 60 / 370,197, filed Apr. 5, 2002. This is a continuation-in-part application. The entire contents of the above application are incorporated herein by reference in their entirety.

  Optical methods are increasingly being used to detect diseases. In particular, near-infrared Raman spectroscopy has proven to be a useful tool for both disease diagnosis and disease progression studies because of its chemical specificity. Over the past decade, Raman spectroscopy has been applied to many diseases and biological problems, and great progress has been made in vitro. More recently, there have been reports of in vivo studies, however, they are limited to studies of skin or other readily accessible organs or are unreasonably long for practical clinical use Either using fiber optic configurations that require collection time. The majority of applications require remote sampling via optical fibers, and the size of the probe and fiber bundle is strictly limited by the application. One particular example that current commercial systems cannot provide is the ability to assess in vivo atherosclerotic lesions in real time with angiographic catheters and thus assist cardiologists in directing the most appropriate treatment in each individual case It is. These objectives are not met by the current system.

  In addition, prior art probes for remote Raman detection using several different filtering methods of the fiber's spectral background represent very low optical throughput or are too bulky for use in blood vessels Either. One problem with prior art designs involves having a 4 cm long rigid tip that prevents their incorporation into a percutaneous catheter to reach the coronary artery. Secondly, in the data collected using these probes, a significant portion of the fiber Raman spectrum still remains. In addition, a 30 second or longer order data collection time is typically required for the collection of signals with an acceptable signal-to-noise ratio (SNR).

  For example, there remains a need for improved systems and methods that include probes for Raman spectroscopy that are sized for medical applications and provide improved spectral signatures from tissues. .

  The systems and methods of the present invention relate to the use of spectroscopy, such as Raman spectroscopic methods, for the diagnosis of tissue conditions such as vascular disease or cancer. The system and method of the present invention for several applications: optical breast biopsy and breast analysis by ductoscopy, percutaneous blood analysis and monitoring, vascular stenosis, assessment of gastrointestinal cancer, dysplasia in the pancreatic duct With scanning as well as skin analysis.

  According to a preferred embodiment of the present invention, the tissue measurement system includes a proximal end, a distal end and a fiber optic probe having a diameter of 2 mm or less. This small diameter allows the system to be used for the diagnosis of coronary artery disease or other small lumens or soft tissues with minimal trauma. The delivery optical fiber (s) is included in a probe connected to a light source at the proximal end. A delivery fiber filter is included at the distal end. The system includes collection optical fiber (s) in a probe that collects Raman scattered radiation from the tissue, and the collection optical fiber is connected to a detector at the proximal end. The second filter is located at the distal end of the collection fiber. An optical lens system including a delivery waveguide connected to the delivery fiber, a collection waveguide connected to the collection optical fiber and a lens is disposed at the distal end of the probe.

  The delivery waveguide comprises a rod and the collection waveguide comprises a cylindrical tube that is concentric around the rod. In an alternative preferred embodiment, the delivery waveguide comprises a first tube and the collection waveguide comprises a second cylindrical tube that is concentric around the first tube. In addition, the lens includes a ball lens optically connected to the delivery fiber and the collection fiber.

  In a preferred embodiment, the probe further comprises a sleeve that optically isolates the delivery waveguide from the collection waveguide. The sleeve can be a metal such as palladium, silver or gold. The glass rod tube and sleeve can be attached together with an adhesive. An outer retaining sleeve can attach the distal optics to the optical fiber.

  The probe further includes a first plurality of collection fibers disposed concentrically around the delivery fiber at a first radius, and a concentric circle disposed at a second radius greater than the first radius around the delivery fiber. A second plurality of collection fibers formed.

  According to another aspect of the invention, the probe includes a controller that gates an acquisition time that is less than 2 seconds. In one aspect, the optical lens system has a length of less than 10 mm. In a preferred embodiment, the optical lens system has a length of less than 4 mm. The diameter of the distal optics is preferably in the range of 1-2 mm. The optical lens system delivers and collects radiation in a radial direction (which can be defined as any off-axis direction). The light source has a wavelength longer than 750 mm, and one preferred embodiment uses an argon laser pumped Ti: sapphire laser emitting at 830 nm. In an alternative embodiment, a diode laser such as an InGaAs laser emitting at 785 nm or 830 nm may be used.

  In a preferred embodiment, the radial Raman probe of the present invention for use in the diagnosis of atherosclerosis is incorporated into a catheter of the type used, for example, for angiography. It includes a balloon to replace blood and other fluids and place the catheter in the artery. One preferred embodiment includes a channel for balloon inflation. Furthermore, the catheter system includes the ability to flush blood temporarily with a liquid, such as physiological saline. One or several optical fibers may be configured to direct the excitation light in a radial direction, either to the side or at an angle ranging from 45 ° to 90 °. In one such embodiment, a side-positioned balloon is used to bring the fiber into contact with the arterial wall and evict blood or other intervening fluid.

  Alternatively, the delivery fiber may be positioned to direct light into a circular pattern at an angle with respect to the probe axis. Various collection fibers collect light simultaneously from various parts of the illuminated surrounding area. In this embodiment, the probe is enclosed in an inflatable balloon that is inflated to replace blood and other liquids prior to light delivery and / or collection. In a preferred embodiment, the balloon is of the type used in arterial applications such as angiography and the excitation light is passed through the arterial wall and the Raman light generated at the arterial wall is passed through the balloon. Made of thin material so that it can be returned to the collection fiber.

  The present invention encompasses the diagnostic classification of atherosclerotic plaques in human coronary arteries by quantitative assessment of their morphological composition using Raman spectroscopy. The rapid and non-destructive nature of Raman spectroscopy provides an opportunity to diagnose coronary plaques in vivo when applied in clinical settings using fiber optic technology. When so used, preferred embodiments of the present invention classify atherosclerotic lesions, and foam cells (FC), necrotic centers (which can be used to assess plaque instability and the extent of disease progression. NC) and its morphological features such as the presence of cholesterol crystals (CC), and thereby provide an in vivo quantitative assessment of the risk of life-threatening complications such as thrombosis and acute plaque bleeding. When so used, the methods of the present invention can provide insights into the dynamics that have so far been poorly understood in the development of atherosclerotic lesions and the effects of lipid lowering and other therapies .

  Chemical composition and morphology rather than anatomy (the degree of stenosis) determine atherosclerotic plaque instability and predict disease progression. In a preferred embodiment, modification of the Raman spectroscopy reference data can also be used to identify microscopic morphological structures comprising plaques, and these microscopic to the macroscopic arterial Raman spectra. A diagnostic algorithm based on the relative contribution of the morphological structure can be used to accurately assess the pathological state of the artery.

  In a preferred embodiment, eight atherosclerotic classifications are used for comparison with previous studies using principal component analysis (PCA) and chemical reference data. These 8 categories are reduced to 3 categories. Upon pathological examination, the presence of FC, NC and CC is an important predictor of plaque instability and disease progression. Aspects of the present invention show that Raman spectroscopic analysis of these identical morphological structures can be used to diagnose atherosclerotic lesions in intact coronary arteries without the need for microscopy. Show. This is because Raman spectroscopy is easy not only with quantitative chemical information but also with current diagnostic imaging techniques such as intravascular ultrasound (IVUS), magnetic resonance imaging (MRI) and angiography. It also suggests that quantitative morphological information regarding the composition of atherosclerotic lesions, such as the presence of CC not available, may also be provided.

  In a preferred embodiment, the spectral signatures of normal and atherosclerotic arterial tissue in situ cells and extracellular morphological components are measured using confocal Raman microscopy. Specific morphological structures are selected by their role in normal arterial structures (eg elastic lamina) and / or atherosclerotic plaque formation (eg foam cells, necrotic centers, cholesterol crystals). Least squares minimization of the linear combination of the fundamental spectrum of 12 biochemical components provides information on the biochemical composition of various morphological structures. These biochemical components were known to exist in high concentrations in normal arterial tissue and / or atherosclerotic plaques (eg, collagen, elastin, and free and esterified cholesterol) or Is a strong Raman scatterer (eg β-carotene). Glycosaminoglycans (eg hyaluronic acid, chondroitin sulfate, dermatan sulfate and heparan sulfate) that can contribute 3% of the dry weight of arteries are probably weak Raman scatterers (ie they have a small Raman cross section) ) And was excluded from the reference data, so it did not contribute significantly to the fit of the biochemical model and the reference data.

  Embodiments of the present invention interpret Raman spectra with respect to morphology. For example, a Raman spectrum can be associated with a morphological structure, such as foam cells that can be associated with a particular compound. In addition, the number of spectra can be reduced, for example, from multiple chemical spectra to as few as eight unique spectra associated with morphological structures, thereby reducing fit errors. Diagnostic methods that can be used to identify and monitor vulnerable plaques using the fiber optic catheter system of the present invention include chemical composition, morphological structure, thickening of the intima layer and overlying collagen layers. Includes the use of information about thinning. A preferred embodiment involves the determination of collagen depth by measuring the percentage of collagen. In addition, the presence of calcification is monitored, and any edges are identified and located with respect to collagen as indicators of potential rupture and clots. In addition, the reduced fractional fit contribution of collagen fibers in non-calcified plaques is an indication of reduced plaque stability.

The foregoing and other features and advantages of the system and method for in vivo biological tissue spectroscopy are described below with respect to preferred embodiments of the system and method, as specifically illustrated in the accompanying drawings. It will be clear from the specific description. In the drawings, like reference characters refer to the same parts throughout the different views. The drawings are not necessarily drawn to scale, and instead focus on specific illustrations of the principles of the invention.
Detailed Description of the Invention
The present invention is directed to systems and methods for using tissue Raman spectroscopy. An attribute for developing systems and methods for in vivo applications that use angiographic catheters to assist cardiologists in assisting cardiologists is to deliver low energy laser light to in vivo tissue and then The development of an optical fiber probe for Raman spectroscopy that can efficiently collect the resulting Raman spectrum signature. The preferred embodiment probe is small and uses micro-optic design principles.

  Methods of performing Raman spectroscopy for tissue diagnosis and treatment are described in US Pat. No. 5,615,673 issued April 1, 1997, US Pat. No. 5,304 issued April 19, 1994. No. 173, International Patent Application No. WO 92/15008 published September 3, 1992 and International Patent Application No. WO 96/29925 published Oct. 3, 1996 (all references). The entire teachings of which are incorporated herein by reference).

  There are at least two difficulties to be overcome in the manufacture of such probes. The first is due to the spectral background signal generated in the delivery and collection fibers themselves, which may be an order of magnitude greater than the signal from the tissue site being examined. This background signal includes Raman light from a fused silica core, impurities and fluorescence from dopants used to design fibers with a specific numerical aperture (NA), and signals from various jacket materials. . The laser light from the delivery fiber typically creates a strong fiber background due to the long optical path length traversing the 3 to 4 meter fiber. This fiber spectrum is scattered from the tissue surface and collected by the collection fiber along with the tissue's Raman spectrum, often resulting in a tissue Raman signal generated from a sample of only about 1 mm due to the relatively short transmission of light into the tissue. Shield.

  In addition, laser light backscattered from the tissue is also collected by the collection fiber, and this scattered laser light produces an additional fiber spectrum that emits in the collection fiber, which is the spectrum of the tissue that reaches the detector. The quality is further impaired. In addition to blurring and distorting the spectrum of interest, a strong fiber background adds some level of short noise to the signal, and this noise can often be greater than the tissue Raman band. Analyzing both the delivery and collection fibers shows that they both produce approximately equal amounts of this fiber spectrum background. In a preferred embodiment, two different filters (one for delivery and one for collection) are used to suppress unwanted fiber background. In addition, it is desirable to terminate the delivery fiber with a short wavelength pass or band pass filter that propagates the laser excitation light while blocking the longer wavelength spectral background generated in the delivery fiber. In a preferred embodiment, the collection fiber can be preceded by a long wavelength pass or notch filter that propagates the Raman spectrum of the tissue while blocking the laser light backscattered from the tissue. Any filters used will also perform adequately over a range of angles corresponding to the allowable angles (NA) of the fibers to which they are connected.

  The second difficulty relates to signal collection. This has two components, the first of which relates to the inherently weak nature of the Raman effect. Approximately one out of every billion excitation photons is converted to Raman photons. In a preferred embodiment, the high-throughput optical probe device collects signals with a signal-to-noise ratio (SNR) sufficient to be useful in a clinically realistic time frame. In order to be clinically useful and commercially feasible, one preferred embodiment collects a high SNR spectrum in approximately 1-2 seconds. The second component also deals with optimization of collection that is further impaired by absorption and scattering in the tissue resulting in a diffuse spread of light over a large area and angle.

In a preferred embodiment, the collection capacity of the optical system is limited by its throughput given approximately by the product of the collection area (A) and the solid angle (Ω) (AΩ product). The AΩ product is preserved throughout the system. In a typical Raman spectroscopy system, the throughput is determined by a spectrometer / CCD acquisition detection system. In a preferred embodiment, the spectrometer is f / 1.8 (NA = 0.278) corresponding to a solid angle of Ω = 0.242 sr with a maximum slit height of 16 mm. A slit width of 0.2 mm is used to achieve sufficient spectral resolution for biological Raman spectroscopy. Therefore, the maximum collecting area is A = 3.2 mm ^2, resulting in a theoretical maximum throughput of AΩ = 0.77mm ² -sr. In a preferred embodiment, a CCD detector is used to ensure that the effective Raman light source generated in the tissue by incident excitation light is optimally collected, whatever the brightness. Light is believed to be emitted over one large area and a solid angle of 4π, but is limited by the collection angle 2π. Thus, the optimal tradeoff between collected solid angle and area is determined in a preferred embodiment of the present invention.

  In a preferred embodiment system, the spectrometer / CCD is replaced by a higher throughput system. For example, one such arrangement consists of a series of dichroic beam splitters, filters, and photodiodes. The filter wavelength is determined to optimize multivariate spectral analysis with a minimum number of wavelengths. The exact number of wavelengths and bandwidths of the detector elements depends on the spectral characteristics of the chemical / morphological structure to be detected. Such a system provides much greater throughput than prior art spectrometer / CCD systems, and is smaller, cheaper and does not require cooling, further eliminating bulk and cost.

  1A and 1B graphically illustrate the comparison of the logic, simulation, and results of Raman emission data of a turbid sample of blood tissue for the radial and angular distributions of Raman scattered light, respectively, according to a preferred embodiment of the present invention. . Biological tissue is a collection of similar cells and the intercellular material surrounding them. The four basic tissues in the body include epithelial tissue; connective tissue including blood, bone and cartilage; muscle tissue; and nerve tissue. Most tissues except the cornea are cloudy. This is because they exhibit a high degree of elastic scattering due to microscopic structures and variations in the refractive index contained therein, and thus the light entering such tissues is greatly diffused. Thus, the sample is characterized as a turbid sample in FIGS. 1A, 1B, 10A-10C, 38A-38E. In a preferred embodiment, a simulation such as a Monte Carlo simulation is performed to predict the spatial and radial distribution of both excitation light and Raman scattered light. In a preferred embodiment, the radial distribution is approximately Gaussian as shown in FIG. 1A, while the angular distribution is Lambertian as shown in FIG. 1B. Using these parameters and an optical throughput theorem that involves a transformation of the product of the area and solid angle of light propagating through the optical system, the optimal collection area and angle are determined. It should be appreciated that the product AΩ is constant and the selection of the optimal combination of A and Ω is important as shown in FIGS. 10A-10C. In a preferred embodiment, a radius of approximately 0.35 mm and a collection parameter of 55 ° is optimal for blood tissue. Optimal acquisition parameters for arterial tissue are a radius of approximately 0.4 mm and 20 °. Thereafter, the results of the analysis are incorporated into an optical design program such as Zemax program to determine the appropriate optical system for maximum signal collection.

  FIG. 38A graphically illustrates the results of Raman firing data 1360 of a turbid sample of arterial tissue as opposed to blood tissue described with respect to FIG. 1A for the excitation and radial distribution of Raman scattered light of one preferred embodiment of the present invention. I will explain it. Curve 1362 illustrates a fit using three Gaussian distribution fits. FIG. 38B further illustrates graphically the results of Raman firing data 1360 of a turbid sample of arterial tissue for the angular distribution of excitation and Raman scattered light according to a preferred embodiment of the present invention. Similar to FIGS. 10A-10C, which specifically illustrate blood tissue distribution and collection efficiency; FIGS. 38C-38E illustrate integrated radial distribution, integrated angular distribution, and optimality of arterial tissue, respectively, according to a preferred embodiment of the present invention. It is an illustration of the collected collection efficiency.

  Optical elements are used to transmit light collected from the tissue to the distal end of the optical fiber in the probe. The proximal end of the fiber bundle is then reshaped to match the shape, area and NA of the spectrometer. These procedures are followed to provide effective coupling without reducing light transmission efficiency. The choice of collection fiber NA and collection fiber diameter is dictated by limitations set by spectrometer NA, desired spectral resolution, and optics matching considerations, and filter incident angle. The system trade-off includes the chosen spectrometer and the desired resolution determines the slit width. At the output end, the collection fibers are arranged in a straight line, which is imaged on the entrance slit by the alignment optics. Considering the throughput theorem, the requirements for the collection fiber include that the product of fiber NA and diameter is equal to the product of spectrometer NA and slit width. The need to use matching optics is eliminated when selecting a fiber that satisfies the stronger condition that the fiber diameter is equal to the slit width and the NA of the fiber is equal to the NA of the spectrometer, and the probe is in the spectrometer. Connected directly to. Matching optics are required if only product requirements can be met. In an alternative embodiment, the spectrometer uses a curved slit and the output end of the collection fiber can be modified to match any slit shape. The upper limit on the number of collection fibers is defined by the height of the fiber array image, which is smaller than the slit height or the CCD chip, whichever is smaller. However, less space may be set by the space available in the collection tip.

  2A-2C are morphological models and reference graphical representations 30, 40, 50 of coronary arteries of one preferred embodiment of the system. The study uses biochemical composition in determining plaque stability and plaque progression. Morphological factors are described in Hendrik P. et al., Published in Cardiovascular Pathology 10 (2001) 69-82. By Buschman et al. “Raman microspectroscopy of human corrosary aerosclerosis: Biochemical assessment of cellular 10” As discussed by Buschman et al., “Diagonosis of human coronary aerosclerosis by morphology-based Raman spectroscopy”, the full teachings of which are incorporated herein by reference.

  FIG. 3 is a graphical representation 60 of the coronary Raman morphometry method of one preferred embodiment of the present invention. Relative fit factors are plotted against various conditions in normal arteries, arteries with non-calcified plaques and calcified plaques.

  Depending on the preferred aspect of the endovascular application, all of the parameters that can include, for example but not limited to, optical filtering and high-throughput optics designed to collect from a diffusion source increase the tip diameter. Or without sacrificing its flexibility. Many prior art commercial probes are designed to be used with 785 nm excitation. The method of the present invention encompasses the recognition that the fluorescence background generated in tissue with 785 nm excitation is at least four times greater than that generated with 830 nm excitation. Operating at 785 nm requires longer data acquisition times, which prohibits in vivo applications. The longer the operating wavelength, the better the fluorescence background. In a preferred embodiment, the use of 830 nm is governed by the fundamental long wavelength limit (1100 nm) of silicon-based charge-coupled device (CCD) detectors that are dominated by the silicon band gap. Alternative preferred embodiments may use 785 nm or 1064 nm excitation light with appropriate detector technology.

  One preferred aspect of the present invention is to remove the fiber optic background to allow data collection time of a few seconds or less, e.g. length of rigid distal tip to facilitate use in coronary catheter placement. It includes a fiber optic Raman probe that limits the thickness to less than a few millimeters and a diameter of about 2 mm, uses excitation at 830 nm, and maximizes signal collection from the diffusion source.

  One preferred embodiment includes a rod and tube configuration, in which the rod and tube of the optical filter module are individually coated, which is two separate coatings, a central one for filtering excitation light and a collection. It is easier than coating a single disc with one of the edges for filtering light. These embodiments are preferred over coating individual fibers. This is because the filter can adhere better due to the increased surface area. In addition, a two-tone disc is preferred over a single disc. This is because it is difficult to deposit a concentric coating on a small diameter with a smooth circular interface without steps or overlapping regions. Furthermore, it is difficult to put a 3 meter fiber length into the deposition coating chamber. Each filter may contain a large amount of dielectric thin film. Such thin film filters can be secondary processed by Research Electro-Optics Inc, Boulder, Colorado.

  4A-4B are longitudinal and cross-sectional views, respectively, of a preferred embodiment of a device that includes a Raman probe. Device 70 includes a two-piece multi- (e.g., two-wavelength) micro-optic dielectric filter module to minimize and preferably eliminate the Raman background of the fiber in the delivery and collection fibers. This module consists of a rod 82 that carries the excitation dielectric filter coating on one plane, fitted in a tube 78 that carries the collection dielectric coating on one plane of the tube. The rods and tubes are used in a manner that is manufactured from either sapphire or fused silica that are individually coated with their respective filters prior to assembly. The rod is wrapped or covered with a thin sheet of metal 80 to provide optimal isolation between components. The module is then placed at the distal end of the probe between the fiber bundle and a lens system for collimating a light beam having a lens 86 such as a ball lens. The lens collects light from a large angle and large area that effectively overlaps the excitation and collection regions. Ball lenses are secondarily processed and supplied by Edmund Industrial Optics, NJ. In a preferred embodiment, a sapphire lens coated with an anti-reflective coating and having an appropriate angular tolerance index, such as 1.77, is fabricated by MK Photonics, Albuquerque, NM. Although obtaining a high quality interference filter of this scale is expensive, the cost of the filter does not depend on the number of pieces to be coated, so it can coat many filters at once, thereby assembling each probe. Costs can be reduced. In addition, the filter size can be adjusted to create smaller diameter probes for a variety of applications with additional coating operations. In a preferred embodiment, the filter is attached to sapphire or quartz rods and tubes for proper registration with the fiber.

  4C and 4D are a longitudinal view and a cross-sectional view, respectively, of an alternative preferred embodiment having a parabolic mirror attached to the lens system. The collection angle can range from 0 to about 55 ° with a collection diameter of about 1 mm. Parabolic mirrors collect light from a wider range of angles and larger areas.

The transmission characteristics of the excitation and collection fibers incorporating these filters are shown in FIG. 5, where 0 cm ⁻¹ = 830 nm.

  According to a preferred embodiment, the choice of fiber diameter and numerical aperture (NA) is low, such as the following considerations, eg, the fiber Raman signal (which produces undesirable background) is proportional to the square of NA and independent of fiber diameter The NA is governed by better and the diameter has no effect.

  For the pump fiber, it is useful to use a lower NA fiber, however, there is a disputed point therewith. At the input end, it makes the coupling of energy into the fiber more difficult. In one preferred embodiment, reasonable attention in mounting the fiber and alignment optics avoids this problem when excited with a laser with low beam divergence. The beam is more limited at the output end. This makes the fiber configuration simpler and more efficient, but illuminates a larger area to minimize the potential for tissue damage by limiting the incident beam power to a smaller area (spot). It can also be important. However, even smaller diameter laser excitation light spots incident on the tissue will spread to encompass a larger area (typically 1 / 2-1 mm diameter) due to the turbidity of the elastic scattering described above, and thus this consideration will be reduced. ease. In a preferred embodiment, larger diameter fibers or a number of distributed smaller fibers are used. A preferred embodiment is that low NA fibers typically exhibit an increased spectral background caused by the dopant used in the core and the cladding of the fiber that reduces NA, and hence a moderate core size and NA for the pump fiber. Balance with the fact of using.

  The situation is different for collecting fibers. The collected Raman energy is proportional to the square of NA. Thus, from signal versus background analysis, there is one advantage in using a high NA collection fiber whose size is limited by the NA of the spectrometer. Here, the best choice of fiber NA and fiber diameter is determined by the limits set by the spectrometer NA, the desired spectral resolution and alignment optics considerations, and the allowable angle of the filter. In one preferred configuration, one or several delivery fibers are used. This is because the energy of the laser source can be efficiently coupled into the delivery fiber (s). However, as shown in FIG. 4B, a greater number of collection fibers is important to increase the collection area. Since it is important to optimize the collection of Raman light, the collection area is maximized. Taking into account all these considerations, it is best to keep the number and diameter of the delivery fiber (s) to a minimum and use the available cross-section of the fiber optic probe for the same collection fiber It is.

  Preferred embodiments include the following trade-offs. For the selected spectrometer, the desired resolution determines the slit width. Again considering the throughput theorem, the requirement for the collection fiber is that the product of the fiber NA and diameter is equal to the product of the spectrometer NA and slit width. The need to use alignment optics is eliminated if it is possible to choose a fiber that satisfies the more aggressive conditions where the fiber diameter is equal to the slit width and the NA of the fiber is equal to the NA of the spectrometer, and The probe can be connected directly to the spectrometer. Matching optics are required if only product requirements can be met. At the output end, the collection fiber is arranged in a straight line, which is imaged on the entrance slit by the alignment optics. Sometimes the spectrometer uses a curved slit; the output end of the collection fiber can be modified to match any slit shape. The upper limit on the number of collection fibers is that the height of the fiber array image is smaller than the slit height or the CCD chip, whichever is smaller. However, less space may be set by the space available in the collection tip.

  In a preferred embodiment, the fiber section of the probe includes a single central excitation fiber having a NA of 0.22 and a core diameter of 200 μm. The fiber buffer layer is matched to the diameter of the excitation filter rod to facilitate proper fiber / filter registration and has an aluminum jacket to provide optical isolation from the collection fiber. Collection fibers with a core diameter of 200 μm are arranged in two different shapes in two alternative embodiments. The first embodiment consists of two concentric rings of 10 and 17 fibers in the inner and outer rings, respectively. The second embodiment has a single ring of 15 collection fibers. Although the second design has a slightly reduced collection efficiency, it is more flexible and is still capable of collecting high SNR spectra in short exposure times. The collection fibers all have a 0.26 NA such that they are f / # matched to the spectrometer for optimal throughput as specifically illustrated in FIGS. The diameter of the probe in a preferred embodiment is less than 2 mm for reaching the coronary artery.

  One preferred aspect provides flexibility with respect to the particular choice of optics for high-throughput collection such that a variety of optical elements can be used to collect the desired AΩ product. In a preferred embodiment, the ball lens is calculated collected over a radius of 0.35 mm for blood tissue (0.4 mm for arterial tissue) while still collecting over a large angle as shown in FIGS. 1A-1B and 4A-4B. Provides a highly efficient collection for front-viewing fiber optic probes that closely match. A collection efficiency greater than 30% is achieved when a small space is maintained between the sample and the lens, and a collection efficiency greater than 10% is achieved when in contact with the tissue (possible and more reproducible in vivo arrangement). Is done. A high quality spectrum obtained in one preferred embodiment using a probe in contact with tissue is presented in FIG. 6, which shows the Raman spectrum of non-calcified atherosclerotic plaques collected in 1 second with an excitation power of 100 mW. In contrast, FIG. 7 shows the spectrum of a normal artery taken with another preferred embodiment system at the same excitation force at 10 seconds.

  FIG. 8 is an illustration for specifically explaining a tissue measurement system according to a preferred embodiment of the present invention. A light source 206 emitting at a wavelength longer than 750 nm is used, such as an argon pumped Ti: sapphire laser system or a diode laser. The diode laser may be an InGaAs laser emitting at 785 nm or 830 nm, for example, as secondary processed by Process Instruments, Salt Lake City, Utah. The laser output is bandpass filtered and coupled into a delivery optical fiber that is included in the probe 204. Probe 204 is inserted into artery 202 to diagnose and possibly treat, for example, plaque buildup in the artery. FIG. 15 is a partial cross-sectional view of a portion of a coronary artery showing a probe according to a preferred embodiment of the present invention. The system may include a guide wire and a guide catheter for passing large arteries. In a preferred embodiment, in an artery that is partially occluded by fatty material 416, a guide wire is first extended and then a catheter containing balloon 420 is extended into the artery. The probe assembly is housed at the tip of the catheter and has a collection window 418. Once the balloon has entered the artery 414, the probe assembly provides the surgeon with a cross-sectional view of the artery. The balloon temporarily blocks blood flow to provide a clear field of view, stabilizes the probe and minimizes the effects of heart movement. In a preferred embodiment, a fluorescent marker may be used. Light is incident on the tissue and Raman scattered light from the tissue is collected by a collecting optical fiber. The collected light is notch filtered and projected onto the entrance slot of the spectrometer. The notch filter removes Rayleigh scattered laser light. A diffraction grating disperses the light into the CCD detector 228 inside the spectroscope. A CCD interface and data storage and processing are provided on a computer such as a personal computer. A program such as Winspec software provided by Princeton Instruments may be used to connect the CCD interface to a personal computer that performs data processing and storage functions. In an alternative embodiment, the CCD interface is connected to a personal computer using the Labview program by National Instruments, Austin, Texas. The Raman signal may be read from the CCD, collected by a computer, and stored on a computer readable medium for later analysis, or used for real-time analysis in clinical situations.

  FIG. 9 illustrates the excitation light diffusing through the tissue of one preferred embodiment of the present invention. 10A-10C are illustrations of the integrated radial distribution, integrated angular distribution, and optimized collection efficiency, respectively, for blood tissue according to one preferred embodiment of the present invention. FIG. 10C illustrates the collection efficiency by varying A and Ω, specifically illustrated in FIGS. 10A and 10B, respectively, but keeping the product AΩ constant and equal to that of the spectrometer. Throughput theorem preserves etendue, where etendue is the product of area and solid angle. The radial and angular distributions are integrated and their product determines the optimization curve 300. The collection optics are designed to run at the maximum of the efficiency curve 300. Specific graphical illustrations for arterial tissue are provided in FIGS. 38C-38E.

  FIG. 11 is a graphical representation 320 of the excitation spot diameter according to one preferred embodiment of the present invention. FIG. 12 is a specific illustration of a ray diagram 340 of the distribution of excitation light in the tissue of one preferred embodiment of the present invention. FIG. 13 is a specific illustration of a ray diagram 360 of the collection efficiency of the probe of one preferred embodiment of the present invention.

  FIG. 14 is a diagram illustrating the collection efficiency of the probe according to one preferred embodiment of the present invention. The efficiency from the distribution measured for tissue and air is specifically described. All components of the probe are constructed from medical grade materials that can withstand standard cold gas, ethylene oxide sterilization. Alternative sterilization methods such as those provided by Steris Corporation of Ohio, such as a pasteurization treatment system may be used. The filter module in the probe tip is assembled and bonded to the fiber bundle using high purity sodium silicate such as index matching cement. The advantages of sodium silicate as an index matching cement in Raman spectroscopy are unique and its utility goes beyond current applications. The advantages are that it does not produce an interference Raman spectrum, has a refractive index close to that of fused silica, thereby greatly reducing the loss of reflection from the mating surface, and has low light absorption at near IR. And therefore it does not introduce any appreciable loss of absorption, has a binding property that facilitates the assembly of the miniature optical components involved, and it is acceptable for use in many industrial applications and transactions It is included.

  Sodium silicate is a ternary compound created by mixing various combinations of water, silicon dioxide and sodium hydroxide in alternative sodium oxide. The optical and mechanical properties of the final product can be adjusted by varying these ratios. Other alkali silicates such as lithium silicate, potassium silicate have similar properties and may also be used in certain applications.

  It is important not to have any adhesive between the ball lens and the filter so that there is no exponential matching that can detract from the lens effect provided by the curvature of the lens. The lens is sealed with a medical grade epoxy in accordance with a preferred embodiment of the present invention to protect it with a brazed retaining sleeve and prevent liquid from leaking into the probe tip.

  The module nature of the preferred embodiment probe is very flexible and can accommodate many optical embodiments. Additional aspects of side-viewing probes, as well as other forward observation aspects for alternative applications, are encompassed by the system of the present invention. For example, the use of angled and mirrored ball lenses, prisms or micro-optical parabolic mirrors allows for efficient side observation probes. The tapered tip allows for incorporation into a needle probe for optical breast biopsy, and a slightly smaller diameter in an alternative preferred embodiment allows breast analysis by ductoscopy. Other potential uses are for skin analysis, transcutaneous blood analyte monitoring, and assessment of gastrointestinal cancer.

  FIG. 16 illustrates the signal 450 and function from the laser output ball lens of one preferred embodiment of the present invention. The forward viewing Raman probe uses a sapphire ball lens to concentrate the excitation light and collect the Raman signal from the tissue. The Raman spectrum from the sapphire lens can be used as an internal standard to calibrate the collected signal with respect to the excitation laser power and thereby obtain intensity information. This intensity information is typically not exploited in biological Raman spectroscopy, but can provide enhanced diagnostic power. A plot of the graph is generated using data taken with a preferred embodiment of the Raman probe and a clinical Raman system, and the magnitude of the signal from the sapphire lens is shown as the excitation force as the probe is held in the air. Express as a function. It implies a natural internal standard for measuring the power delivered to the tissue using a preferred embodiment of the present invention.

  FIG. 17 illustrates graphically a comparison of data from normal arteries collected using a preferred embodiment of the probe 472 and experimental system 474. A similar type of tissue is examined to demonstrate and verify the function of the Raman probe. A similar spectrum has been shown to be obtained and the S / N ratio (SNR) of the spectrum (an indication of the system's performance in order to convert a higher SNR to a system with less noise and show better performance) Are also inspected.

  FIG. 18 specifically illustrates a Raman spectrum 490 of normal breast tissue examined using the probe of the present invention. Tissues other than arteries were examined to show that the probe has multiple uses for disease diagnosis. The probe may be used in a ductal endoscopy procedure for early diagnosis of breast cancer.

  FIG. 19 illustrates graphically a comparison of breast cancer Raman spectra as diagnosed as malignant according to a preferred embodiment of the present invention and as predicted by the reference data of the present invention. Data 502 collected using a preferred embodiment of the probe is compared to data 504 generated by reference data. The reference data coefficients are summarized in Table 1.

  FIG. 20 illustrates the comparison of morphological reference data with calcified aorta data collected using the probe of one preferred embodiment of the present invention. FIG. 20 illustrates the Raman spectrum 522 of the calcified aorta obtained using a Raman probe and a clinical Raman system in 1 second with an excitation power of 100 mW. A fit with morphological reference data is shown as spectrum 524. The residual spectrum 526 is plotted below on the same scale. The lack of features in the residual indicates a high level of agreement between the observed data and the reference data, and the preferred embodiment reference data generated using an experimental in vitro system is Prove that it can be applied to the data taken by the probe. It should also be noted that there is no feature corresponding to any probe background in the residual spectrum 526, indicating that any remaining unfiltered background noise can be accurately removed. Furthermore, the Raman spectrum obtained from diseased arteries is not as diffuse as that obtained from normal arteries, thus providing a spectrum with a better level of signal-to-noise ratio (SNR). It should be noted that SNR is affected by both the Raman cross section of the tissue and the light distribution.

  In a preferred embodiment, the non-axial Raman probe of the present invention for use in the diagnosis of atherosclerosis is incorporated into a catheter of the type used, for example, for angiography. It includes a balloon for evicting blood and other fluids and placing a catheter in the artery. One preferred embodiment includes a channel for balloon inflation. In addition, the catheter system includes the ability to flush blood temporarily with a liquid such as physiological saline. One or several optical fibers may be configured to direct the excitation light in a non-axial direction either at the side or at an angle ranging from 45 ° to 90 °. In one such preferred embodiment, a side-positioned balloon is used to bring the fiber into contact with the arterial wall and evict blood or other intervening fluid.

  Alternatively, the delivery fiber may be arranged to direct light in a circular pattern at an angle to the probe axis. Various collection fibers collect light simultaneously from various portions of the illuminated perimeter region. In this embodiment, the probe is encapsulated in an inflatable balloon that is inflated prior to light delivery and / or collection to eradicate blood and other fluids. In a preferred embodiment, the balloon is of the type used in arterial applications, such as angiography, and collected by passing excitation light through the artery wall and Raman light generated at the artery wall through the balloon. Made of thin material so that it can be returned to the fiber.

  FIG. 21A illustrates a longitudinal view 550 of an alternative preferred embodiment of a lateral observation probe for tissue measurement of the system of the present invention. The embodiment includes a modified axicon 556 in which the angled surface is oval. FIGS. 21B and 21C are preferred alternative embodiments that include a minimum of two different radii of curvature on the angled surface to provide a surrounding image. Ambient imaging is obtained in one aspect by providing a beam over a range of angles of approximately 45 ° to 90 ° and rotating the probe to obtain an ambient image. Alternatively, the delivery fiber can provide light to the tissue and an image, such as 6 images, is collected with the collection fiber to obtain a surrounding image. In a preferred embodiment, the volume between the filter 554 and the angled portion of the axicon 556 comprises solid glass, preferably sapphire, where light redirection is through total internal reflection. Occur.

  In an alternative preferred embodiment, as specifically illustrated in FIG. 21B, the angled surface 562 of the axicon 556 is imaged, which allows reflection. The laser light is directed radially or non-axially on the tissue 564. Further, surface 565 is elliptical and is fabricated using sapphire. The volume between the filter 554 and the axicon 556 may be filled or empty. This embodiment, which is specifically illustrated in FIG. 21C, temporarily blocks blood flow to flow fluid, such as saline, into the artery while the spectral data is acquired or by inflating a balloon. Open central channels 558, 566 are provided that can be used to If central channels 558, 566 are used, the probe includes placing several excitation fibers around the periphery of the central channel. The focus of the axicon is adjustable. The shape of the rod in the tube of the filter described in the previous embodiment is modified to the shape of the tube in tube, ie the central tube of the excitation filter with the central hole for the central channel and the outer tube for the collection filter. The

  FIG. 21D illustrates a side or non-axial viewing probe of one preferred embodiment of the invention, and a catheter 610 that delivers light onto a portion of tissue 612, and a point on the second surface collected from the tissue. The figure of the preferable one aspect | mode of the Raman light re-imaged by (1) is demonstrated concretely. The side viewing probe includes an inflatable balloon 614 or a flexible wire placed adjacent to a side-viewing element transverse to the viewing opening. Upon balloon inflation or wire bending, the opening is pushed into contact with the arterial wall 616 to replace blood and / or establish a well-defined collection shape.

  As briefly discussed above, recent studies have shown that chemical composition and morphology rather than anatomy (the degree of stenosis) determine atherosclerotic plaque instability and disease progression and thrombus and acute plaque bleeding Like to predict the risk of life-threatening complications. For example, the presence of cholesterol esters may soften plaques, while cholesterol without crystals may have the opposite effect. Raman spectroscopy can identify cholesterol esters from free cholesterol as illustrated in FIGS. 35F and 35G. Prior art clinical diagnostic techniques provide an accurate assessment of plaque anatomy, but have limited ability to assess in vivo plaque morphology. In addition, prior art diagnostic imaging techniques such as intravascular ultrasound (IVUS), MRI and angiography predominately provide anatomical information about the extent of luminal stenosis, but of lesions. Only produces limited information about the composition. Coronary angiography (still a “golden standard” for diagnosing coronary artery disease) indicates the degree of luminal stenosis, but does not provide chemical or morphological information about the plaque. In fact, unstable atherosclerotic plaques are often “asymptomatic” in angiography. IVUS, the most accurate and quantitative technology currently in clinical use, reviews the tissue density using the reflection of sound waves delivered by an intravascular catheter and provides imaging information. It has greatly advanced understanding of atherosclerosis by showing widespread atherosclerosis in the coronary arteries that appear normal on angiography. However, although IVUS can identify the presence of an atheroma center, it cannot specifically identify foam cells (FC) or cholesterol crystals (CC) and does not provide any chemical information. MRI has the advantage of being a non-invasive technique and again uses radio waves generated by applying a magnetic field gradient to scrutinize tissue density and provide imaging information. Like IVUS, it can be used to analyze anatomy and morphology to a lesser extent. However, conventional proton MRI techniques used in the clinic almost ignore and often suppress chemical shift information. Thus, at present, plaque morphology and chemical composition can only be assessed by microscopic examination of the removed tissue after endarterectomy or atherectomy.

  Preferred embodiments of the present invention include diagnostic methods based on the morphology of atherosclerosis in coronary arteries using Raman spectroscopy, which can potentially be performed in vivo using fiber optic technology. In a preferred embodiment, Raman tissue spectra are collected from various disease progression stage atherosclerotic coronary artery samples (n = 165) from normal and explanted transplant recipient hearts (n = 16). Elastic plate (EL), collagen fiber (CF), smooth muscle cell (SMC), outer membrane adipocyte (AA) or fat cell (fat cell), foam cell (FC), necrotic center (NC), cholesterol The Raman spectra from crystals (CC), β-carotene-containing crystals (β-C) and calcium mineralization (CM) are used as the basis spectra of the linear least squares minimization (LSM) model to coronary tissue spectra Calculate the contribution of these morphological structures to A preferred embodiment is a set of initial calibration data as either non-atherosclerotic, calcified plaques, or non-calcified plaques using fit-contributions of various morphological structures. Includes a series of diagnostic instructions to classify the 97 coronary artery samples in it. The series of instructions currently prospectively classifies 64 (94%) of 68 coronary artery samples. Raman spectroscopy provides information about the morphological composition of intact human coronary arteries without the need for excision and microscopy. Accordingly, one preferred embodiment uses Raman spectroscopy to analyze the morphological composition of atherosclerotic coronary lesions in vivo and assess plaque instability and disease progression.

  The present invention involves obtaining quantitative morphological information regarding the composition of lesions from coronary arteries by Raman spectroscopy using a modification of mathematical reference data. This morphological information can be used for diagnostic purposes. Chemical and morphological information obtained by Raman spectroscopy can be the basis for diagnostic evaluation of human coronary artery disease.

  In principle, both quantitative chemical and morphological information regarding the composition of atherosclerotic lesions can be obtained from the same Raman spectrum. One preferred aspect of the present invention analyzes coronary artery tissue by modeling Raman tissue spectra using spectra of morphological structures rather than biochemical components as a basis set. The fundamental spectrum of the reference data includes collagen fibers (CF), internal and external elastic lamina (EL), smooth muscle cells (SMC), outer membrane adipocytes (AA) or fat cells, fat cells (foam cells) FC), necrotic centers (NC), cholesterol crystals (CC), β-carotene-containing crystals (β-C) and calcium mineralization (CM), universally observed in arterial walls and atherosclerotic plaques Obtained from the resulting morphological structure. These base spectra can then be used to linearly fit a set of initial calibration spectra for a coronary artery sample. An algorithm for classifying an artery as atherosclerotic or non-atherosclerotic, as in a biochemical model, using various morphological structural fit contributions is included in a preferred embodiment. The diagnostic performance of the preferred embodiment can be tested by applying morphologically based reference data to a second prospective validation set of coronary arteries.

  In a preferred embodiment, the tissue preparation comprises human coronary artery samples (n = 200) from 16 patients representing various stages of atherosclerosis from the hearts of explanted recipients within 1 hour after heart transplantation. Including obtaining. Seven patients had heart failure due to dilated cardiomyopathy and nine due to severe ischemic heart disease. Immediately following the incision from the explant, the arterial section was rinsed with a neutral buffered saline solution, snap frozen in liquid nitrogen, and stored at -85 ° C until use. Arterial samples were collected in duplicate, the first set contained 113 (calibration set) and the second contained 87 samples (prospective verification set).

  These aortic samples could and could be used for macroscopic and microscopic Raman spectroscopy. For macroscopic examination, samples (97 and 98 from the first and second sets, respectively) are passively warmed to room temperature, opened in the longitudinal direction, placed in an aluminum holder with the luminal side facing up, The test was then performed under 10x magnification to select the area to be evaluated. After spectroscopic examination, each spot examined was marked with a small spot of colloidal ink and fixed in 10% neutral buffered formalin.

Unstained transverse tissue sections (6-8 μm) were cut from coronary artery samples for Raman spectrum collection using a microspectrometer. Four sections of each sample were mounted on glass microscope slides and stained for light microscopy, while serial unstained transverse sections were BaF ₂ or MgF ₂ flats (International Scientific Products, Tarrytown, NY, And Spectra-Tech, Stanford, Conn.), Kept moist with phosphate buffered saline (pH 7.4) and transferred to a microscope stage for spectroscopic experiments. A cover glass was not used. Major morphological structures were selected under white light illumination and recorded on videotape under 10x and 63x magnification.

  A formalin-fixed macroscopic tissue sample was processed, paraffin-embedded and cut through a marked location into 5 μm thick sections, stained with hematoxylin and eosin, and two experienced cardiovascular vessels Examined by a system pathologist. Tissue sections were classified according to the revised Systemized Nomenclature of Human and Veterinary Medicine (SNoMed). Samples in both calibration and validation data sets were (1) normal (n = 12 and 1 respectively), (2) intimal fiber growth (n = 61 and 25), (3) atherosclerotic plaque (n = 3 And 0), (4) Adenoma plaques (n = 6 and 16), (5) Calcified atherosclerotic plaques (n = 1 and 3), (6) Calcified adenoma plaques (n = 7 and 13) , (7) calcified fibrosclerotic plaques (n = 5 and 10) or (8) calcified intimal fiber growth (n = 2 and 0). Since some of these categories had a small sample size, the 8 categories were reduced to 3 diagnostic categories for the development of diagnostic algorithms. Class I, non-atherosclerotic tissue (category 1 and 2; n = 73 and 26); class II, non-calcified atherosclerotic plaque (category 3 and 4; n = 9 and 16); and class III, Calcified atherosclerotic plaques (category 5-8; n = 15 and 26).

Macroscopic and microscopic Raman spectra were obtained using the Raman spectroscopy system shown in FIG. Near-infrared (NIR) laser light (830 nm) is generated by an Ar ⁺ -pumped Ti: sapphire laser system 572,574 (Coherent Nova 90 / Spectra Physics 3900S, Coherent, Santa Clara, Calif.). The laser output was bandpass filtered (Kaiser Optical Systems HLBF, Ann Arbor, MN) and projected onto a tissue sample in a macroscopic situation with the insertion of prism 578, or projected into a confocal microscope 600 and 63 × Either focused on tissue sections using an infinite correction immersion objective (Zeiss Acroplan, NA 0.9). In the macroscopic situation, Raman scattered light from tissue (sampling volume 1-2 mm ³ ) is collected with a lens, notch filtered 588, and entrance slit of a Chromex 250IS / SM spectrophotometer (Chromex, Albuquerque, NM). Concentrated on. In the context of the microscope (sampling volume approximately 2 × 2 × 2 μm ³ ), Raman light scattered from the tissue is collected with the same objective lens used to focus the light on the sample and pinhole Giving confocal features), notch filtering, and projecting into the entrance slit of the spectrophotometer. Inside the spectrometer 580, a diffraction grating disperses the light onto a deep-depletion CCD detector 582 (Princeton Instruments, Princeton, NJ) cooled to -110 ° C. A CCD interface (ST130, Princeton Instruments) is provided on the personal computer 584 along with data storage and processing.

For macroscopic measurements, the laser power is 350 mW and the signal acquisition time is in the range 10-100 s. For microscopic measurements, the laser power is 80-120 mW, the signal acquisition time is 60-360 s, and the Raman spectrum is collected in the range between 400 and 2000 cm ⁻¹ (resolution 8 cm ⁻¹ ). .

Each spectrum is frequency calibrated and corrected for color variations in the detection of the spectrometer system. A fourth order polynomial is fit to each spectrum and subtracted from the spectrum to correct for the remaining tissue fluorescence. The macroscopic tissue spectrum can be modeled with a Raman shift range of 680-1800 cm ⁻¹ as a linear combination of the fundamental spectrum of the morphological structure by LSM. This Raman shift range is chosen because this range contains most of the spectral information.

  The Raman spectra of the morphological structures can be normalized with respect to their maximum peak intensity. All spectra in the two sets of data can be accurately modeled using the last set of eight morphological basis spectra. The Raman spectrum reference data calculated the contribution of 7 partial fits of the morphological structure. The eighth structure, β-carotene, is a strong Raman scatterer that often contributes to the Raman spectrum of the coronary arteries, but is present only at low concentrations. For this reason, the spectrum is included in the spectral reference, but the partial fit contribution is not calculated.

In calcified atherosclerotic plaque, CM often occupies a large volume of tissue that is examined by Raman spectroscopy. Ignore the contribution of calcium mineralization to obtain information about the remaining non-calcified areas and to compare the contribution of partial fit of morphological structures among various disease categories Spectral spectrum is renormalized and the partial fit contribution of the morphological structure of the non-calcified region (indicated by X _NCR ) is calculated.

Using the relative fit contribution of each morphological structure to the spectra in the calibration data set, develop a description algorithm or series of descriptions to classify the tissue into one of three diagnostic classifications . Using the method of logistic regression analysis, the differential score R _I based on the linear combination of the relative fit contributions (C _l ) of each morphological structure l R _i = α _i + β _1i C ₁ + β _2i C ₂ +... (Α _i is constant and β _li is an adjustable coefficient for each morphological structure). Since logistic regression analysis makes no assumptions about the normality of the fit contribution rate, this method is chosen for differential analysis.

Arterial sample j is non-atherosclerotic (P _jI ) or non-lime using maximum maximum likelihood estimation using analysis tools such as software package STATA (Release 5.0, Datata, College Station, Texas) _Probability of containing atherosclerotic plaques (P _jII ) or containing calcified atherosclerotic plaques (P _jIII ) is 1

As determined. In addition, using the maximum likelihood ratio test on the initial calibration data set, which morphological structure relative fit contributions are important to the diagnosis, and these relative fit contributions It can be determined which sample with the highest diagnostic threshold for rate is correctly classified. The algorithm so developed can then be used to prospectively classify the arterial samples in the second validation data set.

  In order to determine the level of error in the reference data, it is necessary to analyze the S / N ratio (SNR) of the spectrum being used. Since microscopic Raman artery spectra of morphological reference data can be collected at will and for a long time, they depict the Raman spectra of the eight selected coronary artery morphological structures of a preferred embodiment of the present invention. As will be described in detail with reference to FIG. Thus, the limiting source of error in reference is due to noise in the macroscopic spectrum of intact arteries. The in vitro system is limited in shot noise, and therefore the noise for any given sample is equal to the square root of the signal. According to standard multivariate analysis techniques, the concentration error is proportional to the noise in the spectrum.

E = N × B (3)
Where B = P ^T (PP ^T ) ⁻¹ is the calibration vector of the desired morphological basis spectrum and N is the noise in the sample.

  FIGS. 2A-2C described above show three diagnostic classifications along with LSM reference data (normal coronary artery (FIG. 2A), noncalcified atherosclerotic plaque (FIG. 2B), and calcified atherosclerotic plaque (FIG. 2C). ) Shows macroscopic Raman spectra collected from coronary artery samples representing each of The solid line is the macroscopic spectrum and the dotted line implies reference data. Residuals (data-fit) are shown on the same scale. For all spectra, the calculated fit is in good agreement with the measured spectrum, suggesting that the morphological basis spectrum is a reasonable and complete representation of the Raman spectrum of the macroscopic tissue sample.

The Raman spectra of all 97 coronary artery samples of the calibration data set classified by the pathologist into one of three diagnostic categories can be analyzed in the same way. Contribution of relative fit of all eight selected morphological structures in non-atherosclerotic tissue (I), non-calcified atherosclerotic plaque (II) and calcified atherosclerotic plaque (III) The mean ± standard error of each is shown in FIGS. These figures clearly show that the Raman spectroscopy modeling can detect morphological changes in coronary artery tissue. Morphological Raman reference data showed that, as expected, non-atherosclerotic tissue consisted mainly of AA, CF, EL and SMC (FIG. 24A). In non-atherosclerotic arteries, the intima is thin and therefore the contribution of the outer membrane layer (containing a relatively large amount of adipose tissue) to the spectroscopically examined tissue volume is large. This is because NIR laser light penetrates the entire blood vessel wall. In non-calcified and calcified atherosclerotic plaques, morphological Raman reference data showed dramatic changes in morphological composition with disease progression. In non-calcified atherosclerotic plaques with thickened intima, the contribution of AA decreased, while the contribution of FC / NC and CC increased due to the accumulation of lipids in the plaques (FIG. 24B). The Raman spectrum of calcified atherosclerotic plaques is dominated by CM contribution (FIG. 24C). The contribution of AA _NCR and CF _NCR in calcified atherosclerotic plaque is greater than that of AA and CF in non-calcified atherosclerotic plaque. Reduced CF _{NCR in} non-calcified plaques is an indication of reduced plaque stability.

  Although the concentration of β-carotene in arterial tissue is low, the modeling results showed a large difference in carotenoid contribution between disease classifications as illustrated in FIG. The greatest contribution is found in non-calcified atherosclerotic plaques. This is because β-carotene is a lipophilic substance easily dissolved in NC.

Using logistic regression analysis, the optimal separation of the data into the three diagnostic classifications, P <. CM and CF / NC _NCR + CC _NCR fit contributions with 0001 can be used. In addition, the maximum likelihood ratio test determined that no classification improvement resulted from inclusion of any remaining morphological structure (P <0.05). The differential score is R _j1 = −420.4 + 1870.0 × (FC / NC _NCR + CC _NCR ) −6094.3 × CM, and R _j2 = −8.3 + 23.3 × (FC / NC _NCR + CC _NCR ) +47. It is determined to be 6 × CM.

The CM and CF / NC _NCR + CC _NCR fit contributions for each arterial sample may be plotted in a decision diagram as specifically illustrated in FIG. 26A using the corresponding R ₁ and R ₂ values. The boundary separating non-atherosclerotic tissue and non-calcified atherosclerotic plaque regions is given by PI = PII, which is described by the equation CM = −0.07 + 0.31 × (FC / NC _NCR + CC _NCR ). Is a straight line. The boundary separating the areas of non-atherosclerotic tissue and calcified atherosclerotic plaque is given by PI = PIII and has the equation CM = 0.17−0.48 × (FC / NC _NCR + CC _NCR ). The straight line separating the areas of non-calcified atherosclerotic plaques and calcified atherosclerotic plaques is given by PII = PIII and has the equation CM = −0.07 + 0.30 × (FC / NC _NCR + CC _NCR ) . For 95 (98%) of 97 samples in the initial calibration data set, the determinations determined by the Raman-based diagnostic algorithm correlated with those of the pathologist.

  This algorithm was also used prospectively in the preferred embodiment for classifying the arterial samples of the second validation data set into one of the three diagnostic classifications as specifically illustrated in FIG. 26B. Prospectively, the algorithm results were consistent with those of pathologists for 64 out of 68 (94%). A comparison of pathological and Raman spectroscopic diagnoses for both data sets is shown in Table 2.

The classification is (I) non-atherosclerotic tissue, (II) non-calcified plaques and (III) calcified plaques.

Because the in vitro Raman system used to collect the macroscopic arterial spectrum is limited in short noise, the NIR technique used in data acquisition yielded a very high SNR spectrum. The average peak-to-peak noise is less than 0.04 counts on the normalized spectrum. Calculation of the error to the diagnostic morphological component fit contribution yields an error of 3 standard deviations (SD) of 0.041 for CM and an error of 3SD of 0.036 for FC / NC _NCR + CC _NCR .

  In another preferred embodiment, 35 coronary artery samples were taken from 16 explanted transplant recipient hearts and thin sections were prepared. Using a high resolution confocal Raman microspectrophotometer system using 830 nm laser light, a high S / N Raman spectrum was analyzed for the following morphological structures: internal and external elastic plates, collagen fibers, fat, foam cells, smooth muscle Obtained from cells, necrotic center, β-carotene, cholesterol crystals and calcium mineralization. Their Raman spectra are present in arterial tissues including collagen, elastin, actin, myosin, tropomyosin, cholesterol monohydrate, cholesterol linoleate, phosphatidylcholine, triolein, hydroxyapatite calcium, calcium carbonate and β-carotene Can be modeled by using the linear combination of the fundamental Raman spectra from the major biochemicals.

  The results show that various morphological structures have characteristic Raman spectra, which vary little between structures and between arteries. The biochemical model fully describes the spectrum of each morphological structure and indicates that the most essential biochemical components are included in the reference data. Furthermore, the biochemical composition of each structure, as indicated by the biochemical base spectrum fit contribution of the morphological structure spectrum, is very consistent. Thus, the Raman spectra of various morphological structures in normal and atherosclerotic coronary arteries may be used as a base spectrum in a linear combination model for analyzing the morphological composition of atherosclerotic coronary lesions.

  Raman spectroscopy has great potential for biochemical analysis of tissues on both macroscopic and microscopic scales. One of the major advantages of this method is that they are microdegraded in non-destructively intact cells and tissues (ie in-situ) without the use of homogenization, extraction, or dyes, labels or other contrast enhancing agents. Its ability to provide information about the concentration, structure and interactions of biochemical molecules in the environment. In addition, Raman spectroscopy can be performed in vivo using optical fiber technology as described above.

  Using the attribute that morphological factors can be as important as biochemical composition in determining plaque stability and progression, one preferred aspect of the present invention is that in arterial tissue using Raman spectroscopy. Includes diagnosis based on the morphology of atherosclerotic lesions. To that end, in-situ Raman spectra of individual cells and extracellular components of normal and atherosclerotic human coronary arterial tissues are obtained in vitro using the confocal Raman microspectroscopy described above. Obtained. The biochemical composition of these microscopic morphological structures can then be analyzed using a linear combination of biochemical fundamental Raman spectra in a manner similar to that previously used for intact tissue. Determined by modeling. Similar to the biochemical analysis of macroscopic Raman spectroscopy, the morphological analysis of these macroscopic Raman spectroscopy ultimately helps to assess atherosclerotic plaque stability and disease progression in vivo. Can be used in diagnostic algorithms. Human coronary artery samples (n = 35) representing various stages of atherosclerosis were obtained from explanted recipient hearts (n = 16) within 1 hour of heart transplantation. Immediately following incision from the explant, the arterial sample was rinsed with neutral buffered saline, snap frozen in liquid nitrogen and stored at -85C.

Frozen coronary artery samples were mounted on a cryostat chuck using Histoprep (Fisher Diagnostics, Orangeburg, NY). Thin transverse tissue sections (6-8 μm) for light microscopy and Raman microspectroscopy were cut using a cryostat / microtome (International Equipment, Needham Heights, Mass.). Four sections of each sample were mounted on glass microscope slides and stained with hematoxylin and eosin. Subsequently, serial unstained sections were mounted on BaF ₂ or MgF ₂ flats (International Scientific Products, Tarrytown, NY, and Spectra-Tech, Stamford, Conn.) And phosphate buffered saline (pH 7.4). And kept moistened and transferred to a microscope stage for spectroscopic experiments performed at room temperature. No cover glass was used for spectroscopic measurements. When spectra were collected from multiple morphological structures, each section was replaced with a freshly cut section approximately 2 hours later to avoid biochemical changes in the tissue as a result of enzymatic degradation. There was no significant change in the Raman spectrum within the 2 hour examination period. Morphological structures examined were in normal arteries: collagen fibers in various layers of the arterial wall, internal and external elastic lamina, medial smooth muscle cells and epithelial fat cells, and intimal atherosclerotic lesions : Collagen fibers in the fibrous capsule, foam cells, necrotic centers, cholesterol crystals, β-carotene-containing crystals and calcium mineralization.

  An illustration of a preferred embodiment system of the present invention is shown in FIG. All Raman spectroscopic measurements are performed using a confocal configuration to suppress signal Raman light from features on the peripheral surface other than the region of interest of the selected morphological structure. Observation and analysis were performed using a microscope (Zeiss Axioskop 50, Zeiss, Thornwood, NY) equipped with a phase contrast system and a stage controller (Prior Scientific Instruments, Cambridge, MA). Initial microscopic examination of the samples was performed using a phase contrast microscope with 10x magnification (Zeiss Acroplan objective). Detailed microscopic and microspectroscopy were performed using a 63x infinite correction immersion objective (Zeiss Acroplan, NA 0.9). Phase contrast microscopy and morphological structure selection for microspectroscopy were recorded using a CCD color video camera 758 (Sony, Cambridge, Mass.) Connected to a microscope 750 and videotape (VCR). 764 and then digitized the frame (PCVision-plus, Imaging Technologies, Bedford, Mass.).

Near-infrared (830 nm) laser light was generated by an Ar ⁺ laser 742-pumped Ti: sapphire laser system 744 (Coherent Innova 90 / Spectra Physics 3900S, Coherent, Santa Clara, Calif.). Laser output is bandpass filtered 746 (F1) (Kaiser Optical Systems HLBF, Ann Arbor, Minn.), And laser output on a sample 756 of 50-100 mW with adjustable mirror (m1) 748 and dichroic beam splitter (m2) 754 was used to concentrate on the sample. The light emitted from the tissue sample was collected by the same objective lens, passed through a beam splitter, and passed through a pinhole (P: 100 μm diameter) with a removable mirror (m3) 752. This mirror was used to direct light emitted from the sample to either the spectrometer / CCD system or the white light image to the video camera system. The light directed to the CCD / spectrometer is then notch filtered to eliminate Rayleigh scattered light (F2; Kaiser Optical Systems HSNF) and using an achromatic lens (L) Chromex 250IS / SM spectrometer-monochromator (Chromex, Albuquerque, New Mexico). Spectrometer 766 includes grating-dispersed light on a back-illuminated deep depletion CCD detector 768 (Princeton Instruments, Princeton, NJ) cooled to -110 ° C. A CCD interface (ST130, Princeton Instruments) was connected to a personal computer 774 using Winscope software (Princeton Instruments, version 1.4.3) to perform data processing and storage. A minimum of three Raman spectra (between sampling times 10 and 100 s) were obtained from each selected site over a range of 100-2000 cm ⁻¹ (8 cm ⁻¹ resolution).

The confocal Raman microspectrometer light collection or sampling volume estimation method is small (1 to ensure sufficient resolution to collect Raman spectra from small or thin microscopic structures such as individual collagen fibers. ˜2 μm ³ ) Use collection volume. In summary, 1.0 μm diameter polystyrene beads (Polyscience, Warrington, Philadelphia) were moved by a focused laser beam and the Raman signal was collected as a function of bead position with respect to the center of the laser focus. The step resolution of the microscope stage in the horizontal plane was 1 μm. Vertical movement was advanced in 1.1 μm steps. Optimize the position to obtain the maximum Raman signal of the beads. Lateral resolution is determined by alternately measuring one Raman signal at 8 positions in the X or Y direction using the center position and a 1 or 2 μm step from the bead center. The intensity of the strong 1004 cm ^-1 polystyrene Raman band is measured as a function of the distance to the laser focus both in the planar and axial directions. The result in each direction is then fitted with a Gaussian distribution function, and the diameter of the focused ray is determined from the entire width at half width (FWHM). The diameter is about 1 μm for both lateral directions, while the axial direction is 2 μm. The sampling volume is calculated to be about 2 μm ³ .

Data analysis of Raman spectra from morphological structures is performed using Microcal Origin software (version 4.10, Clecom, Birmingham, UK). This analysis includes color variability in filter / spectrometer / CCD detector systems with cosmic ray removal, wavenumber shift calibration using the spectral characteristics of toluene (Mallinckrodt Specialty Chemicals, Paris, Kent.) And a calibrated tungsten light source. Compensation. Thereafter, the tissue spectra were corrected for the contribution of background BaF ₂ or MgF ₂ by subtraction of appropriate spectrum, and for tissue fluorescence by subtracting the 4-position of the polynomial fit to the spectrum by the method of least squares minimization (LSM) to correct.

Each morphological structure spectrum is
r _total = C ₁ r ₁ + C ₂ r ₂ + C ₃ r ₃ (4)
[Where r is a Raman spectrum] using a simple linear combination reference to generate a partial fit contribution (C ₁ ) for each of the 12 biochemical components Model with a Raman shift range of 1800 cm ⁻¹ . The Raman shift range of 700-1800 cm ⁻¹ is chosen because this range contains most of the spectral information.

  Using reagent grade commercial chemicals (Sigma, St. Louis, MO), proteins (type III collagen, elastin, actin, myosin and tropomyosin), non-esterified cholesterol (cholesterol monohydrate), cholesterol esters (cholesterol) Use as a fundamental spectrum of 12 biochemical components including linoleate), phospholipids (phosphatidylcholine), triglycerides (triolein), carotenoids (β-carotene) and calcium salts (hydroxyapatite calcium and calcium carbonate) Get the Raman spectrum for. These twelve biochemical components are selected as the most common Raman-active biochemical species found in normal arterial tissue and atherosclerotic plaques. In addition, a similar set of biochemical components provided a good fit of the reference data to the spectra observed in previous macroscopic histology. The Raman spectra from these chemicals are recorded in a manner similar to the Raman spectra from the morphological structure.

Reference data components cannot be compared by chemical weight. This is because the actual concentration of biochemical substances in various morphological structures is unknown. Accordingly, the intensity of the spectral features of 1440-1455 cm ⁻¹ (representing the binding of CH ₂ bonds in proteins and lipids) is set to unity. The Raman spectra of β-carotene, calcium carbonate and hydroxyapatite calcium, which lack spectral features in this region, are set singly for spectral features of 1159, 1080 and 961 cm ⁻¹ , respectively. This reference thus provides information about the contribution of the relative fit of these chemical components to the Raman spectra of various morphological structures. The contribution contribution of each biochemical component fit is expressed as the largest (ie 1) fraction.

FIG. 28A is a photomicrograph of an unstained coronary artery section showing an internal elastic plate seen under phase contrast. This structure is examined at a total of 54 sites in 21 coronary artery samples. Nine of these samples collected spectra from an external elastic plate. FIG. 28B shows Raman spectra of six different internal elastic plates. The 1664 and 1264 cm ⁻¹ bands are attributed to vibrations of amides I and III, respectively, of structural proteins such as elastin and collagen. The strong band at 1449 cm ⁻¹ can be assigned to the CH ₂ and CH ₃ binding modes of the protein, while the 1004 cm ⁻¹ band is due to phenylalanine. The 1336 and 1104 cm ⁻¹ bands are attributed to desmosine / isodesmosine and are specific for elastin. The 933 and 855 cm ⁻¹ bands can be assigned to the CC extension mode of proline and are present in collagen. These results indicate that the internal elastic lamina contains both elastin and collagen. Furthermore, visual inspection shows that these spectra show little variation between structures, indicating that the biochemical composition of the internal elastic lamina is very consistent within and between coronary artery samples. The spectrum obtained from the external elastic plate is the same as that obtained from the internal elastic plate.

FIG. 29A is a phase contrast micrograph showing collagen fibers (approximately 10 μm in length, approximately 2 μm in diameter) in the connective tissue of the outer membrane. A total of 17 collagen fibers in 10 samples were examined. FIG. 29B shows Raman spectra from collagen fibers from three different arterial samples. Again, these spectra collected from the adventitia (a, b) on visual examination show little variation from fiber to fiber within or between coronary artery samples, and the fibers of atherosclerotic lesions of the intima Identical to that taken from collagen fibers in the protective coating (c). These spectra are also very similar to those from the elastic plate, except for the absence of desmosine and isodesmosine bands (specific for elastin). The collagen fiber spectrum also contains a highly prominent hydroxyproline contribution (855 cm ⁻¹ ) specific for collagen.

FIG. 30 shows four Raman spectra recorded from smooth muscle cells in the media of normal and atherosclerotic coronary artery samples of one preferred embodiment of the present invention. A total of 32 spectra were recorded from 10 coronary artery samples. On visual inspection, no significant difference was observed between spectra taken from individual smooth muscle cells. The main spectral features in the smooth muscle cell spectrum are similar to those observed in the elastic plate and collagen fiber spectra, and 1660 and 1628 cm ⁻¹ (oscillations of amides I and III, respectively), 853, 940, The protein bands of 1034, 1336 and 1451 cm ⁻¹ (C—C or C—H bend), and 1004 cm ⁻¹ (phenylalanine) dominate. The major differences between the spectrum of protein-dominated smooth muscle cells, elastic lamina and collagen fibers are the bands of phenylalanine (1004 cm ⁻¹ ), desmosine / isodesmosine (1336 and 1104 cm ⁻¹ ) and amide II (1268 cm ⁻¹ ). It is in the fluctuation of intensity.

FIG. 31 shows an example of a Raman spectrum collected from fat cells (adipocytes) in the outer membrane of one preferred embodiment of the present invention. A total of 8 adipocytes were examined from 6 coronary artery samples. The spectra collected from adipocytes are very similar and CH ₂ // which shows an ester band (1747 cm ⁻¹ ), an unsaturated carbon / carbon band (C═C; 1654 cm ⁻¹ ), and a triglyceride in combination. Dominated by the CH ₃ bands (1440 and 1301 cm ⁻¹ ).

FIG. 32A is a phase contrast micrograph of foam cells in the intima of an atheroma plaque. Individual lipid droplets in these cells can be easily identified. A total of 30 foam cells in 8 coronary artery samples were examined. FIG. 32B shows the Raman spectra from three foam cells (ac). Although similar in visual inspection, these spectra show variations among foam cells that are larger than those of collagen fibers, internal and external elastic lamina and smooth muscle cells. More specifically, the foam cell spectrum is distinctly different from the dominant spectrum of elastic lamina, collagen fibers and smooth muscle cell proteins, especially for multiple bands below 1100 cm ⁻¹ . The bands at 702, 878, 923, and 957 cm ⁻¹ can be assigned to the steroid nucleus of both unesterified (free) cholesterol and cholesterol esters. The strong bands at 1671, 1439, 1299 and 1270 cm ⁻¹ are due to C═C stretching and CH ₂ / CH ₃ bending modes. The presence of bands at 1735 and 1026 cm ⁻¹ (specific for cholesterol esters) and 1058 and 1328 cm ⁻¹ (specific for free cholesterol) indicate that these foam cells contain both esterified and unesterified cholesterol. Show. As previously discussed, reduced CF _{NCR in} non-calcified plaques implies reduced plaque stability. The bands at 719 cm ⁻¹ (symmetrical choline extension), 762 cm ⁻¹ (symmetrical O—P—O extension) and 878 cm ⁻¹ (asymmetrical O—P—O extension) indicate the presence of phospholipids and 1523 And 1160 cm ⁻¹ indicate the presence of β-carotenoids. However, the foam cell spectrum lacks the triglyceride bands at 1747, 1654, 1440 and 1301 cm ⁻¹ found in outer membrane adipocytes.

  In total, 31 necrotic central areas of 16 coronary artery samples were examined. FIG. 32B also shows two examples of Raman spectra (d and e) collected from necrotic centers. Similar to the foam cell spectrum, there is some variation from structure to structure within the necrotic center. However, the average spectra from foam cells and necrotic centers are very similar, indicating that the chemical content of both morphological structures is very similar.

FIG. 33 shows an example of a Raman spectrum taken from cholesterol crystals of different sizes at the necrotic center of the plaque. A total of 7 coronary artery samples were examined for cholesterol crystals. The main spectral features of the spectrum of cholesterol crystals are 1668 cm ⁻¹ (C═C stretching), 1443, 1238 and 1274 cm ⁻¹ (CH ₂ , CH ₃ bending) and 1176 and 1085 cm ⁻¹ (CC stretching). is there. The spectral features below 1000 cm ⁻¹ are attributed to steroids and indicate the presence of unesterified cholesterol. There was slightly more variation between the spectra from the individual crystals, mainly due to variations in the band intensity that implied the difference in the ratio of free cholesterol to esterified cholesterol in the crystals themselves or in the components of the tissue surrounding the crystals.

In the necrotic center region, yellow crystals could sometimes be identified under phase contrast. Figures 33 (d and e) show the Raman spectra of two such crystals from two different coronary artery samples. A total of 7 of these crystals from 3 samples were examined. The main spectral features are at 1523 and 1160 cm ⁻¹ , which are implied by C—C and C═C elongation and β-carotene. Given the presence of the 1449 and 956 cm ⁻¹ bands, these yellow crystals appear to also contain some structural proteins and cholesterol esters.

FIG. 34A is a photomicrograph of atherosclerotic plaque containing calcification. A total of 15 calcium mineralizations in 6 coronary artery samples were examined. A Raman spectrum representing different stages of calcification in two atherosclerotic plaques is shown in FIG. 34B. The main features of these spectra are 1071 and 959 cm ^{− which} are attributed to CO ₃ ²⁻ (symmetric) / PO ₄ ³⁻ (asymmetric) and PO ₄ ³⁻ (symmetric) elongation, implying calcium carbonate and hydroxyapatite calcium, respectively. ¹ band. Larger calcium mineralizations (FIGS. 34B, a and b) show different spectral features than those of micropoint calcifications of necrotic centers (FIGS. 34B, c). The main difference is the presence of additional features in the spectrum of punctate calcium mineralization attributed to lipids and / or phospholipids (1433 cm ⁻¹ ) due to the most likely surrounding necrotic centers.

  Using the fundamental spectrum of pure chemicals as specifically illustrated in FIG. 35, the spectra of the individual morphological structures were fitted to the biochemical model. Each figure in FIGS. 36A-H shows one Raman spectrum of the morphological structure and the result of a least squares minimization fit of the biochemical model. Residuals (data-fit) are shown on the same scale. Only the foam cell fit results are shown because the Raman spectra from foam cells and necrotic centers were very similar. Judging from the observed residual fit to the spectrum, which is in the order of noise and does not show a consistent pattern between the spectra, the Raman spectrum of each morphological structure (Figures AH) has 12 raw spectra. It is well described using the fundamental spectrum of chemicals.

  For each morphological structure examined, the contribution of each biochemical component was determined. Figures 37A-H confirm that each morphological structure has a characteristic biochemical composition. In general, each morphological structure is composed primarily of one or two major biochemical components combined with one or more less abundant biochemical components.

  The inner and outer elastic lamina (FIG. 37A) is composed mainly of elastin with less collagen component, while the collagen fibers of both normal artery and atherosclerotic lesion fibrous capsule (FIG. 37B) have mainly less elastin component. Consists of collagen. Smooth muscle cells (FIG. 37C) were almost completely modeled by actin and few tropomyosin components. Myosin did not contribute at all.

Outer membrane fat cells (FIG. 37D) contain triglycerides (triolein) almost exclusively with a small contribution of phospholipids (phosphatidylcholine). In contrast, foam cells and necrotic centers (FIG. 37E) are associated with less contribution of collagen, phospholipids and β-carotene, mainly in a ratio of about 2: 1 cholesterol ester (linoleate) and free cholesterol (monohydrate). Product). Foam cells appear to be similar in current data and are indistinguishable from necrotic centers based on their biochemical composition. However, they can be distinguished in the following evaluation. For example, spectral features of approximately 1750 cm ⁻¹ can be used to distinguish foam cells from necrotic centers. Cholesterol crystals (Figure 37F) contain free cholesterol and cholesterol esters in a ratio of about 3: 1. Yellow crystals (FIG. 37G) consist almost entirely of β-carotene with a small contribution of cholesterol. This may indicate that these crystals are in fact cholesterol crystals containing a high concentration of β-carotene. Calcium mineralization (FIG. 37H) is composed primarily of hydroxyapatite calcium with a small contribution of collagen, triglycerides and calcium carbonate.

  The presence of foam cells and other inflammatory cells may also play a role in plaque instability. Thus, morphological factors such as the presence of crystal-free cholesterol or foam cells may be as important as the biochemical composition in determining the stability and progression of atherosclerotic plaques.

  As shown in FIG. 36, the biochemical model of FIG. 35 fully describes the spectrum of each morphological structure, indicating that most of the essential biochemical components are included in the reference. Means. The biochemical composition of each structure as shown by the contribution of the fit of the biochemical base spectrum to the spectrum of the morphological structure is very consistent (FIGS. 36A-H). The greatest biochemical variation was found in foam cells, necrotic centers, cholesterol crystals and calcium mineralization. Biochemical variations in both calcium mineralization, cholesterol crystals and β-carotene-containing crystals may be due to differences in their progression stage (as reflected by size). The cause of biochemical fluctuations within the foam cells and necrotic centers (differences in collagen, β-carotene and cholesterol esters) is less obvious. Variations in lipid composition of atherosclerotic plaques at various stages of progression have been previously described in in vitro studies of homogenized or extracted tissue and cultured monocyte-derived foam cells. However, these biochemical data are the result of an atherosclerotic plaque component analysis of an in situ or in vitro cell culture model without the disruptive effects of tissue preparation. More detailed in situ Raman microspectroscopy studies of foam cells and necrotic centers in atherosclerotic plaques at various stages of disease may help further elucidate the origin of this variation.

  Although biochemical models have provided valuable information regarding the biochemical composition of microscopic cellular and extracellular morphological structures, it has its limitations. One of the main limitations of this model and / or reference was specifically explained by the fit of the smooth muscle cell spectrum. Previous in vitro studies have shown that smooth muscle cells comprising the majority of the media of muscular arteries such as coronary arteries contain approximately three times more actin than myosin, but approximately equal amounts of myosin and tropomyosin It was shown to contain. However, the contribution of fit in the biochemical model indicated that smooth muscle cells contained actin almost exclusively with a small amount of tropomyosin and virtually no myosin. These unexpected results may be due to a conformational difference in the spectroscopic characteristics of myosin between tissue extracted myosin and in situ intracellular myosin. In addition, as seen with glycosaminoglycans, the contribution of weak Raman scatterers may be underestimated.

  The observed variation may also be due to the contribution of biochemical compounds not included in the reference. For example, only one class of collagen was included. This should not be a major concern since little difference was observed in vitro in the Raman spectra of different classes of collagen. However, there may be significant changes in the Raman spectrum of collagen in vivo due to increased cross-linking as atherosclerotic lesions progress.

  Despite these limitations, the results of previous quantitative Raman spectroscopy biochemical analysis of normal and atherosclerotic arterial tissues using the same biochemical model are sufficiently comparable to standard analytical techniques. did. Previous studies have also shown the basis of a diagnostic algorithm in which the biochemical analysis of these quantitative Raman spectroscopy has accurately classified arterial tissue as either non-atherosclerotic or calcified or non-calcified atherosclerotic plaque It can also be used as The results of the preferred embodiments of the present invention show that modifications of the biochemical model can be used to perform a relative comparison of cellular and extracellular morphological components of normal and atherosclerotic arterial tissue. Furthermore, another preferred embodiment shows that these relative morphological comparisons can be used as the basis of an algorithm that allows diagnosis of coronary atherosclerosis. This develops the morphological analysis of quantitative Raman spectroscopy with the goal of accurately classifying normal arteries and atherosclerotic plaques ex vivo and in the future predicting plaque stability and disease progression in vivo Is the first stage.

  Using the biochemical model reference of the preferred embodiment, individual microscopic (such as foam cells and necrotic centers) in intact arterial tissue that cannot be isolated or purified using conventional analytical techniques A confocal Raman microspectroscopy method to be used to perform in situ biochemical analysis of morphological structures is specifically described. In addition, the various morphological structures have characteristic Raman spectra, which vary little from structure to structure or from artery to artery as expected, and ex vivo normal and atherosclerotic coronary artery morphology Can be used as the base spectrum of the morphological reference for performing relative comparisons. This non-destructive technique ultimately evaluates plaque stability and disease progression in humans, as well as atherogenesis in animal models and lipid metabolism in cell cultures in vitro. May be used for research.

  Embodiments of the present invention interpret the Raman spectrum with respect to morphology. For example, the Raman spectrum can be associated with morphological structures, such as foam cells that can be associated with a particular compound. Furthermore, the number of spectra can be reduced, for example from a large number of chemical spectra to as few as eight unique spectra associated with morphological structures, thereby reducing fit errors. Diagnostic methods that can be used to identify and monitor vulnerable plaques using the fiber optic catheter system of the present invention include chemical composition, morphological structure, thickening of the intima layer and overlying collagen layers. Includes the use of information about thinning. A preferred embodiment involves the determination of collagen depth by measuring the percentage of collagen. In addition, the presence of calcification is monitored and any edges are identified and located with respect to collagen as an indicator of potential rupture and clots. As previously discussed, the reduced partial fit contribution of collagen fibers in non-calcified plaques is an indicator of unstable plaques.

  A preferred embodiment implements an optical system design that fully utilizes the throughput of the system by characterizing the Raman distribution from the tissue. The embodiment optimized the collection efficiency, minimized noise, and resulted in a small diameter highly efficient Raman probe capable of collecting high quality data in 1 second. The performance of the embodiments has been tested by simulations and experiments using tissue models and several in vitro tissue types, indicating that these embodiments can advance Raman spectroscopy as a clinically feasible technique.

  Raman spectroscopy has proven to be a valuable and accurate tool for diagnosing disease and studying biological tissues. Laser excitation is used to provide detailed information about the vibration and rotation of molecular bonds. Since each chemical moiety in the sample has a unique molecular structure, its composition can be evaluated by spectroscopic analysis of inelastically scattered excitation light.

  In vitro studies have established the medical potential of Raman spectroscopy. In fact, many diseases are being considered because the Raman spectrum has the ability to provide specific information about a wide range of chemical and morphological components that cannot be obtained with other spectroscopic methods. For example, the rupture of unstable atherosclerotic plaques in the coronary arteries is responsible for the majority of fatal myocardial infarctions. It has been established that the likelihood of plaque rupture is related to chemical composition, and Raman spectroscopy may have the unique ability to identify most criminal lesions based on this. If this method can be successfully implemented in clinical practice with real-time analysis, direct diagnosis of cardiovascular disease and guidance of effective treatment, or characterization of malignant tumors and their completeness by monitoring the margin of resection. It can be used for a wide range of applications such as confirmation of resection.

  Many medical applications require remote sampling using optical fibers where anatomical considerations severely limit the size of the probe and fiber bundles. For example, the ability to clinically assess coronary atherosclerosis and breast cancer is approximately 2 mm in diameter or less so that they can be incorporated into standard cardiovascular catheters or configured for optical needle biopsy Need a probe that is In addition, the data acquisition time must be limited to a few seconds at most.

  With regard to coronary atherosclerosis, the detection of fragile atherosclerotic plaques is critical for the prediction and prevention of cardiac events. These vulnerable plaques are present in clinically asymptomatic blood vessels and are characterized by the presence of biochemical changes, ie foam cells, lipid pools, inflammatory cells, thin fibrous capsules less than 65 μm thick and thrombosis obtain. A preferred embodiment of the present invention uses a Raman probe to provide quantitative biochemical information and morphological analysis regarding the presence of the aforementioned factors to characterize vulnerable plaque. The use of Raman spectra is spectroscopically advantageous with the use of a narrow vibration band, is chemical specific and rich in information.

  For diagnostic purposes, the use of Raman spectroscopy is advantageous because no biopsy is required; it also directly measures low concentrations of the molecule to provide its chemical composition and morphological characteristics. Raman spectroscopy can be used to assess plaque stability, monitor disease progression, and assess therapeutic intervention by confirming plaque regression and restenosis.

  There has been great progress in the design of fiber optic probes over the past decade, indicating that Raman spectroscopy is a potentially useful clinical technique. Low-OH fused silica has been determined to be the optimal fiber support for use in the near infrared. The need for a proper optical filter has been established and numerous probe configurations are being explored.

  Whether in vivo studies, often using commercially available probes, are restricted to skin and other readily accessible organs or have an optical system configuration that is not optimized for tissue studies It was either. In addition to other difficulties such as choice of excitation wavelength, non-optimal probe designs have resulted in acquisition times that are too long for practical clinical use. Thus, an important impediment to realizing the clinical potential of Raman spectroscopy is the small diameter fiber optic Raman probe that can deliver excitation laser light to in vivo tissue and efficiently collect Raman scattered light Development.

  A preferred embodiment of the Raman probe includes a small (approximately 2 mm) probe that is flexible to reach remote organs. The probe is capable of collecting high signal-to-noise (SNR) spectra in approximately 1 second for accurate clinical application of the spectral model used for analysis. This is accomplished with a safe level of laser exposure and by minimizing the total noise source while maximizing throughput and efficiency.

  There are several noise sources that are minimized in the preferred embodiment. Detector dark charge and read noise are reduced by using a low temperature cooled charge coupled device (CCD) detector. The choice of excitation wavelength also affects the SNR. Excitation at 785 nm, as done in many Raman probe designs, results in a minimum 4-fold increase in tissue fluorescence when compared to 830 nm excitation. This increased fluorescence adds significant shot noise to the data. Although longer excitation wavelengths further reduce tissue fluorescence, the Raman cross section is simultaneously reduced. This is because they depend on the excitation frequency for the fourth force. Furthermore, the absorption coefficient of water increases rapidly at longer wavelengths, thereby decreasing the depth of penetration and attenuating the signal. Excitation wavelengths greater than 830 nm also interfere with the use of CCD detectors, thereby compromising the ability to collect complete Raman spectra with a single exposure.

  One major noise source specific to fiber optic Raman probes is the long recognized problem of the spectral background generated by the delivery and collection fibers themselves. This background is several orders of magnitude greater than the signal from the tissue site being examined. It consists of Raman light from a fused silica core, impurities and fluorescence from dopants used to produce fibers with a specific numerical aperture (NA), and signals from various jacket materials. The laser light in the delivery fiber creates a strong fiber background with a long path length (typically 3 to 4 meters) across the fiber. This fiber spectrum is scattered from the tissue surface and collected by the collection fiber along with the tissue Raman spectrum. The background often masks the tissue Raman signal generated from a sample of only about 1 mm due to the relatively short penetration of light into the tissue. Laser light backscattered from the tissue also enters the collection fiber, creating additional fiber background and further degrading the quality of the tissue spectrum reaching the detector. In addition to obscuring and distorting the spectrum of interest, a strong fiber background adds short noise to the signal. This noise can often be larger than the tissue Raman band, and it is therefore necessary to remove as much background as possible by using an optical filter.

  Another consideration in the preferred embodiment probe design to optimize throughput and maximize collection efficiency has two components. The first relates to the inherently weak nature of the Raman effect. Only about one out of every billion excitation photons is converted to Raman photons. Therefore, it is critical to design high-throughput optics to collect signals with sufficient SNR for accurate analysis in clinically realistic time frames. The second component relates to the optical characteristics of the tissue itself. The signal of interest is directly attenuated via absorption of the excitation laser and the generated Raman light. Furthermore, the collection of Raman light is confused by light scattering that diffuses photons extensively over large areas and angles. Thus, simple probe designs used to examine other types of spectrometers or non-turbid samples are not ideal for this application.

  Preferred embodiments remove most of the fiber optic background, use 830 nm excitation, and maximize signal collection from Raman sources generated in the tissue in a few seconds or less (1 or 2 seconds). Includes fiber optic Raman probes that allow data collection and operate at a safe fluence level while limiting the rigid distal tip to less than a few millimeters in length and less than about 2 mm in diameter.

  The analysis shows that the spectral background of the detected fiber is equally occurring in both the excitation and collection fibers. Two different filters (ie, one for delivery and one for collection) are required at the distal end of the probe to suppress unwanted signals. The delivery fiber is terminated with a short wavelength pass or band pass filter that propagates the laser excitation light while blocking the longer wavelength spectral background from the fiber. The collection fiber is preceded by a long wavelength pass or notch filter that propagates the Raman spectrum of the tissue while blocking the laser light backscattered from the tissue. Filters perform these functions over a range of angles corresponding to the NA of the fiber on which they operate.

  In order to adapt the filter to the distal end of the probe, a preferred embodiment includes a filter module. The module also consists of a rod carrying a short pass dielectric filter coating on a plane, fitted in a tube carrying a long pass dielectric coating on a plane. They are separately coated with their respective filters prior to assembly and, for example, Research Electro-Optics, Inc., Boulder, Colorado. Rods and tubes made from either sapphire or fused silica, which are secondarily processed by, are used in a preferred embodiment. The rod is wrapped or covered with a thin sheet of metal to prevent crosstalk by providing optical isolation between components. The module is placed at the distal end of the probe between the fiber bundle and the collection optics. The performance curve of the filter used in the preferred embodiment of the Raman probe is shown in FIG. 5 (0 cm−1 = 830 nm). Peak propagation is typically greater than 90%, while unwanted light rejection is greater than 96%.

In general, the throughput (ie Etendue) of an optical system is given by the product of its collection area (A) and the projected solid angle (Ω ′), where
Ω ′ = πsin ² (θ) (5)
And the half-angle value (θ) of the collection measured with respect to the optical axis. Ignoring the limitations of fiber coupling along with the loss of reflection and propagation of all optical components (a factor that can be easily optimized), the collection capacity of the system is limited by its maximum limiting factor throughput. And this amount is preserved throughout the system.

In an ideal spectroscopy system, the throughput is determined by the spectrometer / CCD detector. In a preferred embodiment, the spectrometer has an NA = 0.278, such as Ω ′ _D = 0.225 sr (Holospec f / 1.8i, Kaiser Optical Systems, Inc., Ann Arbor, Minn.). It is also connected to a back-illuminated deep-depletion CCD detector (Spec-10: 400BR, Roper Scientific, Trenton, NJ) that also has a height of 8 mm and thus does not compromise throughput. In order to achieve sufficient spectral resolution (approximately 8 cm ⁻¹ ) for biological Raman spectroscopy, a slit width of 0.2 mm is used at the entrance of the spectrometer. Therefore, the maximum area of collection AD = 1.6 mm ² , resulting in a theoretical maximum throughput of A _D Ω ′ _D = 0.360 mm ² −sr for the detection system. The detection of light from a Raman light source is limited by this product, and the collection optics are designed to preserve system throughput.

  Diffuse scattering in tissue results in the emission of Raman light over a large area with a specific distribution and a solid angle of 4π. Optimization of signal collection from such light sources requires two steps. First, the distribution of Raman light emitted from the turbid medium is determined. This distribution defines the potential light collection efficiency for a given Raman source within throughput constraints as the collection area (or angle) is varied. These characteristics are used to determine the optimal tradeoff between collection solid angle and area so as to maximize the efficiency for the design of the collection optics.

  For fiber optic probes, the optics at the distal end of the probe are also designed to convert Raman scattered light for efficient coupling to a collection fiber that is chosen to have the same NA as the spectrometer. . Furthermore, in order to optimize signal collection, it is necessary to maximize the area of the distal probe tip utilized for the collection fiber. This is achieved by using a single central excitation fiber surrounded by a tightly packed ring of as many collection fibers as can be accommodated by the spectrometer / CCD and incorporated into the probe diameter. Achieved. The circular bundle of fibers at the distal end is then reshaped into a straight line at the proximal end for connection to the spectrometer.

  Proper selection of the excitation optics is also critical. The intensity of the background generated in the fused silica optical fiber is proportional to the square of NA, but is relatively independent of the core diameter. Therefore, although it is desirable to match the NA of the collection fiber to the spectrometer, it is preferable to use an excitation fiber with a smaller NA. Lowering this NA also provides reduced beam divergence at the distal end, thereby improving short pass filter performance. The results show that fibers with very small NA (0.12) represent a substantially increased broadband fiber background, possibly generated from the doping material used in the cladding. Therefore, it is most effective to use moderate NA (0.22). In that case, an appropriate excitation fiber diameter may be chosen to ensure a safe illumination fluence in the tissue while limiting the spot size to facilitate efficient collection.

  Determining the optimal collection shape requires characterization of the distribution of Raman light from the tissue. This light source has a predetermined brightness B (r, θ) emanating from the surface with a convoluted dependence on both the light source radius r and the launch angle θ. The total amount of Raman light available for collection from the organization is

Where dA = (2πr) dr and dΩ = 2πsin (θ) dθ. The integral is carried over to the total area and solid angle of the light source, the latter being limited to 2π for backscattering geometry.

B (r, θ) is independent of the individual distributions B ₁ (r) (a function of only the radial distance from the excitation light) and B ₂ (θ) (depends only on the angle from the normal surface) Each can be measured experimentally. Using these distributions,

[Wherein I ₁ (r) and I ₂ (θ) are integrated radial and angular distributions originating from the tissue]
The light emitted from the light source is approximated as follows.

  The efficiency of the optical system for this Raman light source is calculated by integrating the radial and angular luminance over the characteristics of the collection optical system normalized by the total light emitted from the light source and constrained by throughput conservation. This resulting efficiency curve

Is used to guide the design of the probe optics by specifying the optimal trade-off between collection radius and solid angle. Angular efficiency η ₂ (θ) can be converted to a function of radius η ₂ (r (θ)) by using throughput conservation, resulting in a single variable function η _T (r) of total collection efficiency.

  Guided by the application to diagnose atherosclerosis, the radial and angular distribution of Raman light from arterial tissue was examined. Normal arterial tissue was typically used to represent the weakest signal compared to other arterial disease states due to its optical properties (eg, scattering and absorption coefficients) and relative Raman cross sections.

The characterization of the spatial distribution of Raman light B ₁ (r) was determined in a preferred embodiment of the invention. Briefly, the 830 nm excitation was focused and directed to the sample via a small prism. The excitation spot diameter was approximately 100 μm and the sample was kept in a quartz cuvette containing phosphate buffered saline (PBS, pH 7.4). Backscattered light was collected by a Cassegrain objective in the 14 ° to 33 ° angular range. This entire range of Raman light was collected, collimated and notch filtered to eliminate Rayleigh scattered light. A single 100 μm central optical fiber collected images of spatial regions separated by translating the beam laterally with a step size of 250 μm. Explaining the magnification from the objective lens, this corresponds to a step size of approximately 104 μm on the sample surface. The fiber was connected to an f / 1.8 spectrometer and the light was dispersed on a CCD detector. The intensity of the 1450 cm ⁻¹ Raman band from —CH ₂ bending mode was integrated, normalized to the maximum signal, and plotted as a function of radial distance from the excitation source. The results on either side of the excitation beam are nearly symmetric, and the average B ₁ (r) for the two sides is presented in FIG. 39B (◯).

  Optimally fitting this radial distribution to a multi-Gaussian distribution (FIG. 39B, line),

(R is mm). The narrow distribution represented by the first term is likely to relate to the Raman light produced by the ballistic laser light that produces the strongest Raman energy distribution. The other term is responsible for diffuse light that is also affected by the layer structure of arterial tissue. This data is then integrated and normalized to obtain the radial collection shown in FIG. 39C as a function of distance from the excitation beam, along with a least squares fit showing the expected Gaussian distribution (line). The efficiency η ₁ (r) is determined.

The angular distribution was determined using a slight modification of the Raman system of open-air optics. Rather than measuring a separate angular distribution B ₂ (θ), I ₂ (θ) was measured directly. Briefly, 830 nm excitation light was directly incident on the tissue by a small mirror between the collection lens and the sample. Backscattered Raman light is collected by an f / 1.2 camera lens that collimates the beam prior to notch filtering and then focused on the f / 4 spectrometer via an f / # matching lens for detection by a CCD detector. It was. The excitation light was concentrated to approximately 100 μm diameter and the collection radius was approximately 1 mm. The collection lens was preceded by an aperture iris that allows for a direct measurement of the collection angle variation and the integrated angular Raman distribution. FIG. 40B plots the experimentally determined η ₂ (θ) (◯) of Raman light from the tissue. The distribution is flat around 20 ° due to the limited angle collected by the lens. The light exiting the tissue generally follows cos (θ) or Lambertian dependence, where θ is an angle with respect to the normal surface. The integrated distribution should therefore have a (line) sin ² (θ) dependence also plotted in FIG. 40B, showing a reasonable good agreement between experiment and theory over the range of collection angles. .

Taking the product of η ₁ (r) and transformed η ₂ (θ) → η ₂ (r (θ)) from FIGS. 39C and 40B yields an efficiency curve η for this particular combination of system throughput and tissue. _{Resulting in T} (r) (FIG. 41). An optimal efficiency of 8.62% occurs with a collection radius of 0.191 mm and a corresponding 90 ° angle. This indicates that it is preferable to collect a complete angular range of Raman light from the strongest illumination region rather than collecting a smaller range of angles while stretching to a weaker tailing at the edge of the distribution.

  After that, the results of the Raman light source characterization study are shown in Zemax v. 10.0, Focus Software, Inc. of Tucson, Arizona. Incorporated into the optical design criteria provided by, determines the optimal optical system for maximum signal collection and conversion of the collected light for efficient coupling to an optical fiber. Although it is possible to design sophisticated optics to operate near these specific parameters, the spatial constraints imposed by a given application appear to be extremely difficult to build. In fact, studies with Zemax have shown that the use of a simple ball lens provides reasonable performance when using a high refractive index. Several supports have been investigated and it has been determined that the highest refractive index glass is highly fluorescent due to the doping material. However, sapphire with a refractive index of 1.77 expects a wide angle collection. Sapphire does not exhibit fluorescence, has only a single sharp Raman band, and is optically transparent through a very broad wavelength range. In addition, sapphire is extremely hard, thus making it an excellent choice for multi-use Raman probes.

  The resulting Raman probe design is presented in FIGS. 4A and 4B. The left side shows a longitudinal view of the probe tip, while the right side shows a cross-sectional view at the fiber / filter interface level. There is a central excitation fiber with an aluminum jacket for optical isolation to prevent crosstalk with the collection fiber. This fiber is placed in registration with the short path excitation rod. The rod is placed inside the long pass collection filter tube and the two are optically isolated by a metal sleeve. The excitation fiber is then buffered to ensure proper alignment of the registered collection fiber with the center of the long pass filter tube. The central excitation fiber has a 200 μm core with 0.22 NA. The collection fiber also has a 200 μm core, but has a NA of 0.27 (closely matched to that of the spectrometer). The filter is fixed to the fiber with a refractive index matching optical cement and the entire fiber bundle / filter module is wrapped in black Teflon for connection and protection. The probe length is 4 meters.

  The filter rod and tube are 1 mm long to ensure proper spatial placement of the sapphire ball lens. This shape handles two considerations. First, in this fiber and lens separation, the excitation light is roughly collimated and not concentrated in a dense spot on the tissue, thereby reducing the energy density incident on the sample and preventing possible damage . Second, since the ball lens converts the large angular distribution exiting the tissue into a well collimated beam that is within the NA of the fiber, excellent coupling of Raman scattered light to the collection fiber is ensured. The ball lens is secured to the pleated stainless steel tube with epoxy to ensure that no liquid leaks to the tip. The stainless steel tube is then secured to the fiber bundle / filter assembly. There is no adhesive used on the inner surface to maximize the collection efficiency of the ball lens.

  The total diameter of this probe is below 3 mm. The factor limiting the current size is the diameter of the ball lens (2 mm to accommodate the entire width of the filter tube). The size of this filter was chosen because this shape was used to build a Raman probe with a collection fiber (2 rings totaling 27 fibers) that more fully utilizes the spectrometer's throughput. . In practice, a single ring probe can be used because it provides excellent signal collection and is much more flexible and easier to construct. Recent studies have shown that the probe diameter can be reduced without significantly reducing the collection efficiency. The diameter of the central collection rod was chosen to be 0.55 mm for ease of construction. All components of the probe are constructed from medical grade materials that can withstand standard cold gas ethylene oxide sterilization for surgical procedures.

  The design of the Raman probe was tested by simulation and experimentally. The simulation experiment was conducted using a Zemax probe model, and two aspects of probe design were examined. First, the excitation spot diameter was examined to ensure safety. Second, the collection efficiency for various Raman light sources was examined to determine probe performance over a range of conditions.

  The result of the simulation of the excitation spot size is shown in FIG. Two configurations were investigated: one with a probe placed in the air (O) and one with a probe immersed in a simulated tissue model (□; more likely clinical shape). The tissue model was constructed using the refractive index of water and the scattering properties of the artificial tissue, ie g = 0.9, mean free light path = 0.27 mm. As can be seen from the figure, when the probe is in the air, there is a slight concentration about 1 mm away from the lens to a full width at half maximum (FWHM) of about 175 μm. However, in the case of scattering, the beam begins to diverge immediately from the surface of the ball lens and does not fall below the FWHM of less than 200 μm. This spot size produces a fluence well below any reported damage threshold at the laser power and exposure time used in the clinical context. The data collection protocol is designed using approximately 100 mW excitation in less than 5 seconds, thus producing a fluence far below that typically reported.

  Collection efficiency was determined from the Zemax model in a similar manner (Figure 14). Various radius Lambertian light sources were placed in contact with the probe and the fraction of light exiting the proximal end of the collection fiber was measured. For these simulations, a model of a double ring probe with 27 collection fibers was implemented. This is because the throughput of the spectrometer is more fully utilized. Again, two situations were examined. The first had a light source in contact with the probe, but kept the lens in the air so that the outer surface of the ball lens did not experience any index matching (circle with solid line). The second also had a light source in contact with the probe, but now both the lens and the light source were immersed in the simulation tissue model described above (circle with a dashed line). The probe exhibits high collection efficiency (up to 35%) for small light sources, but efficiency declined as the light source became larger. The tissue model results were also excellent, yielding an efficiency of about 1.75 times less than when the probe was not index matched.

  An additional configuration of another preferred embodiment was also considered in which the measured Raman distribution from the light source was modeled and placed at the end of the probe. This resulted in a collection efficiency of 3.5% and 1.7% for the air and tissue interfaces, respectively. The maximum collection efficiency of 8.6% results from fully utilizing the throughput, however, by using optical fibers, only 53% of the spectroscopic slit area is used. Therefore, the maximum expected collection efficiency is 4.6%, indicating that the ball lens is operating near the peak. By multiplying the efficiency curve by the same 53% caused by the reduced throughput, it can be seen that the efficiency of the ball lens exhibits a collection radius of 0.27 mm corresponding to a collection angle of 45 °. Although there is room for improvement, this is excellent collection efficiency and ease of implementation is very practical.

  The performance of the Raman probe was experimentally tested in three ways. First, a variety of known Raman scatterers were examined to evaluate filter performance. Second, a tissue phantom was developed to assess the effects of scattering and absorption on signal and background collection. Finally, in vitro spectra of arterial and breast tissue were collected and evaluated with spectroscopic models.

  An illustration of the experimental system used in these studies is shown in FIG. Light from an 830 nm diode laser (Process Instruments, Salt Lake City, Utah) is collimated by two cylindrical lenses (c1, c2), directed through a bandpass filter (BP, Kaiser) and gold coated It was redirected by a mirror (M) and focused on a Raman probe excitation fiber by a 10 × microscope objective (Newport, Irvine, Calif.). A line in the line proximal to the collection fiber from the Raman probe parallelizes the light before it is notch filtered (NF), concentrated on the slit and re-parallelized for dispersion by the holographic gate (HG). To the f / 1.8 spectrometer. Finally, the dispersed light is focused on a liquid nitrogen cooled back-illuminated deep depletion CCD detector interfaced with a laptop computer. The slit (S) allows laser excitation from the probe during data acquisition. This reduces laser exposure to the patient and increases user safety. In addition, the laser is connected to a computer for more precise control of the laser power, monitoring of the laser power, and the incorporation of a feedback loop that automatically sets the power from the probe to a set level that is independent of the system's optical alignment. Interfaced.

The tissue phantom study was designed to mimic the range of artificial tissue scattering and absorption properties. Samples were monodisperse 1.03 μm latex microspheres for scattering (Duke Scientific Corp., Palo Alto, Calif.), Hemoglobin for absorption (Sigma, St. Louis, Mo.), and India Ink (Triangle Biochemical Sciences, Durham, NC). As well as a combination of deionized water. A stock solution of NaClO ₄ was added to the sample as a target Raman molecule at a constant concentration. By mixing various combinations of components, ClO with absorption and decrease scatter coefficients of constant volume and the respective 1.31,1.79 and 2.25 cm ^-1, and 22, 29 and 36cm ^-1 ₄ ^- Nine phantoms of concentration were produced. Concentrations to produce these optical properties were determined using a modified version of Bohren and Huffman's Mie standard.

The phantom was placed in a 2 "deep, 3/4" wide glass vial and the probe tip was immersed just below the surface of the sampling liquid. The sample was continuously circulated with a magnetic stir bar to prevent microsphere precipitation. Spectra were collected for a total integration time of 10 seconds using 100 mW excitation with a dual ring Raman probe. The 928 cm ⁻¹ band of ClO ₄ ⁻ was integrated to determine the probe's ability to collect Raman under these varying conditions. Probe background was assessed by examining the intensity of the maximum signal collected (approximately 420 cm ⁻¹ ). Evaluation of background alternative integrates the Raman band of 750 cm ^-1 which are caused by the Raman band or sapphire ball lens of 800 cm ^-1 from the background quartz. All of these produced similar results.

  Finally, in vitro tissue samples were examined with a single ring Raman probe using 100 mW excitation power and acquisition times ranging from 1 to 60 seconds. Aortic tissue samples were collected post mortem, while breast samples were collected during surgical resection. Samples were snap frozen in liquid nitrogen immediately after collection and stored at -85 ° C until the time of examination. Samples were passively warmed to room temperature in a PBS bath prior to testing. The spectrum was corrected for the spectral response of the filter and CCD using a tungsten white light source. The remaining fiber background was removed by subtracting the signal generated by directing excitation light on the roughened aluminum surface. Tissue fluorescence was removed by subtracting the fifth polynomial. Finally, the spectrum was applied to a spectroscopic model developed in the laboratory. Breast data was fitted to the model described below, while the arterial spectrum was fitted to the morphological model. Residuals are calculated as data-fit and are shown on the same scale.

For biological tissue spectroscopy, the Raman characteristic of 600 to 1800 cm ⁻¹ is important. FIG. 44 shows the Raman spectrum of filled BaSO ₄ (a well-characterized Raman scatterer). Unlike liquid samples, which typically produce little detected fiber background due to the presence of minimal backscatter, BaSO ₄ is highly reflective when packed and strong in unfiltered probes Create a fiber background. This spectrum shows the effectiveness of the filter module and the optical isolation of the probe because there is little detectable feature of the fiber background other than a slightly increased tilting background above 550 cm ⁻¹ . Even a strong silica band of 800 cm ⁻¹ is not recognizable.

  The results of the tissue phantom examination are presented in FIG. Signals from perchloric acid (circles and solid lines) and probe background (squares and dashed lines) are plotted as a function of transport length. The plotted line draws a constant absorption and is drawn to show how signal collection increases as scattering increases. Conversely, the collected signal decreases with increasing absorption for a given scatter value. Similar trends are seen for both fiber background and Raman signal, however, the effect of scattering is more dramatic for background.

  46A and 46B are a comparison between the performance of a single-ring Raman probe and the experimental system described above by looking at a normal aorta (a tissue type that shows very little variation from site to site). The experimental system is an open optical system configuration that is not constrained by the demands of micro-optical systems, and has been used to develop many useful Raman spectral models.

FIG. 46A shows data taken with open optics (dots) and single ring Raman probe (lines) with equivalent excitation power and acquisition time. The data is corrected for the spectral response of the system and the gain of the CCD detector. Nevertheless it is difficult to separate the Raman bands on fluorescence background tissue, the band observations of 1004 cm ^-1 by a band or phenylalanine -CH ₂ bending of 1450 cm ^-1 is slightly from the Raman probe Shows increased signal collection. Despite efficient filtering in the Raman probe, there is still some evidence of the probe background observed in this data. Unlike BaSO _4, Raman signal from the tissue is for very weak. However, FIG. 46B shows the result after subtracting the fiber background and tissue fluorescence. All spectral features of the fiber background have been accurately removed, and these spectra of normal aorta were used to fix the 750 cm ⁻¹ peak due to Raman scattering from the sapphire ball lens and the lens. It looks almost identical except for a small peak just below 1600 cm ⁻¹ from the epoxy. Despite the slightly increased background in the raw data from the Raman probe, on average, the two processed spectra have the same SNR due to the increased collection efficiency of the probe. The double ring version of the Raman probe shows greatly improved performance over the experimental system.

  Here, the normal aorta, which is weakly Raman scattered, is evaluated because it is homogeneous. Experiments with more highly scattered tissues suggest that the probe performs even better in experimental or laboratory systems than for normal tissues. However, these differences are difficult to compare directly due to tissue heterogeneity.

  A preferred embodiment of the Raman probe provides the ability to collect an interpretable spectrum of tissue on a clinically relevant time scale. 2A-2C and 47A-47B show the processed spectra collected with the Raman probe in just 1 second using 100 mW excitation. Data are presented as dots, model fits are shown as overlapping lines, and residuals are plotted down on the same scale. FIG. 47A shows the spectrum of normal breast tissue, while 47B shows that of malignant breast cancer. The model fit is excellent even for spectra with reduced signal collection (ie normal aorta and malignant breast cancer), and all features remaining in the residual are noise.

  For fiber optic probes, maximizing the signal involves optimizing collection efficiency within the constraints of the spectrometer / CCD detector system of a preferred embodiment of the present invention. This is done by maximizing the area of the probe available for Raman light collection and characterizing the Raman light source for proper design of the collection optics. Noise minimization also requires some consideration in accordance with a preferred aspect of the present invention. First, the inherent noise source from the CCD detector is reduced by using a back-illuminated detector cooled to a very low temperature. Second, the excitation wavelength was carefully chosen to minimize shot noise from tissue fluorescence. For biological samples, excitation at 830 nm has been found to be ideal. Finally, in the case of fiber optic probes, fiber background is eliminated as much as possible to reduce its associated shot noise.

  The dual-purpose filter module used in the preferred embodiment of the Raman probe of the present invention effectively reduces the fiber background to a level where there is minimal spectral distortion once the remaining background is removed. High quality spectra of several tissue types can be collected in as little as 1 second using an excitation power well below the tissue damage threshold. The magnitude of the residual from the fits specifically illustrated in FIGS. 2A-2C and 47A-47B depends on the intensity of the Raman signal collected from the tissue, but in most cases they are purely noise And does not show the structure of the spectrum. The only exception to this is for non-calcified atherosclerotic plaques, in which case there is some small structure in the residual due to spectral resolution discrepancies between the data and the model.

  Results from phantom studies indicate that the probe decreased collection with increased absorption due to attenuation of both excitation and Raman light. Greater signal collection efficiency is observed with increased scattering. This increased signal from the highly scattered sample is the result of a Raman light source that is limited closer to the excitation beam that the ball lens collects most efficiently. Similar trends in the fiber background have also been observed, but the effect of scattering has been strengthened. Thus, if the concentration and cross section of the Raman scatterer is constant, only a slight SNR change is seen for tissues with various scattering and absorption characteristics. In addition, further analysis of the background signal, especially from sapphire ball lenses, may provide an internal calibration for the scattering and absorption properties of the sample, giving additional information about the diagnosis of the disease.

  The single ring probe of a preferred embodiment of the present invention can reduce the probe diameter. A smaller diameter means that the number of fibers in the probe must be reduced, thereby reducing the collection area. However, as the number of fibers is reduced, they are also brought closer to the excitation beam where the strongest scattering occurs. Also, the reduction in ball lens diameter results in a larger lens curvature, which leads to increased collection from a more central area of the Raman light source. Thus, there is an additional efficiency curve for the trade-off between the Raman probe collection area and the Raman light source sampling volume. At worst, the collection efficiency may decrease linearly with the number of fibers. Reduction to a total of 9 collection fibers results in a 1.5 mm total outer diameter and 1.2% collection efficiency Raman probe that produces a reasonable SNR for clinical practice.

  A preferred embodiment of the Raman probe of the present invention has general applicability and performs well for both arteries and breasts. The modular nature of the design provides great flexibility with respect to the particular choice of optics for high-throughput collection so that a variety of optical elements can be used to collect the desired spatial and angular distribution from the target tissue. Enable.

  Lateral observation probes may be used in alternative applications of a preferred embodiment of the present invention. For example, the use of angled and mirrored half-ball lenses, prisms, or micro-optical parabolic mirrors allows for efficient radial collection. The tapered tip allows for incorporation into a needle probe for optical breast biopsy, and the slightly smaller diameter allows breast analysis by ductoscopy for detection of dysplasia. Due to the collection capability of the Raman probe of one preferred aspect of the present invention, other potential uses include skin analysis, transcutaneous blood analyte monitoring and gastrointestinal cancer assessment.

  48A and 48B illustrate a clinical probe having a total diameter of less than 3 mm according to one preferred embodiment of the present invention.

  A preferred embodiment of the present invention is used for collecting clinical data. Laser power calibration is set using Teflon and is approximately 100 mW and ranges between 82-132 mW. The operating room was turned off as in the case of the blood vessel endoscope. Data was collected during peripheral vascular surgery. A spectrum of atheroma was collected during endarterectomy of the carotid artery. During femoral artery bypass surgery, spectra of the anastomosis site and posterior artery wall were collected.

  FIGS. 49A and 49B illustrate clinical data of normal artery intimal fiber proliferation according to one preferred embodiment of the present invention, where FIG. 49A is the acquired Raman spectrum and FIG. 49B shows the corresponding histology. This will be specifically described. The spectra were collected for a total of 5 seconds with a probe held orthogonal to the artery wall. There were 20 accumulations of 0.25 seconds each. Both 1 second and 5 second data were analyzed. All data was accumulated for 1 second.

  FIGS. 50A-50C illustrate clinical data of atheroma plaques of a preferred embodiment of the present invention, where FIG. 50A illustrates the Raman spectrum and FIGS. 50B and 50C illustrate the corresponding histological artery. I will explain it. The same data analysis method was applied to all clinical data collected as for normal arteries. 51A and 51B illustrate clinical data acquired for a preferred embodiment of the calcified plaque of the present invention, where FIG. 51A illustrates the Raman spectrum and FIG. 51B illustrates the corresponding histology. explain. 52A-52C illustrate the clinical data acquired for a ruptured plaque of one preferred embodiment of the invention, where FIG. 52A illustrates the Raman spectrum and FIGS. 52B and 52C illustrate the corresponding histology. I will explain it. 53A-53C illustrate the clinical data acquired for calcified plaques with thrombus according to one preferred embodiment of the present invention, where FIG. 53A shows the Raman spectrum and FIGS. 53B and 53C show the corresponding tissues. Explain the study specifically.

  Table 3 specifically explains the model components that were the results of analysis of clinical data representing various arterial conditions.

  FIG. 54 illustrates a side-viewing Raman probe 1900 that includes a single central excitation fiber 1902 according to a preferred embodiment of the present invention. The fiber buffer layer is matched to the diameter of the excitation filter rod 1916 to facilitate proper fiber / filter registration and has an aluminum jacket 1910 to provide optical isolation from the collection fiber 1908. . The configuration of the side observation probe is similar to the forward observation probe described with respect to FIGS. Probe 1900 includes a dielectric filter to minimize and preferably eliminate the Raman background of the fiber in the delivery and collection fiber, and also includes a rod 1916 fitted to a tube 1912 carrying the collection dielectric coating. Include. The module is placed at the distal end of the probe between the fiber bundle and the lens system to collimate the light beam having the half ball lens 1918. The probe diameter is approximately 1.5 mm and has a sheath disposed around it.

  55A-55C illustrate the effect of blood on signal collection according to a preferred embodiment of the present invention.

  FIG. 55A illustrates raw data collected using a system having a Raman probe according to one preferred embodiment of the present invention.

  FIG. 55B specifically illustrates the spectrum once the fluorescence has been removed, while FIG. 55C specifically illustrates the spectrum once the data has been normalized and processed. Perfused calcified arteries with blood to determine the effect of blood on signal collection. The blood spectrum is less than 10% of the arterial spectrum, and absorption and scattering due to the presence of blood reduces the arterial signal by approximately 1/3.

  FIG. 56 illustrates an illustration of a system that can be used in clinical practice of a preferred embodiment of the present invention. The embodiment shown uses a laser output 2001 that is focused by a lens 2004 in a Raman probe 2006 that is switched by a shutter 2002 and inserted into a biopsy channel of an endoscope 2008, which illuminates tissue over a region 2012. Deliver it to the tissue site 2010 as is possible. Acquisition optics 2231 provides a return of the Raman signal from the probe to the processor 2236, which is the spectrograph / CCD combination. The Raman signal and endoscopic camera are operated individually in system 2236.

  Endoscopic camera 2220 obtains its white light illumination from broadband xenon arc lamp 2224 with its own fiber optic illuminator 2222. A non-standard shutter 2228 under control 2232 of computer 2230 may be attached. Image signal 2234 may be processed by processor 2236 to produce a standard video signal 2238 (digitized by a frame grabber in computer 2230). A processed image signal 2240 with that information regarding the condition of the observed tissue is transmitted to the monitor 2242. The entire diagnostic procedure can be initiated by a foot switch 2244 connected to the computer by cable 2246.

  FIG. 57 illustrates the flow of method 2300 used in acquiring data for in vivo Raman spectrum diagnosis of one preferred aspect of the present invention. Upper loop 2301 is used to ensure proper excitation laser power in a sterile operating room. A calibration spectrum is acquired prior to clinical treatment to characterize system performance. A spectrum of a Teflon standard is obtained to determine the expected signal with the desired excitation power. During the procedure, the spectrum of the same sterile Teflon block is measured using a sterile Raman probe in this feedback algorithm. The automated correction for the laser power in step 2312 continues until the target Teflon intensity is obtained or until a predetermined threshold power is reached. Once the correct output is set, tissue spectra can be acquired. The laser is shut off by the shutter until data accumulation is started. The start of acquisition opens the shutter by step 2316, collects the spectrum by step 2318, and closes the shutter by step 2320. The collected data is then processed and displayed in real time along with spectral diagnostics. The system is then ready to examine the next tissue site. Details of data acquisition and processing are presented below.

FIG. 58 depicts the data flow used in the real-time analysis Raman system of a preferred embodiment of the present invention. Control of the laser, shutter, and CCD detector are all accomplished using software that can include, but is not limited to, LabView. Drivers for detector control are described in R ³ -software, Inc. Written in software as provided by. Any unfiltered probe background is characterized by collecting the excitation laser light reflected by the aluminum block. The spectral response of the system is characterized by collecting the spectrum of a calibrated white light source that is diffusely scattered by the reflectance standard (BaSO ₄ ). The spectrum of 4-acetamidophenol (Tylenol) is acquired for Raman shift calibration.

  Both the raw tissue spectrum and the aluminum spectrum are corrected for the spectral response of the system by dividing by the normalized white light spectrum. The remaining probe background is then removed from the tissue data by subtracting the aluminum spectrum. Tissue fluorescence includes a fifth-order polynomial fit, or some other means that can include, for example, but not limited to, Fourier filtering, point difference derivatives, spline fitting, Savitsky-Golay derivatives or weighted subtraction. To remove.

  Tissue characterization is performed by ordinary least squares fitting of data using established Raman spectral models. The resulting fit factor is used to provide a diagnosis based on an in vitro diagnostic algorithm generated by logistic regression analysis. The processed data, model fit and residuals (data fit) are then plotted in real time along with diagnostics and fit coefficients. Clinicians can use real-time data to make diagnosis and treatment decisions.

  In 1999, approximately 176,000 new cases of breast cancer were diagnosed in the United States alone, resulting in 44,000 deaths. There has been an increasing interest in the use of optical techniques to diagnose breast cancer in situ in the last 20 years.

  Current methodologies such as x-ray mammography and ultrasound look for density changes in the breast. These techniques cannot reliably distinguish between benign and malignant tumors and can therefore only be used to detect suspicious lesions and cannot be used for diagnosis. A tissue biopsy must be performed to determine if the lesion is malignant, and 70-90% of breast biopsies are found to be benign upon pathological analysis. However, instead of removing the tissue for pathological analysis, optical techniques such as Raman spectroscopy can be used to provide diagnostic information about the suspected lesion in situ. Raman spectroscopy as described above examines spectral sidebands generated by light scattered from a sample illuminated with monochromatic excitation light. Each chemical that is present has its own unique Raman spectral signature. By inserting a fiber optic needle device into the breast, it is possible to collect Raman spectroscopic measurements from the lesion and extract chemical information almost immediately. Obtaining such information using a Raman needle device results in a more objective and rapid (real-time) diagnosis and reduced trauma to the patient compared to currently used biopsy techniques.

  In addition to Raman spectroscopy, several other optical techniques are currently being explored. These include optical tomography, fiber optic ductoscopy and fluorescence spectroscopy. Optical tomography uses visible or near infrared light to illuminate a point on the surface of the breast, while the detector records light that is diffusely reflected or propagated at other points. In addition to providing information regarding the attenuation of the light signal as it traverses the breast, scatter and absorption information can also be extracted to quantitatively measure water, lipid and oxy / deoxyhemoglobin concentrations. The use of this information to distinguish between benign and malignant tumors is under study. In addition, a number of light sources and detectors can be used to form a measuring cup, allowing for three-dimensional imaging.

  Fiber optic ductoscopy is an adaptation of an endoscope developed to detect cancer in organs such as the colon, cervix, and esophagus to the examination of the ducts. Because most breast and precancerous cancers begin in the inner lining of the ducts and lobules, very small fiberscopes (less than 1 mm in diameter) are introduced into the ducts through the nipple (mainly papillary lesions). Search for. Illuminate and view the inside of the milk duct through an optical fiber. The apparatus can be used to observe the breast duct and its branches.

  Fluorescence spectroscopy has been successfully used to examine cancerous lesions in the esophagus, colon, bladder and oral cavity in vivo. Breast fluorescence spectroscopy has also been studied ex vivo and, while showing some promise for diagnosis, so far there is little understanding of the chemistry behind these results. Fluorescence based diagnosis is limited by the number of endogenous fluorescent chromophores (primarily collagen and NADH) present in breast tissue associated with cancer. In contrast, there are many more Raman-active molecules present in tissues associated with cancer development, such as collagen, fibrinogen, DNA, hydroxyapatite calcium and a variety of glycosaminoglycans.

  Raman spectroscopy has been used for chemical analysis for many years, but only recently has researchers begun to apply it to biomedical problems. The ability to acquire Raman spectra in clinical situations has been made possible by the development of new technologies such as small diode lasers, CCD detectors and holographic notch filters. Each of these components contributes to the secondary processing of small, high-efficiency systems for medical diagnostics previously unachievable.

Normal, benign using an excitation of 784nm to gather the Raman spectrum from (fibrocystic disease) and malignant (infiltrating ductal carcinoma) breast tissue, the 1450 cm ^-1 band 1439cm ^-1 in normal tissues shift was observed in malignant tissues (due to changes in the chemical environment of the CH ₂ bending mode). By using an area ratio of 1654 (by a combination of C = C extension and amide I band) and a band of 1439 cm ⁻¹ it is possible to distinguish between malignant and normal tissue. This difference can be attributed to increased protein concentration in the malignant sample. However, this test cannot be used to distinguish benign lesions from malignant lesions.

Prior art tests used excised biopsy samples fixed in formalin. The fixation process mainly crosslinks the collagen proteins to chemically alter the tissue and thus affects the Raman spectrum signature of the tissue. Using Raman spectroscopy, tissue can be diagnosed in vivo using frozen and unfixed tissue. Raman spectra of normal, benign and malignant breast tissue samples (approximately 0.5 cm ³ ) using 830 nm excitation have been previously reported. Principal component analysis of this data allowed discrimination between normal, benign and malignant tissues based on important spectroscopic features. However, principal component analysis did not allow identification of the chemical or morphological origin of these spectroscopic signatures, and the data sets at that time were too few for cross-validation (from 13 patients). 61 samples).

Clinical measurement of breast tissue using a fiber optic Raman needle probe typically samples a tissue region with a volume of 1 mm ³ . Cancer-related changes in the breast are accompanied by slight changes in the biochemical and morphological composition of the tissue. These changes occur at the microscopic level. As a result, it is important to characterize the characteristics of the Raman spectra of individual morphological components in order to develop diagnostic algorithms that provide insight into the microscopic state of the tissue. In that case, a model that uses these microspectral features as components to describe the macroscopic spectrum may be used to extract information about the composition of the tissue at the microscopic level. By identifying specific contributors to the Raman spectrum, a robust diagnostic algorithm can be developed.

  In previous embodiments, Raman spectroscopy was used for quantitative biochemical analysis of atherosclerotic lesions in aorta and coronary artery tissue in vitro. In these studies, tissue Raman spectra were modeled using a linear combination of Raman basis spectra collected from major biochemicals present in arterial tissue. A related approach was instead to create models based on the Raman spectra of individual morphological features found universally in the aorta, and to use these as the basis spectra for modeling . Similar morphological models for breast cancer diagnosis are used in preferred embodiments of the invention.

  Because it can be very difficult to determine which chemicals should be used to represent certain morphological features, the fundamental spectrum derived from morphological features is mainly used. For example, it may not be possible to identify all chemicals in complex mixtures such as those found in cells or tissues. More importantly, identifiable components such as collagen can be present in human tissue in many different forms, each having a slightly different Raman spectrum. The collagen found in breast tissue is actually a combination of several different types of collagen, but if each type of collagen is individually included in the model, this is over-fitting. Can lead to By using a single, morphologically derived collagen spectrum, an image of the chemical component in its microenvironment in normal or diseased tissue is obtained. Finally, chemicals that are purified in the laboratory or that come from commercial sources are not in their natural state. For example, proteins such as collagen may have been exposed to corrosive acids or other organic solvents. All of these problems are avoided by using the Raman spectrum obtained from the breast tissue itself. However, if necessary, synthetic or commercially available chemicals can be used.

  A morphological model of human breast tissue is developed using the Raman confocal microimaging system of a preferred embodiment of the present invention. This model can characterize all of the spectroscopic features observed in macroscopic samples of both normal and diseased breast tissue. It identifies the morphological components present in breast tissue by their unique Raman spectra and uses them as components to describe the morphological features of the macroscopic sample.

Breast tissue samples were obtained from surgical biopsy specimens. Samples were snap frozen in liquid nitrogen and stored at -85 ° C until spectroscopic examination. Samples were then mounted on a cryostat chuck using Histoprep (Fisher Diagnostics, Orangeburg, NY, USA) and 6-8 μm using a microtome (International Equipment, Needham Heights, Mass., USA). Sliced into thick sections. These sections were then mounted on MgF ₂ flats (Moose Hill Enterprises, Sperryville, VA, USA) selected by their small Raman background signal, and with phosphate buffered saline (pH 7.4) Kept moist.

  Raman spectral images generated using confocal Raman microscopy were collected from unstained tissue sections and correlated with phase contrast images of identical sections and serial hematoxylin and eosin stained sections. Images were overlaid for comparison. Where possible, examples of each morphological element were identified from various patients and disease states. The spectra were then classified according to their morphological origin (ie, collagen fibers or epithelial cells) and the disease classification of the tissue sample. For example, initially, the extracellular matrix spectra from normal and malignant samples were kept separate. Once a library of spectra for each morphological element was acquired (usually 60-80 spectra from 5-6 patients), they were analyzed for their degree of variation. If the spectrum of morphological elements was large or did not vary consistently, the spectra were averaged to create a morphologically derived base spectrum used in the model. When consistent differences were observed, as was true for cellular components, the number of independently varying contributors was identified and used to extract independent baseline spectra. If a single spectrum had additional Raman bands when compared to other spectra in its morphological category, those spectra were removed from that category and analyzed independently to determine the additional spectrum. It was ensured that the spectral features could be explained by other elements in the model. If the spectral features could not be explained by other elements of the model, a new base spectrum was added to the model and a database of Raman microimages was searched for similar spectral signatures. All micro-image phase contrast images and serially stained sections containing this new spectrum were re-evaluated. This methodology allowed new morphological features to be identified.

  Raman spectra were also collected from macroscopic breast tissue samples and a variety of chemicals, either synthesized in the laboratory or obtained from commercial sources. The Raman spectrum is the following commercially available chemicals for use in model development and image analysis (Sigma, St. Louis, Mo.): actin (chicken gizzard), β-carotene, hydroxyapatite calcium, cholesterol, Obtained from cholesterol linoleate, collagen (Usiachiles tendon, type I), deoxyribonucleic acid (calf thymus), ribonucleic acid (calf liver), phosphatidylcholine and triolein, and also calcium oxalate synthesized in the laboratory.

An illustration of the experimental equipment is shown in FIG. The same system was used for both macroscopic tissue samples and microimaging. Raman excitation light (830 nm) provided by an argon ion laser pumped Ti: sapphire laser (Coherent Innova 90 / Spectra-Physics 3900S, Coherent, Santa Clara, Calif., USA) is a bandpass filter (Kaiser Optical Systems, Ann Arbor, Minn., USA) ) And an aluminum holder for macroscopic tissue samples via a prism for Raman micro-imaging using two mirrors or a top illumination microscope (Zeiss Axioskop 50, Zeiss, Thorn, NY, USA) (Wood: axial resolution approximately 1 μm). The microscope objective lens concentrated both excitation and collected Raman scattered light in the form of backscatter. A combination of dichroic beam splitter and mirror redirects Raman scattered light from the microscope through a variable diameter confocal pinhole to increase axial resolution. When a macroscopic assembly was used, the camera lens collected Raman scattered light. The diameter of the light spot on the macroscopic tissue sample was approximately 1 mm, and the sampled tissue volume was typically 1 mm 3. For both configurations, the light was passed through a holographic notch filter (Kaiser Optical Systems) and then connected to a liquid nitrogen cooled CCD detector (Princeton Instruments, Princeton, NJ) at 0.25m f / 4 imaging. Focused on a spectrometer (Model 250IS / SM spectrometer monochromator, Chromex, Albuquerque, New Mexico, USA). With the smallest confocal aperture diameter (approximately 100 μm), the spatial resolution of the microscope system was approximately 2 μm ³ .

The spectrometer itself had an adjustable slit and a turret holding three diffraction gratings (Chromex) for the measurement range. For Raman examination, a 600 groove grating, blazed at 1 μm per mm, was used with a spectrometer inlet slit setting of 140 μm to provide a resolution of approximately 8 cm ⁻¹ . Since most biological samples did not exhibit a Raman bandwidth narrower than 10 cm ⁻¹ , the 140 μm width of the spectrometer entrance slit was generally used to provide maximized optical throughput (sometimes 70 μm entrance slit was used for macroscopic measurements).

A CCD camera on the microscope (Sony, Cambridge, Mass., USA) anticipated registration of a focused laser spot with a white light transillumination image and recording of the image on videotape. The microscope itself was equipped with a range of objective lenses with both normal and phase differences. Typically, a 63x infinite correction immersion objective lens (Zeiss Acroplan, numerical aperture 0.9) was used for Raman inspection. Both the detector and microscope translation stages were controlled by a computer. A complete Raman spectrum is collected at each tissue location and then a microscopic image of the tissue spectrum is moved by moving the translational state (Primer Scientific Instruments, Cambridge, Mass., USA) with a raster scan pattern under the microscope objective. Created. This method typically produced a 50 × 50 μm ² image, each pixel containing the entire Raman spectrum from 400 to 1850 cm ⁻¹ . The step size in both x and y directions is typically 2 μm, consistent with the spatial resolution of the confocal microscope. The spectrum was typically collected for 20 seconds at each pixel location with an output between 50 and 100 mW, thus requiring approximately 3.5 hours for the entire image. A drop of phosphate buffered saline kept the tissue sections moist during data collection. A test was also performed to test for photochemical damage to the tissue. At 220 mW, the intensity of the fluorescent signal was observed to decrease with increasing exposure time (over several minutes), while the Raman signal remained unaffected. Photobleaching occurred at this output, but its influence on the Raman signal was negligible. This photobleaching effect was not observed at lower excitation power. As a result, the power was kept below 100 mW during 20-30 seconds exposure to reduce the effects of photobleaching during the collection of spectroscopic data.

The acquired Raman spectrum was subjected to processing to ensure the reproducibility of the daily data. First, they corrected for the spectral response of the system using a tungsten light source. The data was then frequency calibrated using the known Raman line of toluene. The MgF ₂ background spectrum was then subtracted, and a broad and slowly varying fluorescence background was removed by fitting the spectrum to a fifth-order polynomial (in wavenumber) and then subtracting this polynomial from the spectrum. In addition, if necessary, a differential filter was used to remove contributions from cosmic rays.

  Microimaging or macroscopic histological spectroscopic data was simultaneously applied to the model base spectrum using MATLAB's positive least squares fitting algorithm or sequence of instructions (MathWorks, Natick, Mass., USA). Principal component analysis was used for the estimate of the number of independent variable components (MATLAB). In order to use either the least squares fitting technique or principal component analysis, each Raman spectrum was represented as a vector of intensity values corresponding to each wavelength.

  Another important issue when using a linear model is the orthogonality of the fundamental spectrum. The degree of orthogonality of elements is the equation

[Wherein x and y represent fundamental spectra of two morphological components arranged as Raman intensities at each wavelength (x ^T is a transpose matrix of x)]
Was used to test. A value of zero indicates that the vectors are orthogonal, and a value of 1 means they are identical.

The level of error in the morphological model is determined by the S / N ratio of the spectrum being used. If the model base spectra are not identical within the noise limits (ie they are more orthogonal than the same spectrum "changed" by noise), they are separated using the normal least squares method Can do. Since the fundamental spectrum is the average of many data points collected as long as necessary, they are virtually noise free. Thus, the limiting source of error in the model is due to noise in the fitted data. The error in the contribution contribution of a particular fundamental spectrum is proportional to the noise in the fitted spectrum:
E = NB (11)
Where B = P ^T (PP ^T ) ⁻¹ is the calibration vector of the morphological basis spectrum P and N is the noise in the sample.

  Raman spectroscopy was used to extract information about the morphological and chemical components present in relatively large abundance in the breast tissue reviewed here. The breast contains two types of tissue: glands and stroma. The gland element consists of leaflets and ducts. The leaflets are a dense population of epithelial cells that produce milk and secrete it into the ductal system, which transports milk to the teat. The tube consists of an inner layer of epithelial cells surrounded by a layer of myoepithelial cells. Both layers are surrounded by a basement membrane made primarily from collagen. Stromal elements provide a supportive network for these gland units and include the extracellular matrix, fibroblasts, fat and blood vessels. While the glandular element of the breast is almost cellular, there are only a few cells in the stroma. The majority of these cells are fibroblasts that are responsible for the production of the extracellular matrix (supporting network of structural proteins and carbohydrates, mainly collagen and glycosaminoglycans). Fat is the only other major morphological structure present and constitutes the bulk of normal breast tissue. Sometimes there is also a crystalline deposit of β-carotene (a lipophilic precursor to vitamin A).

  Many of the morphological structures in benign and malignant breast lesions are similar to those in normal breast tissue. For example, fibrosis occurs in both benign and malignant breast lesions and is accompanied by stromal proliferation. Fibrous tissue, like most of the extracellular matrix, is primarily collagen in composition, with an increased presence of proteins such as fibrinogen and fibronectin.

  However, some of the morphological characteristics of diseased breasts differ from those in normal breast tissue. For example, breast cancer is most commonly used for nuclear enlargement, polymorphism (variation in size and shape) and hyperstaining (stains more intensely), atypical mitosis, and DNA aneuploidy (chromosome). Emanating in the leaflets and ducts as rapid proliferation of epithelial cells with increased or lost). These morphological changes are not associated with large-scale production of new chemicals, but rather are associated with changes in the relative concentrations of chemicals already present in the breast. For example, the above morphological changes are accompanied by changes in the nuclear to cytoplasm ratio (a malignant qualitative indicator used by pathologists).

  Two additional morphological features that can be observed in breast cancer are calcification and necrosis. Calcification is important because they are high in radiation density, can be detected by mammography, and are often found with cancer. There are two main types with similar morphological features on mammograms. Type I calcification is calcium oxalate dihydrate crystals, while type II calcification is primarily hydroxyapatite calcium, but also contains other calcium phosphates and possibly also calcium carbonate. Calcium oxalate crystals are found benign and more often than in malignant lesions, and are thought to be products of cellular secretion, while hydroxyapatite calcium deposits are found in both benign and malignant lesions , And is believed to be the result of cell degradation or necrosis (death).

  With this basic knowledge of breast chemistry and structure and changes induced by disease progression, it is possible to explain all of the major Raman spectral features of normal and diseased breast tissue.

  More than 60 Raman images were collected from normal, benign and malignant breast tissue samples. A Raman image of normal breast ducts is shown in FIGS. Microimages of collagen, cytoplasm, and cell nuclei are generated by normal least squares fitting of each data point in the image with these base spectra. Serially stained sections are shown for comparison. It is clear that the structure observed in the Raman image correlates well with the tissue structure.

  From the microimaging data, nine important fundamental spectra were identified: cytoplasm, cell nucleus, collagen, fat, cholesterol-like, β-carotene, hydroxyapatite calcium, calcium oxalate dihydrate and water. Some features, such as cell membranes, have been identified using Raman imaging, but because they are not present in large quantities and have a small Raman cross-section, and therefore do not contribute significantly to the macroscopic tissue spectrum, Not included in the model. Others have already been found in the model to have virtually the same chemical composition as the element, and were therefore true in the case of the basement membrane (consisting mostly of collagen like the extracellular matrix) Could not be included as separate morphological features. The number of spectra used to determine the model component spectra also depended on the abundance of its morphological elements and also the issue of S / N ratio. For example, fat has a very strong Raman cross section. As a result, little fat spectrum was required from each patient to produce a clean spectrum. Both the extracellular matrix and cellular components discussed required more spectra to increase the S / N ratio. The fundamental spectrum used for the complete model of breast tissue is shown in FIG.

  FIG. 61 compares the spectra of fibroblasts and epithelial cells taken from normal, fibrocystic and malignant ducts. Statistical analysis showed that there are two main independently varying components emanating from the cytoplasm and cell nucleus. Although the spectrum of DNA [FIG. 62A] was very similar to that of the cell nucleus [FIG. 62B], the spectrum of the cell nucleus also contained minor features of RNA and histones. Similarly, the actin spectrum [FIG. 62C] was the main contributor to the cytoplasmic spectrum [FIG. 62D]. The cytoplasmic spectrum also included minor features related to other elements found in the cytoplasm.

  Since the ability to collect pure spectra from the cytoplasm and cell nucleus was limited by the collection volume of the Raman confocal microscope, the two fundamental spectra were mathematically separated. To separate the two components, a Raman imaging system was used to collect spectra of hundreds of cells (all types) from 8 patients. First, the spectra were fitted to two basic spectra, one taken from a cell region with a low nuclear content (determined by looking at the Raman signal) and one from purified DNA. Subsequently, the spectrum with a particularly high DNA fit factor (rich in DNA) corresponding to the spectrum taken from the nuclear region was separated from the spectrum with little or no DNA (poor DNA) collected from the region in the cytoplasm. The average of the DNA rich spectrum was then scaled and subtracted from the average of the DNA poor spectrum to give a new cytoplasmic spectrum without DNA content. This new cytoplasm-only spectrum was then subtracted from the average of the DNA-rich spectrum to remove all cytoplasmic features, leaving a spectrum representing only nuclear material. The original data (both DNA-rich and DNA-poor) were then fitted to these two modified baseline spectra. The procedure was repeated using two modified base spectra rather than purified DNA and a low nuclear content spectrum to give a final cytoplasmic and nuclear spectrum. This iterative process was used to minimize artifacts due to the inability of the purified DNA spectrum to model the nucleus (containing DNA, RNA, histones and others). These two fundamental spectra can be used to extract important diagnostic information about the cell, such as the nucleus to cytoplasm ratio.

Both the extracellular matrix and the basement membrane are mainly composed of collagen. Other structural proteins such as fibrinogen and fibronectin and proteoglycans also exist, but in such a small amount and small Raman cross section that they did not contribute significantly to the overall Raman spectrum. FIG. 63 compares the spectrum of morphologically derived collagen (mostly type I, but some III, IV and V were also present) and that of purified collagen (type I). Although they are very similar, some small differences can be observed in the region between 800 and 1200 cm ⁻¹ . The morphologically derived collagen spectrum was an average of 215 spectra taken from 7 patients, mainly from the area of the extracellular matrix.

  Fat is one of the strongest contributing factors to the Raman spectrum of normal breast tissue. It exists in large quantities and has a strong Raman cross section. Its storage in humans mainly takes the form of triglycerides, especially triolein. FIG. 64 compares the Raman spectrum of fat obtained from breast tissue with that of triolein and showed that triolein was the main contributor to the spectrum, as expected. The fat spectrum included in the model and shown in FIG. 64 was an average of 28 spectra collected using data from 5 patients.

  Necrosis within the lumen of a malignant breast duct or the center of a malignant tumor is specifically a product of cell degradation. As a result, its composition varied greatly from position to position even within a single duct. Analysis of Raman spectra from three patients showed that necrotic material contained fat, collagen, calcification (hydroxyapatite calcium), free cholesterol and cholesterol ester (linoleate) in addition to cellular material (both cytoplasm and nucleus) It showed that. Because the ratio of these factors can vary widely, the spectrum encompassed by the model (“cholesterol-like”) is a common component of these spectra (primarily cholesterol) that is not represented elsewhere in the model collected from a single patient. Component). Chemical modeling tells us that the “cholesterol-like” spectrum has a large contribution from cholesterol and cholesterol linoleate with a small contribution from cellular material (cytoplasm and nucleus) and fat. FIG. 65 shows a normal least squares fit to the data using these five elements. Pure chemical spectra of cholesterol and cholesterol linoleate were not used because there are one or more types of cholesterol esters that cannot be determined individually. As was true for collagen, it is best to model the tissue using a biologically derived mixture rather than using one or two pure components. Necrotic material was not the only element in breast tissue containing cholesterol and cholesterol esters. Although the cell membrane also contains both of these chemicals, they also include other chemicals such as phospholipids. It was therefore found that spectra based on “cholesterol-like” are present in small amounts in the whole tissue spectrum (not just malignant samples).

  Both hydroxyapatite calcium and calcium oxalate dihydrate have very strong Raman spectra [Figures 66A and 66B]. However, they were not universally found in frozen breast tissue samples. This is because calcification is important for medical diagnosis and therefore tissues containing calcification are generally not released for scientific research. Although calcifications were found in occasionally frozen samples, they were often punctate calcifications and difficult to study. For these reasons, spectra were obtained from 6 μm thick deparaffinized sections of breast tissue fixed in formalin where calcification was greater and greater. The fixation process changed tissue proteins, but did not affect the relatively inert mineral deposits of calcification. Thus, it was possible to use deparaffinized sections to analyze more calcifications identified for us by experienced pathologists from a range of patients and disease states.

  Hydroxyapatite calcium was identified from frozen sections from 3 patients and from deparaffinized tissue sections from an additional 11 patients. The spectra from the frozen and deparaffinized samples were identical. Spectra based on the hydroxyapatite calcium used were obtained from the combination of these spectra. Calcium oxalate was only observed in one deparaffinized tissue section due to its rarity. Although its presence in breast tissue is well documented, calcium oxalate dihydrate is significantly less common than hydroxyapatite calcium in breast tissue. Accordingly, calcium oxalate dihydrate was synthesized in the laboratory for incorporation into a model of a preferred embodiment of the present invention. Both calcification spectra were in agreement with previously published spectra.

  β-carotene is resonance enhanced when excited with 830 nm radiation. As a result it has a very strong Raman signal. Although the peak is conspicuous, it is often found with fat throughout the breast. To eliminate the need to extract fat content from the morphologically derived β-carotene spectrum, the spectrum obtained from commercially available β-carotene [FIG. 66C] was used. Using the morphologically derived Raman spectrum of β-carotene, it was confirmed that the commercially available sample was an accurate representation of β-carotene found in tissues.

Although water is a weak Raman scatterer, it contributes to the spectrum by a huge amount. Water constitutes approximately 80% by weight of human tissue and is present in phosphate buffered saline used to keep the tissue moist. Previously, arterial studies determined that water did not contribute significantly to the Raman spectrum of human tissue. However, in these studies of breast tissue, inclusion was found to be essential to fit the data properly. Water has a single relatively broad Raman band concentrated at 1650 cm ⁻¹ . If water was not included in the model, the fitting of this band by other morphological components was incomplete.

  One of the key requirements for successful morphological modeling was that there was little patient-to-patient variation in the Raman spectrum of a given morphological structure. Developing a model by averaging several spectra from many patients can ensure that the model includes a common element of all morphological features. The spectrum of the extracellular matrix was similar. The spectrum of the extracellular matrix is mainly collagen regardless of the patient. FIG. 67 shows the extracellular matrix spectra from five patients. The variability between patients is similar for all morphological features. The small differences observed were due to the close proximity of other morphological features. That is, fat droplets close to the collagen fibers being examined may have resulted in a small amount of fat being observed in addition to collagen. In developing a Raman model of breast tissue, it was found that linear variability not accounted for by other underlying spectra in the model was not significant.

  When analyzing the orthogonality of model components, it was found that the four components, namely cytoplasm, fat, collagen and cholesterol-like, had values greater than 0.5 when compared to each other (Table 4).

These four elements have many of the same functional groups (CH ₂ bending, C—C extension, etc.), yet they are sufficiently orthogonal to be particularly distinguished when using normal least squares fitting. It was. Water and cell nuclei also overlapped considerably. Nevertheless, the key to a successful fit has been to use as few elements as possible while retaining adequate spectral information to avoid spectral overdetermining. Despite having the highest degree of overlap (0.89), the difference between fat and cholesterol-like spectra is greater than the noise component of the data fit using the models of FIGS. 68A-68C and 69A-69C. Hence their incorporation into the model is reasonable. As previously discussed, in this situation, the predicted value of the model depends on the S / N ratio of the fitted data.

  Using the morphological model developed here, the spectral characteristics of a range of macroscopic tissue samples can be described with respect to the morphological composition of each sample. 68A-68C and 69A-69C, Raman spectra from normal, fibrosis, glandular disease, fibrosis / cystosis, fibroadenoma and invasive ductal carcinoma tissue samples are combined into a primary combination of morphological model base spectra. Fit. The fit factor generated by the model normalized to the sum to 1 (also shown in FIGS. 68A-68C and 69A-69C) is normalized to the whole tissue spectrum (excluding independently varying water) Represents the proportion of chemical and morphological base spectrum contributions made. For example, the fibroadenoma and malignant samples shown in FIGS. 69A-69C both have a large cytoplasmic content (31 and 34%, respectively) while the normal samples shown here do not. This observation reflects the greater cellularity of invasive cancer and fibroadenoma when compared to normal tissue or even other benign lesions, which is due to subsequent microscopic analysis of the sample by experienced pathologists. confirmed. The strong correlation between the model's fit factor and the morphological changes known to be insidious to the disease demonstrates the accuracy of the model. The small residuals observed in FIGS. 68A-68C and 69A-69C indicate that all major spectroscopic features are accounted for by the model. Similarly, small residuals were observed when 101 macroscopic tissue spectra collected from 37 patients representing a range of disease states were fitted with morphological models.

  By comparing the Raman image with a phase contrast image and also with a serially stained section of the same image, it is possible to monitor the chemical changes across the spectrum and thus the tissue surface. For example, not only can the spectrum of ductal epithelial cells found in malignant tissue be compared to that found in normal or benign tissue, but it is also affected by the disease process in the same tissue section as a region of invasive cancer Progressive changes in these spectra can also be monitored as they move between those that do not. Imaging also allows the identification of chemical / morphological differences that are not made visible by phase contrast or staining. With such information, a better understanding of the disease process and how it affects both tissue morphology and chemistry can be gained, and morphological models can be developed.

  The construction of a morphological model of breast tissue has three assumptions: first, the Raman spectrum of the mixture was equal to the weighted linear sum of the individual components of the mixture; Morphological features had identical Raman spectra among patients; and third, the fundamental spectra included in the models were sufficiently different in nature to allow their discrimination based on their Raman spectra I depended on that.

  Although only some of the collected Raman micro-images were used to create the presented model, all of them were used to test the comprehensiveness of the model. Using spectral data from a wide range of patients with diverse pathologies ensured that the model accounted for all the major spectral features found in breast tissue, including breast cancer. A good model fit also confirmed that the Raman spectrum of breast tissue was equal to the weighted linear sum of the spectra of the nine morphological / chemical elements included in the model. Each of the included elements had a strong spectroscopic signature, hardly varied between patients, and was present in large quantities except for calcium oxalate dihydrate. Some elements were not considered independently because their Raman spectra overlapped too much with those of other elements. This overlap is a problem for many cell types (epithelial cells, fibroblasts, etc.), basement membranes and cell membranes (which contributed weakly to the tissue spectrum but was very similar to the necrotic material spectrum). there were. Other chemicals present in breast tissue contributed little to the total Raman spectrum as they were not significant. For example, glycosaminoglycans are present in large amounts in the extracellular matrix, but have a very weak Raman cross-section, while matrix metalloproteinases are present in small amounts. None were observed in the Raman spectrum of breast tissue.

  The chemical composition of the morphological features identified by Raman microimaging was as expected. For example, the extracellular matrix was found to be mostly collagen, while the fat droplets were primarily triolein. The cell types examined (fibroblasts, epithelial cells from a range of normal and diseased states, and inflammatory cells) are all composed of the same basic components: cholesterol and cholesterol linoleate, actin and DNA. It was. Each cell is surrounded by a cell membrane (a lipid bilayer composed mainly of phospholipids, cholesterol, triglycerides and several proteins). The cytoplasm (mainly the cytosol or aqueous solution that fills the cells) constitutes the bulk of the cells. In the cytoplasm, a cytoskeleton composed primarily of actin filaments that allows controlled movement and organization within the cell; RNA involved in the cellular machinery (primarily regulating the production and production of more proteins) and There are proteins; as well as various organelles. The largest of these organelles is the cell nucleus. The nucleus is rich in DNA, RNA and histones, which are involved in helping DNA form small structures.

  Depending on the function of the cell, it has varying amounts of these components and possibly some additional ones respectively. For example, fibroblasts are responsible for creating and maintaining the extracellular matrix. In order to do so, they must produce collagen, fibrinogen and glycosaminoglycans in their cytoplasm and transport them to the extracellular space. However, for the development of a Raman model of breast tissue, these components are already included in the collagen spectrum and therefore do not need to be considered independently.

  The greatest difference between cells within or between types can be observed in the ratio of cytoplasm to cell nucleus. It is natural that there is some variation in this ratio, but it should be greatly exaggerated in malignant cells due to the occurrence of aneuploidy and used by pathologists to diagnose malignancy. Parameters such as the nuclear to cytoplasm ratio may be measurable in future macroscopic tissue samples.

  Numerous non-cellular components, namely collagen (extracellular matrix and basement membrane), fat, cholesterol-like (necrosis), hydroxyapatite calcium, calcium oxalate and β-carotene are also important in modeling the Raman spectrum of breast tissue It was found that Some of these, such as β-carotene, were important only because they were strong Raman scatterers and were therefore required for a good model fit. Like “cholesterol-like”, others are also an important feature used by pathologists to diagnose malignancy.

The proteins that contribute the most to the Raman spectrum of breast tissue are collagen and actin. Collagen represents the extracellular matrix, whereas actin is found in cells. Since both are proteins, their Raman spectra are very similar, especially in the region between 1440 and 1660 cm ⁻¹ , where researchers have previously looked for differences between normal, benign and malignant lesions. However, when the information contained in these fundamental spectra is used to fit the macroscopic tissue spectrum of the model, information about the relative amounts of cellular material (actin) and extracellular matrix (collagen) in a particular sample Can be extracted. This information is used to develop an algorithm based on Raman spectroscopy for diagnosing breast cancer as described below.

  Screening mammography is an important tool in the early detection of breast cancer. One feature of particular diagnostic significance is the presence of microcalcifications on mammograms. Two major types of microcalcification are found in breast tissue. Type I deposits consist of calcium oxalate dihydrate (birefringent colorless crystals), while Type II deposits consist of calcium phosphate, primarily hydroxyapatite calcium. Type II microcalcifications are typically basophile and non-birefringent in optical microscopy with H & E staining.

  There is no reliable way to distinguish between type I and type II microcalcifications in clinical mammograms, but the type is thought to correlate with disease. Calcium oxalate dihydrate crystals are most frequently found in benign ductal cysts and are rarely found in cancerous lesions, while calcium phosphate deposits are most often seen in proliferative lesions, including carcinomas. This distribution is consistent with the hypothesis that type I microcalcifications are products of secretion, while type II calcifications are derived from cell degradation or necrosis.

  Type II microcalcifications are estimated to occur two to three times more frequently than type I. Non-tactable type II microcalcifications found by mammography frequently geometrically target the location of the most important anomalies within the breast. Therefore, calcification is an important component that radiologists look for in mammograms. Several algorithms have been proposed that attempt to correlate parameters such as shape, size, number and roughness of microcalcifications detected by mammography with disease. However, these studies often rule out cases due to dense mammographic background, low density calcification spots or densely clustered calcifications and are therefore limited to certain patients and mammograms only. ing. The highest reported sensitivity and specificity for the mammography based cross validation algorithm is 71% and 74%, respectively. Although these studies show promising results, the diagnosis of breast cancer using microcalcification detected by mammography remains undefined. While the mammographic appearance of microcalcifications has some relationship to the histological type of the lesion, currently no diagnosis can be made reliably on this basis.

  Because calcium deposits in breast tissue are classified only morphologically, examining them using more rigorous methods may provide important insights. Raman spectroscopy is a technique based on energy exchange between light and matter. It is an inelastic scattering process in which photons incident on a sample transfer energy into or out of the sample's vibrational or rotational mode. It is a two-photon process and can be considered as simultaneous absorption of incident photons and emission of Raman photons. The difference between the energy of these two photons corresponds to the transition of the molecule from one energy level to another. Because the energy level is unique to every molecule, the Raman spectrum is chemical specific. Individual bands in the Raman spectrum are characteristic of specific molecular movements. Raman spectroscopy is particularly amenable to in vivo measurements because the power and excitation wavelengths used are non-destructive to the tissue. Since Raman spectroscopy can provide a definitive chemical analysis of these morphological structures in vitro, it is more well suited for the study of microcalcifications in breast tissue. In fact, Raman spectroscopy has been successfully used to examine calcific deposits in several other organs such as the kidney and urinary tract.

  A preferred embodiment of the present invention uses Raman spectroscopy to highlight differences in the chemical composition or structure of microcalcifications present in different lesions in the breast. The results from this aspect will advance the understanding of the chemical changes associated with the onset and progression of breast disease and provide important parameters for the diagnosis of breast disease using Raman spectroscopy.

Raman spectra were obtained from 6 μm thick deparaffinized sections of formalin-fixed paraffin-embedded breast tissue. Because of their diagnostic importance, microcalcifications in fresh breast tissue typically cannot be left for scientific studies, and therefore the preferred embodiment system is suitable for microcalcifications in fixed tissue sections. Limited to inspection. Since microcalcification is relatively inert, the protein cross-linking effect of the fixative should be minimal. Furthermore, the shape of the Raman spectral lines from the examined calcifications is consistent with previously published data obtained from unfixed tissue of other organ systems. Samples were mounted on MgF ₂ flats (Moose Hill Enterprises Inc., Sperryville, VA). Each microcalcification examined was photographed using a phase contrast microscope. Phase contrast images and H & E stained serial sections are reevaluated by an experienced pathologist (blinded for spectroscopic results) and histological diagnosis of the disease state in the area from which the data was obtained did. A total of 30 Type I and 60 Type II microcalcifications in breast biopsies from 11 patients were examined using Raman spectroscopy, with 74 histologically benign ducts And 16 were from histologically malignant ducts. Histological diagnosis of benign ducts ranged from ductal epithelial hyperplasia, sclerosing gland disease, fibrocystic disease and fibroadenoma to Menkeberg arteriosclerosis, while all 16 malignant ducts were glandular. Diagnosed as ductal carcinoma in situ (DCIS). Invasive cancer was not encountered in the area where data was acquired. All 11 patients were white women with an average age of 53.4 years (range 41-85 years) and had undergone an excised breast biopsy for suspected microcalcifications found on mammography. These patients had no palpable breast lesions and no mass lesions of other important findings on mammography, with the exception of fibroadenoma.

Data was acquired using the Raman microscopy system (described previously) shown in FIG. Briefly, Raman excitation light (830 nm) was launched into a confocal microscope and concentrated to a spot size of approximately 2 μm. Both objective lens, 63x (NA 0.9); Zeiss Acroplan) concentrate the excitation and collect the Raman scattered light in the form of backscatter. A charge coupled device camera on top of the microscope expects registration of the focused laser spot with a white light illuminated image. A combination of dichroic beam splitters and mirrors redirects the Raman scattered light to the spectrometer system, where it is recorded by a deep depleted CCD detector (Princeton Instruments, Princeton, NJ) cooled to -100 ° C. The dispersion of Raman scattered light on the CCD results in 1.6 cm ⁻¹ per pixel. All of the Raman spectra were acquired with an integration time of 60 seconds and a spatial resolution of 8 cm ⁻¹ . The average laser excitation power used varied between 100 and 150 mW.

  All of the data processing was performed using a software algorithm, such as in MATLAB 5.30. The spectrum was Raman shift frequency calibrated using known spectral lines of toluene. A fifth order polynomial was fit to the spectrum by least squares minimization and then subtracted to remove the slowly varying fluorescence background. Cosmic rays were removed by using a differential filter.

FIG. 71A is a radiograph of a sample representing features that imply the presence of microcalcification, while FIG. 71B displays a phase contrast image collected from a slice of this sample. FIG. 71C shows the Raman spectrum of type I microcalcification obtained from the deposit highlighted by the small box in FIG. 71B. Based on the overall histology of this sample and the specific features evident in the phase contrast images, the lesion was diagnosed as a fibrocystic disease. The vibration characteristics characteristic of calcium oxalate dihydrate can be seen at 912 cm ⁻¹ , 1477 cm ⁻¹ and 1632 cm ⁻¹ . These Raman features are attributed to CC stretching, and CO symmetrical and asymmetric stretching, respectively, and are consistent with the previously published Raman spectrum of calcium oxalate dihydrate.

72A and 72B display a phase contrast image of type II microcalcifications in a malignant duct and a radiograph of the corresponding sample. FIG. 72C shows a Raman spectrum obtained from the deposit highlighted in FIG. 72A by a small box. Examination of this spectrum reveals that microcalcification is not composed of pure hydroxyapatite calcium. The Raman spectrum of pure stoichiometric hydroxyapatite calcium contains four phosphate internal vibrational modes and bands due to hydroxyl ion stretching and vibrational modes. Two of the phosphate vibration modes are from both the spectral range selected for examination as well as the hydroxyl ion mode. The large band at 960 cm ⁻¹ is a fully symmetric elongation mode of v ₁ (PO ₄ ) of the “free” tetrahedral phosphate ion. Another phosphate V ₁ mode exists at 948 cm ⁻¹ , but is obscured by a broad phosphate extension mode of 960 cm ⁻¹ . Overlapping Raman structures from five v ₃ (PO ₄ ) bands can be seen between 1028 cm ⁻¹ and 1061 cm ⁻¹ . A sixth v ₃ (PO ₄ ) mode appears at 1085 cm ⁻¹ . The characteristics of phosphoric acid present are consistent with the previously published Raman spectrum of hydroxyapatite calcium. In addition to the phosphate peak derived from hydroxyapatite calcium, there are several other vibrational modes that exist in this spectrum. Protein contributions can be seen at 1445 cm ⁻¹ and 1650 cm ⁻¹ . A sharp peak at 1004 cm ⁻¹ can also be attributed to phenylalanine. Small contribution from the lipid is evident each 1130 cm ^-1 and 1300 cm ^-1 as C-C stretching and CH (CH ₂₎ bend.

Initially, data obtained from Type I and Type II microcalcifications were separated based on their Raman spectra. The presence or absence of vibration intensity at a specific wavenumber was used to distinguish between type I and type II microcalcifications. While the spectrum containing a large peak characteristic 912Cm ^-1 and 1477cm ^-1 of calcium oxalate dihydrate were grouped into the category of type I represents the intensity characteristic of 960 cm ^-1 to calcium hydroxyapatite The spectra were grouped into Type II categories. In the analysis performed in one preferred embodiment, separation into type I and type II microcalcifications was performed by visual inspection. However, may be performed in one embodiment of an alternative automated computer algorithm distinguishes to any one of them ^{960 cm} -1 based on the intensity values of those with and present normalized spectrum (II type) or 1477cm ^-1 (I type) . All 30 type I microcalcifications examined were formed in the lesions of fibrocystic disease, and therefore all 30 type I microcalcifications were diagnosed as benign. This is consistent with the fact that type I microcalcifications are formed as secretion products and are typically located in cystic lesions. Although type I microcalcifications are found in malignant lesions, especially lobular carcinoma in situ, it is very rare.

  A multivariate statistical analysis method called principal component analysis (PCA) to distinguish type II microcalcifications present in benign and malignant breast lesions. Similar methods have been used to classify tissue samples suffering from several other organ system diseases. PCA uses the entire Raman spectrum and does not assume any knowledge about the chemical composition of the tissue. It is a chemometric technique that resolves an entire set of spectral data into several orthogonal principal component (PC) spectra. These PC spectra can have positive and negative components, and a complete set of data that accurately describes all of the data (within the limits imposed by noise) when the PC is multiplied by the appropriate weighting factor Cassis can be formed. These weighting factors, called scores, are similar to chemical fractions. As a method based on factor analysis / chemometrics, PCA can recognize small spectral variations and thus distinguish samples based on similarity. This analysis method is well suited for testing type II breast microcalcifications. This is because they are mainly composed of hydroxyapatite calcium with minute chemical fluctuations due to biological impurities. PCA inherently and naturally provides little physical information, however, it is skilled in isolating spectral trends that correlate with physical information, and thereby provides the basis for the development of diagnostic algorithms. Furthermore, by comparing the shape of the lines in the diagnostic PC spectrum with the spectrum of pure chemicals, it is possible to attribute them to what they mean. More importantly, this preferred method of analysis provides evidence of the principle that there is really important diagnostic information contained within the Raman spectrum of type II microcalcifications.

A singular value decomposition algorithm for determining a set of PC data is used in a preferred manner. Normalizing the set of data to a peak height of 960 cm ⁻¹ to remove any possible intensity bias, and then removing features common to all of the spectra centered on the average before performing PCA; It emphasizes the spectral dispersion. All 60 of the spectra could be accurately modeled above the noise level using 9 PCs. The first 6PCs account for over 97% of the total variance in the data. A logistic regression analysis (one discriminant analysis) was then used to generate a diagnostic algorithm that was used to classify breast lesions into benign and malignant categories. Logistic regression analysis correlates the PC weighting factor (score) calculated for each Raman spectrum with the diagnostic category. Diagnosis was provided by a blinded pathologist. A robust algorithm was obtained using cross-validation analysis without a single piece.

  Fibroadenoma is a benign tumor of a completely different lineage from all the other lesions examined. It is closely related to phyllodes tumors (the malignant counterpart is malignant in the stroma rather than the epithelium). In addition, clinicians typically know that breast lesions are in the fibroadenoma / phyllodes family based on physical examination and are characterized by mammography other than microcalcifications. Because these lesions must be surgically removed for treatment, it is unlikely that physicians will use techniques such as Raman spectroscopy as an aid tool for the diagnosis of fibroadenoma. For these reasons, the performance of the algorithm is evaluated after excluding samples diagnosed as fibroadenoma from the analysis.

A combination of PCA and logistic regression analysis was used to examine the signature of the Raman spectrum of type II microcalcifications to determine whether a diagnosis of breast disease could be made based on this. Three PC scores were used to obtain a high level of diagnostic accuracy. Important score related to _PC 2, PC ₃ and PC _5. PC ₅ accounts for 1.0% of the total variance in the data, while PC ₂ and PC ₃ account for 8.8% and 5.2%, respectively. Using these three PCs and the single cross validation method, 14 out of 16 DCIS and 34 out of 39 benign samples could be accurately predicted. Thus, type II microcalcifications occurring in benign and malignant ducts could be distinguished with 88% sensitivity and 87% specificity. If all of the sample is retained, sensitivity and specificity are only slightly degraded, maintaining a sensitivity of 88% with a drop in sensitivity to 80%. An illustration of a diagnostic algorithm for type II microcalcifications is shown in FIG. To reduce the algorithm to a two-dimensional display, PC ₅ and PC ₂ (both with higher scores for benign microcalcification) were added together to form a single axis. Based on this algorithm, all of the samples diagnosed with ductal epithelial hyperplasia and sclerosing glandular disease (the most universally benign state morphologically confused with cancer) were accurately predicted. 4 of 5 type II interstitial calcifications present in fibroadenoma, and 2 of 4 arterial calcifications of Menkeberg arteriosclerosis and 13 ductal limes in fibrocystic disease Three of the cases were misdiagnosed.

  In general, only one microcalcification was examined for each lesion. However, two samples examined multiple microcalcifications from the same lesion and there was no significant difference in the spectrum of each given lesion. When the data were combined data from both Type I and Type II microcalcifications, an overall sensitivity of 88% and a specificity of 74% were obtained, as well as a negative predictive value of 97%. A patient operating characteristic (ROC) curve generated from these results is shown in FIG. Based on these in vitro results in fixed tissue, it is shown that Raman spectroscopy has the potential to more accurately distinguish microcalcifications associated with benign malignant breast lesions than mammography. Additional studies conducted in vitro and ultimately in vivo on fresh tissue may better assess the clinical utility of Raman spectroscopy compared to x-ray mammography for the diagnosis of breast cancer.

Examination of the three diagnostic PC spectra can provide insight into the physical basis governing this identification. The most diagnostically relevant PC spectrum was PC ₅ shown in FIG. 75A. Examination of this PC spectrum shows broadening of the 960 cm ⁻¹ phosphate extension peak. This broadening is clearly shown in FIG. 75B, where PC ₅ is overlaid with an average spectrum from all of the Type II microcalcifications. It has been reported in the literature that this peak broadening is due to the presence of calcium carbonate. In these embodiments, application of Raman spectroscopy to the carbonated apatite model system showed phosphoric acid peak broadening with increased carbonate content. The introduction of carbonate ions into the apatite structure reduces the symmetry of the unit chamber. Peak broadening results from the loss of long-range migration order in the apatite structure as the sample's carbonic acid content increases. The analysis found a linear relationship between the FWHM of the 960 cm ⁻¹ phosphate stretch mode and the calcium carbonate content of the sample. Evidence that 960 cm ⁻¹ broadening in PC ₅ may be derived from variations in the calcium carbonate content of microcalcifications is evident in the 1070 cm ⁻¹ peak attributed to the ν ₁ (CO ₃ ) mode of calcium carbonate. It is. However, the difficulty in interpreting the PC spectrum afforded by the inclusion of both positive and negative features requires additional consideration.

If PC ₅ is indeed responsible for the variation in the amount of calcium carbonate present, a spectrum with a higher score for PC ₅ will contain more calcium carbonate than a spectrum with a smaller weighting factor. Since the benign spectrum typically has a PC ₅ score greater than the malignant spectrum, it can be assumed that the type II microcalcifications present in benign lesions of the breast contain more calcium carbonate than the deposits found in DCIS. To confirm this hypothesis, the full width at half maximum (FWHM) was calculated for the 960 cm ⁻¹ phosphate extension peak in each Raman spectrum. In accordance with the theory that type II microcalcifications formed in benign lesions have a higher calcium carbonate content, type II microcalcifications present in benign breast lesions average 18.0 ± 0.5 cm ⁻¹ It was found to have a FWHM. The significance of this difference is reflected in a P of 0.03. This value was calculated based on Wilcoxon rank sum test (not assuming normal distribution of data). Furthermore, the opposite trend is found when examining the FWHM of misdiagnosed samples. The FWHM of a benign sample mistakenly diagnosed as malignant was 15.8 ± 0.5 cm ⁻¹ , while that of a malignant sample mistakenly diagnosed as benign is 17.5 ± 0.5 cm ⁻¹ , The breadth of elongation mode of phosphoric acid is an important diagnostic feature. However, although the peak height of the elongation mode of carbon dioxide at 1070 cm ⁻¹ is on average more than 4 times in the benign sample, it does not correlate linearly with the FWHM of the elongation mode of 960 cm ⁻¹ phosphate. This indicates that additional impurities in the apatite structure contribute to symmetry breaking and thereby broadening the 960 cm ⁻¹ peak. These impurities are evident in the complex vibrational structure of PC ₅ , but have not been identified to date. PC ₅ also contains a number of features representing proteins such as vibration of _CH 2, amide CH ₃ bending mode and 1650 cm ^-1 I of 1445cm ^-1. Unlike calcium carbonate, which has a positive strength, protein features have a negative direction. This indicates that the protein and carbonic acid contents are inversely correlated, and thus the benign sample has a lower protein content than the malignant sample.

The amount of protein and calcium carbonate present during type II microcalcifications in benign and malignant lesions is additionally confirmed by examination of PC ₂ shown in FIG. This spectrum also appears to contain positively directed calcium carbonate features (especially 1070 cm ⁻¹ ) as well as negatively directed protein features and, on average, fine lime formed in benign ducts It greatly contributes to the Raman spectrum. In addition, PC ₂ exhibits a large second derivative-like feature around 960 cm ⁻¹ . This type of structure explains the broadening of the peaks in the data, and in addition, type II microcalcifications formed in benign ducts are more abundant than those formed in malignant ducts This supports the hypothesis that it tends to have a greater broadening of calcium carbonate and hence the peak at 960 cm ⁻¹ .

PC ₃ was also found to be diagnostically important and is shown in FIG. However, PC ₃ contributes significantly to the Raman spectrum obtained from type II microcalcifications in malignant ducts. It has positively directed protein characteristics and thus provides additional support for the theory that microcalcifications formed in malignant ducts have more protein than deposits in benign ducts . The amount of protein during microcalcification in benign and malignant ducts is confirmed by monitoring the peak height of the amide I vibration at 1650 cm ⁻¹ . The intensity of this mode is approximately 1.5 times greater in type II microcalcifications formed in malignant lesions. In addition, the contribution from phenylalanine (an amino acid often found in association with collagen and other proteins) can be seen in PC ₃ at 1004 cm ⁻¹ . PC ₃ represents a large first derivative-like feature at approximately 960 cm ⁻¹ . This feature is responsible for the shift of the phosphate elongation mode peak and is positively correlated with the protein features. The presence of these protein features is a misdiagnosis of interstitial calcification in fibroadenoma and arterial calcification in Menkeberg arteriosclerosis as a result of stromal or arterial degradation similar to cell degradation occurring in DCIS Might explain.

  Preferred embodiments involving Raman probes have shown the diagnostic potential of Raman spectroscopy to distinguish microcalcifications found in benign and malignant lesions. In addition, using PCA, slight differences in the chemical composition of type II microcalcifications present in benign and malignant breast lesions have been discovered. Based on the results, it can be assumed that type II microcalcifications present in benign lesions of the breast have a lower protein and higher calcium carbonate chemical content than those formed in malignant lesions. A preferred embodiment uses a Raman technique in vitro in a biopsy of a breast where little tissue is obtained and the lesion may not be well represented but microcalcification is present. Furthermore, the embodiment may be used as an in vivo appendage to mammography to assist in the selection of patients with microcalcifications that subsequently need to undergo a biopsy.

Raman spectroscopy can provide detailed qualitative and quantitative information about the sample under consideration. Several approaches have been used to acquire a set of Raman imaging data. The three standard approaches are point scanning, line scanning and direct imaging. Direct imaging requires the collection of complete images with a single spectral component. Wavelength selectivity is achieved by using either an acousto-optic or liquid crystal tunable filter that sweeps with a specified wavenumber interval that captures the frame at each. Line scans and point scans collect complete spectra (typically encompassing Raman shifts between 400 and 1800 cm ⁻¹ for biological media) when imaging either lines or single points. The resulting data set from each of these approaches can be viewed as a hypercube of Raman intensity as a function of the Raman shift and the two spatial axes.

  In addition to mapping tissue structure, Raman imaging can be used for in situ chemical studies of disease processes. One such example is atherosclerosis, where the end product of the disease (the ceroid) has been identified as an autofluorescent lipid product of unknown chemical composition. Surface enhanced Raman spectroscopy with imaging can be used to examine the chemical composition of viable cells. In particular, the DNA and phenylalanine content of the cells can be monitored.

  One old technique for the creation of spectral images is by examination of specific peak heights. In this approach, an image is created by plotting the intensity of a particular Raman band at each spatial location. This model is widely used and provides information about the spatial location of all molecules in the sample that give strength to the chosen vibrational frequency. However, this approach uses only a small portion of the available data. In complex biological samples where several distinct parts appear to give intensity to a particular Raman band, it is necessary to capture all of the spectral information to distinguish them. This is accomplished by application of a model that utilizes the full spectrum, such as is done using point and line scans when creating an image. The trick is to compress the information into a form that can be managed and still provide information. Some universal data compression techniques are principal component analysis (PCA), multivariate curve resolution (MCR) and Euclidean distance. Morphological modeling is an approach that is also used in preferred embodiments of the present invention.

  Each of these methods relies on the basic assumption that the Raman spectrum of a mixture of chemicals can be expressed as a linear combination of the spectra of the components of the mixture. The Raman image is generated by fitting the fundamental spectrum contained in the model to the Raman spectrum obtained at each position in the image. In general, the more base spectrum contributes to the data spectrum, the higher the fit factor and the brighter the spot in the image of the component being examined. For PCA and MCR, the fundamental spectrum is derived mathematically, while for Euclidean distance and morphological modeling, the fundamental spectrum is determined experimentally.

  PCA uses singular value decomposition to calculate the fundamental spectrum. The first fundamental spectrum (ie principal component) describes the maximum variance in the data if the data is centered on the average before analysis. The second fundamental spectrum explains the next largest variance, and so on until the fundamental spectrum only explains the noise in the data. These fundamental spectra are created such that they are mutually orthogonal and therefore do not contain overlapping spectral information. A two-dimensional image can be created using the fit coefficients obtained when these principal components are applied to a set of imaging data. This image provides a map of how the spectral features represented by the principal components are distributed in the sample. In turn, this map may correlate with morphological features observed by another optical technique such as phase contrast microscopy or light microscopy using histological staining. The linearity of the principal component may also be correlated with the Raman spectrum of a known chemical, but this is difficult because the principal component contains both negative and positive spectral features.

  MCR is designed to extract a fundamental spectrum that is similar to the real Raman spectrum of the chemicals present in a sample. The initial estimate of the concentration or fundamental spectrum present in the sample is used in constrained alternating least squares optimization. Generate new estimates of the concentration and base spectrum by iterating between the least squares solution for the base spectrum and concentration. These equations can be solved according to non-negative constraints to ensure that both the fundamental spectrum and concentration are all positive and therefore physically related. Optimization continues until the change in concentration and baseline spectrum from one iteration to the next is minimized. The more complex the system, the better the initial estimate needs to get a meaningful solution to these equations. Due to the various components inherent in the data and the high degree of overlap in the spectral characteristics of the noise, the MCR cannot always converge to the correct solution. However, if a solution is found, the resulting fundamental spectrum is similar to the Raman spectrum of the individual chemicals present in the sample. Once again, an image may be produced using the base spectral fit factor.

  Both PCA and MCR are useful techniques when the sample is almost speculatively unknown. They make it possible to extract spectral information without knowing its chemical origin. Both Euclidean distance measurements and morphological modeling both create images using information about the sample's known chemistry. While Euclidean distance only requires knowledge of the few chemicals that are present, morphological modeling requires knowledge of all of the major contributors to the sample's Raman spectrum. Despite the need for most prior knowledge of the sample, morphological modeling produces the most easily interpretable results.

  The Euclidean distance classifies the spectral dispersion of the image data from the fundamental spectrum (usually a pure chemical spectrum) according to the geometric distance of the data. The distance is the equation:

Where d is the Euclidean distance, S is the sample data, P is the pure chemical spectrum, and λ represents the wavelength at which the spectrum is acquired. The greater the image spectrum differs from the fundamental spectrum, the greater the distance.

  Morphological modeling is a new technique for analyzing Raman images and uses a conventional least squares method to fit a set of fundamental spectra to the data. The origin of the fundamental spectrum makes this approach very useful. Base spectra are obtained from the main morphological features found in a set of representative samples using a Raman confocal microscope. By using a range of morphological features acquired in situ, a spectrum representing its morphological components in the in situ chemical microenvironment is obtained. The fundamental spectrum should account for all of the major chemicals present in the sample, but to determine which fundamental spectra can be accurately resolved, the S / N of the data as well as the overlap of the fundamental spectra You must consider both of the degree. Although the fundamental spectrum can be obtained from pure compounds, morphologically derived components are preferred. They are derived from actual samples and are therefore closer to those observed in situ than pure chemical spectra. Sometimes a combination of pure chemical components and morphologically derived components gives the best results if the chemical of interest is not independently present in the sample. If the model is well selected, the resulting image can show detailed morphological and chemical structure in the sample.

  The preferred embodiment applies morphological modeling to human colon cancer cells and Raman images of human breast and arterial samples. This morphological modeling method is compared with other commonly used techniques, mainly: peak height analysis, PCA, MCR and Euclidean distance.

Breast tissue samples were obtained from excised biopsy specimens, while arterial samples were obtained from explants at the time of transplantation. Once removed, the tissue was snap frozen in liquid nitrogen and stored at -80 ° C. Tissue samples were then mounted on a cryostat chuck using Histoprep (Fisher Diagnostics, Orangeburg, NY) and 6 μm thick using a cryomicrotome (International Equipment Company, Needham Heights, Mass.). Cut into sections. Cut several serial sections, one on a MgF ₂ slide (Moose Hill Enterprises Inc., Sperryville, VA) for Raman data acquisition, and at least two others on glass for histological staining. Mounted on a glass slide. The stained glass slide was used for pathological confirmation of the features observed on the Raman map. During the measurement, the tissue was kept moist with PBS, pH = 7.4. In addition to the Raman microimage, a phase contrast image of unstained tissue was recorded via a CCD camera.

Cell studies were performed using the human colon cancer cell line HT29. They use high glucose Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum, 100 units / ml penicillin and 100 μg / ml streptomycin (all Gibco BRL products, Life Technologies, Grand Island, NY). And allowed to grow. Cells were grown to confluency at 37 ° C. in a humidified atmosphere of 5% CO _{2 in} air and dispersed in suspension using trypsin. The cell suspension was placed on a MgF ₂ flat, rinsed with phosphate buffered saline (buffered with PBS, pH = 7.4) and allowed to air dry. Sample drying was necessary to fix the cells throughout the mapping experiment. The dried sample was then rewetted with PBS and a Raman map was then obtained. Raman imaging microscopy data collected from dried cells was compared to data collected from cells still viable in suspension using the bulk Raman system. Spectra obtained from dry cells were used to model spectra obtained from viable cells. No residual from the model fit was observed.

The Raman microimaging setting used to collect the image data presented here was a point scan system. Raman excitation was provided by an argon ion laser pumped Ti: sapphire laser (Coherent Innova 90 / Spectra-Physics 3900S, Coherent Inc., Santa Clara, Calif.). Typically, 50-150 mW of 830 nm excitation light was focused on a spot on the sample with a diameter of less than 2 μm by a microscope objective (63 × Zeiss Acroplan, infinite correction immersion, numerical aperture 0.9). The spectral resolution was approximately 8 cm ⁻¹ . A spectral map of the tissue was created by raster scanning a translation stage (Primor Scientific Instruments Ltd., Cambridge, Mass.) Under a microscope objective. The map was usually acquired using a 2 μm step size consistent with the spatial resolution of the confocal microscope. Although the data acquisition time was dependent on several user-defined parameters such as the image step size, number of steps and spectral acquisition time, the entire Raman image was typically generated in 2-5 hours. The CCD camera on the microscope allowed for registration of focused laser spots with white light illumination or phase contrast images.

All spectral data processing was performed using software such as MATLAB (MathWorks, Inc., Natick, Mass.). The data was corrected for the spectral response of the system using a tungsten light source and then frequency calibrated using the known Raman line of toluene. Cosmic rays were removed using a differential filter and a small background from the MgF ₂ flat was subtracted. The data was then fitted with a fourth or fifth order polynomial and subtracted from the spectrum to remove any fluorescence background. All data were normalized to a peak height of 1. Finally, MATLAB was used to perform various data compression techniques: PCA, MCR, Euclidean distance and morphological modeling. In a preferred embodiment, the PCA used as a fitting algorithm for morphological modeling and the normal least squares algorithm were provided from software such as but not limited to MATLAB, while the MCR algorithm was PLS_Toolbox. (Eigenvector Research, Inc, Manson, WA). The pure chemicals used for spectroscopic modeling of HT29 cells, namely triolein, phosphatidylcholine, cholesterol and DNA (calf thymus) were purchased from Sigma (St. Louis, MO).

  In order to obtain improved image contrast, a smoothing algorithm based on spatial filtering was applied to the preferred mode of data. Spatial filtering relies on the assumption that adjacent pixels in a digital image contain relevant information. A group of pixels that surround and encompass the central pixel is called a kernel. The smoothing algorithm is based on a 3 × 3 kernel size. The preferred algorithm uses a mask that weights the contributing pixels by the reciprocal of their geometric distance from the center of the kernel. The resulting mask is:

Where each fraction represents the weight of the pixel in the kernel.

  Morphological modeling is a powerful tool for collecting structural and chemical information on small scales. 78A-78G, human colon cancer cells (HT29) spectra were developed for phosphatidylcholine (A), DNA (B), cholesterol (C), triolein (D), and “cytoplasm” (E) (breast tissue model). Features such as cell membrane, nucleus and cytoplasm are easily identified when applied to the pure chemical spectrum of the morphologically derived spectrum, primarily actin). The spectrum corresponding to the voxel shown in FIG. 78E can be viewed in G along with the corresponding fit and residual. The contribution rate of the fit of the individual model elements is also shown. The spectral image is consistent with the phase contrast image, indicating that structural and chemical information about the sample can be obtained at the subcellular level using a simple model of the five fundamental spectra. Since the cells shown in the image are horizontally bisected by the focal plane of the confocal microscope, the cell membrane (mainly phosphatidylcholine and cholesterol) is observed as a circular structure, and the cytoplasmic and DNA contributions are clearly seen as separate features within it. Observed. The average nucleus size of HT29 cells is 10 μm, consistent with the dimensions provided by the Raman image of the cell's DNA content.

  Morphological modeling can be applied to human tissue samples as well. FIGS. 79A-79G show phase contrast images (79A and G) of mildly atherosclerotic arteries with Raman images depicting the distribution of some of the morphological structures (79B-F). The image clearly shows that smooth muscle cells (79E) are found more predominantly in the media while cholesterol (79B), foam cells and necrotic centers (79C) are restricted only to the intima. This finding is consistent with the known structure of atherosclerotic blood vessels. There is only a slight demarcation between one smooth muscle cell and the next. This is because they are very closely spaced and even overlap in the media. The image shows the high spatial resolution of this technique and evidence of elastic plate perforation (a process known to occur with the development of atherosclerosis). The perforations can be observed in a Raman image of the internal elastic plate (IEL), ie FIG. 79D. It can be seen that the smooth muscle cells shown in FIG. 79E migrate through the tear in the IEL to the intima. Smooth muscle cell migration is a feature of progression of atherosclerotic disease. In addition, significant collagen fibers (2F) in the intima above the background of scattered connective tissue (features that are difficult to fully recognize from phase contrast images) can be identified.

FIGS. 80A-80G show Raman images of normal human breast ducts obtained using a morphological model created to specifically analyze breast tissue (FIGS. 80A-D). These images were created by plotting the intensity of two Raman bands characteristic of the DNA phosphate extension (1094 cm ⁻¹ ) and the amide I band (1664 cm ⁻¹ ) (FIGS. 80E and F). ). The morphologically based Raman image represents a region (bright region) where a specific component (cytoplasm (80A), cell nucleus (80B), lipid (80C) or collagen (80D)) contributes strongly to the spectrum. Histological analysis of the tissue sample showed a normal breast duct with a diameter of approximately 25 μm. A typical breast duct of this size consists of a ring of epithelial cells surrounded by a basement membrane (mainly collagen). There is some fat in and around the tube. The morphological model images clearly show the structure of the ducts, while the peak height images produced using the Raman bands found at 1094 and 1664 cm ⁻¹ provide much less information. The phosphate extension of DNA (1094 cm ⁻¹ , FIG. 80E) should be found primarily in the cellular region, while the amide I band (1664 cm ⁻¹ , FIG. 80F) suggesting the protein is primarily found in the collagen region. Nonetheless, the resulting image does not show as clearly as the morphological model image shows neither cellular components nor collagen. This is because the amide I band can be found in many proteins, including those that form the cytoskeleton, while the phosphate extension overlaps with the bands present in the collagen spectrum. The inability of peak height analysis to accurately distinguish morphological features due to spectral overlap results in images that provide much less information.

  The Raman spectrum in FIG. 80G represents a single point in the Raman image. The spectrum is a mixture of many chemical components, all of which contribute to the Raman spectrum. By fitting the spectrum with a morphological model, it is possible to explain the characteristics of the main spectrum in the data. The fit residual also shown in FIG. 80G is predominantly noise, indicating that all of the information in the hypercube of Raman imaging data can be represented by a model-based image.

  Although morphological modeling is an effective means of displaying Raman images, it requires much more advanced knowledge of the sample being examined. As previously discussed, PCA, MCR, and Euclidean distance can also be used to compress data into manageable forms and are much more effective when little is known about the system. 81A-81E show side-by-side comparisons of PCA, MCR, Euclidean distance, and morphological modeling. Images generated from the same data set are of normal breast tissue samples containing three ductal units (mainly cells) surrounded by a collagen matrix. As can be seen, the images created by all four techniques are similar. Euclidean distance images are shown as inverted for easy comparison with other techniques (because they represent differences from the input spectrum rather than similarities that other methods represent). The contribution that can be attributed to collagen is shown on the left, while the lesser contribution of the cell nucleus (mainly DNA) is shown on the right. Both PCA (FIG. 81A) and MCR (FIG. 81B) were able to find seven independently varying fundamental spectra. A complete morphological model of breast tissue has nine fundamental spectra, however, it is very pathologically important but is rarely observed in human breast tissue and is completely absent in this sample. It includes several elements such as microcalcifications that are not observed.

  The first two principal components (two of the spectra derived using MCR) and the basic spectra of collagen and cell nuclei are shown in FIG. 81E. The spectrum of the initial principal component and MCR is similar to that of collagen (the largest contributor to the image). The spectra produced by the second principal component and MCR both contain some features of the nuclear spectrum (as negative peaks), but are seen from the produced images (FIGS. 81A and B, right). As can be, they are much less effective than the morphological model (FIG. 81D, right) in the extraction of nuclear contents within the ductal unit. The filled circle of the ductal unit observed in FIG. 81D (right) is consistent with the pathology of this tissue slice.

  82A and 82B show the normalized fit factor for a particular row of Raman images used on the left of FIGS. 81A-D (column shown in FIG. 82A). PCA (△), MCR (□), Euclidean distance (◯) and morphological model (x) are all some morphological transitions from collagenous to cellular regions of tissue as indicated by changes in strength of the fit factor And the transition is sharpest when using a morphological model. Thus, the morphological model not only provides information about more of the sample components (eg, cell nuclei), but it also produces images with higher resolution.

  The simplest way to display a Raman image is to plot the intensity of a specific Raman band or the ratio of two Raman bands. This analysis method uses only a small portion of the data, and most biological samples contain many compounds with similar spectral characteristics and are not applicable to biological systems. Spectral overlap makes it difficult to obtain structural or chemical information about a sample from a Raman image based solely on peak height.

  These imaging techniques, namely PCA, MCR, Euclidean distance and morphological modeling, are applicable not only to Raman, but also to many other spectroscopic imaging methods such as fluorescence. Each method has its advantages and disadvantages. Some require no (PCA) or little (MCR) prior knowledge of the sample being examined, while others require some (Euclidean distance) or complete (morphological modeling) knowledge . The image quality produced is usually related to how much information is known.

  PCA requires minimal input from the user and as a result is the best tool for inspecting new types of samples. PCA is used to pinpoint regions based on their spectral dispersion. Due to the mathematical process in which they are created, the principal components account for all of the spectral features present in the data. However, because the main components themselves are mathematical constructs, they can be difficult or impossible to correlate with known chemicals. Despite this drawback, information obtained from PCA can be used to build more sophisticated models, such as morphological models developed for breast and arterial tissue.

  While MCR is also mathematically derived, non-negative constraints can be applied to ensure that the generated fundamental spectrum has more identifiable features than those produced by PCA. In fact, the spectrum determined using MCR can be very similar to the true chemical spectrum. The disadvantage of MCR is that the more complex the system being examined (especially when there is a large overlap in spectral features), the more difficult it is to perform the analysis. One skilled in the art can recognize if the MCR fails if the system is simple enough and can adjust the parameters accordingly, but this is when more component spectra are added to the sample mixture. Makes it more difficult. In addition, noise plays an increasingly larger role as more curves are resolved in complex systems. Nevertheless, MCR is extremely useful in obtaining spectral linearity that can be used to direct further analysis of a sample.

  If some (but not all) components of the sample are known, the Euclidean distance is very useful. For example, it is not uncommon to have a sample where the spectrum of the particular chemical being tested is known, but the background chemical is unknown. In this case, the Euclidean distance can map the distribution of that particular chemical within the sample without being disturbed by the lack of background knowledge.

  Morphological modeling is the best technique for detailed analysis of the system, especially for producing images with information content similar to pathology glass slides. However, the development of a good morphological model can be time consuming and requires the acquisition of large amounts of data that is independent of the others. If the model is incomplete, the image gives less accurate information. Therefore, morphological modeling is best used when exhaustive research has been performed and model development is part of the experiment. Raman spectral imaging is a powerful tool for determining chemical information in biological samples. The difficulty is to fully utilize all of the spectral information and reduce it to an image with maximum information content. Preferred embodiments include morphological modeling and imaging approaches, ie PCA, MCR and Euclidean distance.

  A preferred embodiment includes the ability to combine imaging aspects and Raman confocal microscopy to produce tissue or cell images. Aspects of the invention are used to monitor subcellular processes in real time using Raman imaging.

  FIGS. 78A-78G specifically illustrate Raman images (AE) of HT29 cells along with corresponding phase contrast images (F), according to a preferred embodiment of the present invention. A chemical map of the cell is constructed by fitting the Raman spectrum with phosphatidylcholine (A), DNA (B), cholesterol linoleate (C), triolein (D) and morphologically derived cytoplasmic (E) spectra. . G: Shows the spectrum (·) acquired from within the box shown in image E together with the corresponding fit (-) and residual (bottom, zeroed). The contribution ratio of each model element is listed next.

  FIGS. 79A-79G illustrate phase contrast images (A and G) of mildly atherosclerotic arteries according to one preferred embodiment of the present invention, showing internal elastic lamina (IEL) and collagen fibers in G Emphasized. Also shown are Raman images of cholesterol (B), foam cells and necrotic centers (C), IEL (D), smooth muscle cells (E), and collagen (F). Important morphological features such as IEL perforations can be observed.

FIGS. 80A-80G are based on a normal least squares fit of morphologically derived components, namely cytoplasm (A), cell nucleus (B), fat (C) and collagen (D), according to one preferred embodiment of the present invention. A Raman image of a normal breast duct will be specifically described. Images E and F plot the intensity of a single band, the amide I (1664 cm ⁻¹ ) peak based on DNA phosphate (1094 cm ⁻¹ ) and protein, respectively. G shows the proof of fitting of the morphologically based model (•) to the spectrum of individual pixels (located in the region containing cellular content) in the Raman image (−). The fit residual is plotted below the spectrum (zero line drawn).

  81A-81E illustrate a comparison of four different analysis methods for Raman images of a region containing a plurality of ductal units separated by collagen, according to a preferred embodiment of the present invention. An image produced by the first two principal component fit coefficients is shown in A. B: This shows two corresponding images produced by multivariate curve resolution (MCR). C: This shows an image based on Euclidean distance using the collagen (left) and nucleus (right) spectra from the morphological model. The image in D was produced using the fit factor generated by the usual least squares fit with the morphological model and shows only collagen (left) and cell nucleus (right), but the complete model is used. E: Base vector used to create the image, top to bottom: first two principal components, corresponding spectrum produced by MCR, morphologically derived spectrum of collagen and morphologically derived cell nucleus Spectrum is shown. The last two spectra were used for both Euclidean distance measurement and morphological modeling.

  82A and 82B show a preferred embodiment of the present invention: A: Raman image in which the third column is indicated by a white line, and (B) four different models: PCA (Δ), MCR (□), Euclidean distance (◯ ) And the corresponding fit factor heights for the indicated columns obtained using the morphological model (x).

  In view of the broad aspects in which the principles of the invention may be applied, it is to be understood that the specifically described aspects are illustrative only and should not be construed as limiting the scope of the invention. For example, the flowchart steps may occur in an order other than that described, and more or fewer elements may be used in the block diagram. While the various elements of the preferred embodiment have been described as being implemented in software, other aspects of hardware or firmware implementation may alternatively be used, and vice versa.

  It will be apparent to those skilled in the art that the methods involved in the systems and methods for measuring and controlling contamination can be illustrated in computer program products including computer usable media. For example, such computer usable media include a hard drive device having computer readable program code segments stored thereon, a readable storage device such as a CD-ROM, DVD-ROM or computer diskette. Can be included. Computer readable media may also include communication or transmission media such as optical, wired or wireless buses or communication links having program code segments carried thereon as digital or analog signals.

  The operating environment for systems and methods for biological tissue spectroscopy includes a processing system having at least one high speed processor and storage system. In accordance with the practice of those skilled in the art of computer programming, the invention is described with reference to acts and symbolic representations of operations or instructions performed by a processing system, unless indicated otherwise. Such actions and operations or instructions are sometimes referred to as “computer execution” or “processor execution”.

  It will be appreciated that actions or instructions displayed with actions and symbols include manipulation of electrical signals by the processing device. An electrical system with data bits is a conversion or degradation of the electrical signal representation, and thereby the management of data bits at memory locations in the storage device for resetting or otherwise changing the operation of the processing unit, and Causes other processing of the signal. A memory location where a data bit is managed is a physical location having specific electrical, magnetic, optical or organic characteristics corresponding to the data bit.

  Preferred embodiments of the present invention include a side observation probe and / or a front observation probe for collecting Raman spectra. A preferred embodiment implements an optical design to fully utilize the system throughput by characterizing the Raman distribution from the tissue. The embodiment optimized the collection efficiency, minimized noise and resulted in a small diameter highly efficient Raman probe capable of collecting high quality data in 1 second. The performance of the embodiments has been tested in simulations and experiments using tissue models and several in vitro tissue types, indicating that these embodiments can advance Raman spectroscopy as a clinically feasible technique. Preferred embodiments of the invention use Raman spectra to highlight differences in the chemical composition or structure of microcalcifications present in various lesions in the breast. The results from this aspect will advance the understanding of the chemical changes associated with the onset and progression of breast disease and provide important parameters for the diagnosis of breast disease using Raman spectroscopy.

  Preferred embodiments involving Raman probes have shown the diagnostic potential of Raman spectroscopy to distinguish microcalcifications found in benign and malignant lesions. In addition, principal component analysis (PCA) was used to find minor differences in the chemical composition of type II microcalcifications present in benign and malignant breast lesions. Based on the results, we hypothesize that the type II microcalcifications present in benign lesions of the breast have both lower protein and higher calcium carbonate chemical content than those formed in malignant lesions.

  The data bits may also be managed on a computer readable medium including magnetic disks, optical disks, organic disks, and any other volatile or non-volatile mass storage system readable by a processing device. . A computer readable medium is a collaborative or interconnected computer readable medium distributed between interconnected processing systems that can be terminated exclusively at the processing system or local or remote to the processing system. Include.

  The claims should not be read as limited to the described order or elements unless stated to accomplish it. Accordingly, all aspects that fall within the scope and spirit of the following claims and equivalents thereto are claimed as the invention.

The logic, simulation and comparison of the results of the Raman emission data of a turbid sample of blood tissue for the radial distribution of Raman scattered light according to a preferred embodiment of the present invention will be specifically described with reference to the drawings. The simulation, logic and comparison of the results of the Raman emission data of the turbid sample of the blood tissue for the angular distribution of the Raman scattered light according to a preferred embodiment of the present invention will be specifically described with reference to the drawings. FIG. 4 is an illustration of coronary artery morphological reference data for normal coronary artery, non-calcified plaque and calcified plaque, respectively, according to a preferred embodiment of the present invention. It is an illustration of the Raman morphometry method of the coronary artery of one preferable aspect of the present invention. 1 is a longitudinal view of an apparatus including a tissue measurement probe according to a preferred embodiment of the present invention. FIG. 4B is a cross-sectional view of the probe specifically illustrated in FIG. 4A of a preferred embodiment of the present invention. FIG. 2 is a longitudinal view and a cross-sectional view, respectively, of an alternative embodiment of a tissue measurement probe with a parabolic mirror of the system of the present invention. Illustratively illustrates the transmission characteristics of excitation and collection fibers incorporating a filter with respect to the Raman shift of a preferred embodiment of the present invention. The Raman spectrum of non-calcified atherosclerotic plaques collected in 1 second with an excitation power of 100 mW according to a preferred embodiment of the present invention will be specifically described with reference to the drawings. The Raman spectrum of a normal artery of a preferred embodiment of the in vitro system of the present invention will be specifically described with reference to the drawings. 1 is a diagram for specifically explaining a tissue measurement system according to a preferred embodiment of the present invention. The excitation light diffusing through the tissue of one preferred embodiment of the present invention will be specifically described. FIG. 4 is an illustration of the integrated radial distribution, integrated angular distribution, and optimized collection efficiency, respectively, for blood tissue according to a preferred embodiment of the present invention. 2 is an illustration of an excitation spot size of a preferred embodiment of the present invention. It is concrete description of the ray diagram of distribution of the excitation light in the structure | tissue of the preferable one aspect | mode of this invention. It is a concrete description of the ray diagram of the collection efficiency of the probe of one desirable mode of the present invention. The collection efficiency of the probe according to one preferred embodiment of the present invention will be specifically described with reference to the drawings. It is a fragmentary sectional view explaining a part of coronary artery concretely showing a probe of one desirable mode of the present invention. The signal from the ball lens as a function of laser power according to a preferred embodiment of the present invention will be described in detail. A comparison of the data collected using the probe of one preferred embodiment of the present invention and a normal aortic experimental system is illustrated in the figure. The Raman spectrum of normal breast tissue examined using the probe according to one preferred embodiment of the present invention will be specifically described with reference to the drawings. A comparison of the Raman spectra of malignant breast cancer collected using the probe of one preferred embodiment of the present invention and as predicted by the reference data of the present invention is illustrated in the figure. The comparison of the morphological reference data with the calcified aorta data collected using the probe of one preferred embodiment of the present invention is specifically illustrated in the figure. The longitudinal view of an alternative preferred embodiment of the lateral observation probe for tissue measurement of the system of the present invention will be described specifically. Illustrated is a side view probe that delivers light imaged on a portion of the tissue of a preferred embodiment of the invention and a view of a preferred embodiment of Raman light collected from the tissue. 1 is an illustration of a composite Raman macrospectroscopy and confocal microscopy system of one preferred embodiment of the present invention. The Raman spectrum of the morphological structure of the eight selected coronary arteries according to a preferred embodiment of the present invention is illustrated in the figure. FIG. 24A specifically illustrates the results of the contribution of seven morphological structure fits to the calibration data set and diagnostic algorithm classification of a preferred embodiment of the present invention, where FIG. 24A illustrates non-atherosclerotic tissue. Specifically, FIG. 24B specifically illustrates non-calcified atherosclerotic plaque, and FIG. 24C specifically illustrates calcified atherosclerotic plaque. The contribution of the spectrum of β-carotene in the calibration data set for the three diagnostic categories of a preferred embodiment of the present invention is specifically illustrated, where carotenoid levels are expressed in arbitrary units. The result of the algorithm developed using the result of each of the initial calibration data set and the prospective verification data set according to a preferred embodiment of the present invention will be specifically described with reference to the drawings. 1 is an illustration of a system including a confocal Raman microspectrometer of a preferred embodiment of the present invention. FIG. 4 is a micrograph of an internal elastic plate and a Raman spectrum of the internal elastic plate in a 6 μm unstained coronary artery section, each seen under a phase difference according to a preferred embodiment of the present invention. FIG. 4 is a micrograph of the adventitia with collagen fibers in a 6 μm unstained coronary artery section seen under a phase difference in each of the preferred embodiments of the present invention, and a Raman spectrum of the fibers. The Raman spectra of four different smooth muscle cells in the media of one preferred embodiment of the present invention will be specifically described with reference to the drawings. The Raman spectrum of four types of fat cells (adipocytes) in the outer membrane according to a preferred embodiment of the present invention will be specifically described with reference to the drawings. FIG. 4 is a micrograph of foam cells in intimal atherosclerotic plaques in 6 μm unstained coronary artery sections seen under a phase difference of each of the preferred embodiments of the present invention and Raman spectra of foam cells and necrotic centers. The Raman spectrum of cholesterol crystals in the atherosclerotic plaque of the inner membrane according to a preferred embodiment of the present invention will be specifically described with reference to the drawings. FIG. 4 is a photomicrograph and corresponding Raman spectrum of calcification in a necrotic center of an intimal atherosclerotic plaque in a 6 μm unstained coronary artery section seen under a phase difference of a preferred embodiment of the present invention. The Raman basic spectra of the 12 biochemical substances used for the linear fit to the morphological spectrum of one preferred embodiment of the present invention are specifically illustrated in the figure. A preferred embodiment of the present invention provides a graphical comparison between observed and reference data of spectra of various morphological structures in coronary arteries. The biochemical composition of each morphological structure according to a preferred embodiment of the present invention will be specifically described with reference to the drawings. The result of the Raman emission data of the turbid sample of the arterial tissue regarding the radial distribution of the Raman scattered light according to a preferred embodiment of the present invention will be specifically described with reference to the drawings. The result of the Raman emission data of the turbid sample of the arterial tissue regarding the angular distribution of the Raman scattered light according to a preferred embodiment of the present invention will be specifically described with reference to the drawings. FIG. 4 is an illustration of each of the integrated radial distribution, integrated angular distribution and optimized collection efficiency of arterial tissue according to a preferred embodiment of the present invention. In one preferred embodiment of the present invention, the spatial distribution of Raman light emitted from the normal aorta is characterized, where FIG. 39B shows the measured (O) individual radial distribution B ₁ (r) and from the excitation beam Shows a multi-Gaussian fit (line) to the data as a function of distance, and the radial collection efficiency (η ₁ (r), ○) along with a least-squares fit (line) to confirm the Gaussian profile Plot in FIG. 39C. Characterize the integrated angular distribution (η ₂ (θ), (◯), and the theoretical sin ² (θ) distribution (line) for the Lambertian light source, illustrated in FIG. 40B, where: The experiments are in good agreement for the angular range measured according to a preferred embodiment of the present invention. Illustrating the efficiency of a preferred embodiment of the Raman probe of the present invention, where the angular (dashed line) and radial (thin line) collection efficiency as a function of radius is shown, and the product of these curves is the total collection Efficiency η _T (r) (thick line). The simulation result of the excitation spot size of the Raman probe according to one preferred embodiment of the present invention will be described in detail, where a slight concentration occurs at 1 mm from the ball lens without scattering (◯ and solid line) but the probe An immediate divergence occurs when is in contact with the scattering medium (□ and dashed line). The illustration of the Raman spectroscopy apparatus used for the experimental test of the Raman probe of the preferable one aspect | mode of this invention is demonstrated concretely. The Raman spectrum of BaSO ₄ collected with a single ring probe showing the efficiency of the filter module of a preferred embodiment of the present invention is specifically illustrated, where there is minimal evidence of fiber background. The results of a tissue phantom examination showing the signal collection of one preferred embodiment of the invention as a function of transport length are specifically described, where the intensity of the desired perchlorate signal (circles and solid lines) is increased scatter Plot with a line of constant absorption fiber background (□ with dashed line) drawn to show the effect of signal collection with. A comparison of a traditional open optical Raman system with a preferred embodiment of the Raman probe of the present invention is illustrated specifically, showing a slightly increased collection from the Raman probe along with the remaining fiber background. the raw data shown in Figure 46A, the removal of the fiber background and tissue fluorescence, except for the peak of 750 cm ^-1, resulting in the same spectrum diagram 46B just below 1600 cm ^-1 from the probe tip components. Illustrating the Raman spectra of normal breast tissue and malignant breast cancer, respectively, where the data are shown as points with corresponding model fits (lines) and the residuals are plotted down on the same scale. A clinical probe having a total diameter of less than 3 mm according to a preferred embodiment of the present invention is specifically described. FIG. 49A illustrates the acquired Raman spectrum and FIG. 49B specifically illustrates the corresponding histology, specifically illustrating clinical data of normal artery and intimal fiber growth of a preferred embodiment of the present invention. FIG. 50A illustrates the Raman spectrum, and FIGS. 50B and 50C illustrate the corresponding histology, specifically illustrating the clinical data of atheroma plaques according to one preferred embodiment of the present invention. The clinical data acquired about the calcified plaque of one preferable aspect of this invention is demonstrated concretely, Here, FIG. 51A demonstrates a Raman spectrum, FIG. 51B demonstrates corresponding histology concretely. FIG. 52A illustrates the Raman spectrum and FIGS. 52B and 52C illustrate the corresponding histology, specifically illustrating the clinical data obtained for a ruptured plaque of one preferred embodiment of the present invention. FIG. 53A illustrates the Raman spectrum and FIGS. 53B and 53C illustrate the corresponding histology specifically illustrating the clinical data obtained for calcified plaques with thrombus according to one preferred embodiment of the present invention. To do. The figure of the side observation probe of the preferable one aspect | mode of this invention is demonstrated concretely. Illustrating the effect of blood on signal collection according to a preferred embodiment of the present invention, the figure specifically illustrates the spectrum of an artery without blood, an artery with blood, and blood alone, respectively. An illustration of an endoscope system having a Raman probe according to a preferred embodiment of the present invention will be specifically described. It is a flowchart explaining the acquisition method of the data for in vivo Raman spectrum diagnosis of one preferable aspect of this invention concretely. It is a flowchart explaining concretely the processing of the data used by the real-time analysis Raman system of the preferable one aspect | mode of this invention. The Raman image of the normal breast duct [(A)-(C)] with the corresponding serially-stained section (D) according to a preferred embodiment of the present invention will be specifically described. Each image represents the contribution of a particular morphological element to the area under consideration. (A) Collagen; (B) Cytoplasm; (C) Cell nucleus. The fundamental spectrum used in the morphological model of the breast according to one preferred embodiment of the present invention will be described specifically. (A) cytoplasm; (B) cell nucleus; (C) fat; (D) β-carotene; (E) collagen; (F) hydroxyapatite calcium; (G) calcium oxalate; (H) cholesterol-like; (I) water. The Raman spectra of four types of cells observed in normal or diseased human breast tissue according to one preferred embodiment of the present invention will be specifically described. (A) fibroblasts (normal stroma); (B) epithelial cells (fibrocystic disease); (C) epithelial cells (normal ducts); (D) malignant cells. A comparison of commercially available [(A), (C)] and morphologically derived [(B), (D)] Raman spectra observed in cells of one preferred embodiment of the present invention. I will explain it. (A) DNA (Sigma); (D) Cell nucleus (breast tissue); (C) Actin (Sigma); (D) Cytoplasm (breast tissue). A comparison of (A) purified collagen and (B) morphologically derived collagen of one preferred embodiment of the present invention will be specifically described. A comparison of (A) purified triolein and (B) morphologically derived fat according to a preferred embodiment of the present invention will be specifically described. The spectrum of necrotic centers ("cholesterol-like") fitted with cytoplasm, cell nucleus, fat, cholesterol linoleate and cholesterol according to a preferred embodiment of the present invention will be specifically described. The spectrum of the breast deposit of one preferable aspect of the present invention will be specifically described. (A) calcium oxalate dehydrated; (B) hydroxyapatite calcium; (C) β-carotene. The Raman spectrum collected from the extracellular matrix of five patients according to one preferred embodiment of the present invention will be specifically described. The quality of the fit of the model to the macroscopic breast tissue sample of one preferred embodiment of the present invention, ie the normal (●) fit with the model (-), fibrosis and glandular diseases is specifically described. Plot the residuals of the fit under each spectrum (zero line drawn). The percentages shown next represent the fit factor of the base spectrum normalized to the sum to 1 (the water fit factor is not included in the sum). A further illustration of the fit of the model to a macroscopic breast tissue sample of one preferred embodiment of the invention, ie, the fit of the model (-), fibrosis plus cysts (●) with fibroadenoma and invasive ductal carcinoma ( The residual is plotted below). 1 is an illustration of a confocal Raman microscopy instrument system of a preferred embodiment of the present invention. The a taken from the section of the same sample according to a preferred embodiment of the present invention, the X-ray photograph and b of the sample, and the phase contrast image will be specifically described. c, Raman spectrum of type I calcification that occurs in connection with luminal secretion of ductal cysts in the lesions of fibrocystic disease. The region where the Raman spectrum was acquired is emphasized by a box. The a taken from the section of the same sample according to a preferred embodiment of the present invention, the X-ray photograph and b of the sample, and the phase contrast image will be specifically described. c. Raman spectrum of type II calcification in malignant breast lesions. The region where the Raman spectrum was acquired is emphasized by a box. The results of the diagnostic algorithm of type II microcalcification based on the scores of three types of PC (▲, benign; ○, malignant) according to a preferred embodiment of the present invention will be specifically described. An ROC curve that specifically illustrates the ability of Raman spectroscopy to separate microcalcifications present in benign and malignant breast lesions of a preferred embodiment of the present invention is illustrated. A simulated ROC curve of two indistinguishable groups represented by dashed lines is included for comparison. A, PC spectrum 5 of a preferred embodiment of the present invention; b, in order to specifically explain the broadening of the peak at 960 cm ⁻¹ (arrow), the spectrum spectrum (solid line) superimposed with the average (dotted line) of the spectrum from all type II microcalcifications explain. Specifically described PC spectrum 2 which represents the characteristics of a large second derivatives like the preferred embodiment 960 cm -1 ^(arrow) near the present invention. One preferred embodiment of a peak of 1445cm ^-1 of the present invention, the vibration of the amide I of 1650 cm ^-1, and 1004 cm ^-1 the PC spectrum 3 representing the characteristics of the protein directed to a positive, such as features of phenylalanine (arrow) This will be specifically described. The Raman image (AE) of HT29 cell with the corresponding phase difference image (F) of one preferable aspect of the present invention will be specifically described. A chemical map of the cell is constructed by fitting the Raman spectrum with phosphatidylcholine (A), DNA (B), cholesterol linoleate (C), triolein (D) and morphologically derived cytoplasmic (E) spectra. G shows the spectrum (•) acquired from within the box shown in image E along with the corresponding fit (−) and residual (bottom, with zero line drawn). The contribution ratio of each model element is listed next. A phase contrast image (A and B) of a mild atherosclerotic artery of one preferred embodiment of the present invention is specifically illustrated, with the internal elastic lamina (IEL) and collagen fibers highlighted in B. Also shown are Raman images of collagen (C), cholesterol (D), internal elastic lamina (E), foam cells and necrotic centers (F) and smooth muscle cells (G). Important morphological features such as perforations in the inner elastic plate can be observed. A Raman image of normal breast ducts based on a normal least squares fit of the morphologically derived components of the preferred embodiment of the invention, namely cytoplasm (A), cell nucleus (B), fat (C) and collagen (D). This will be specifically described. Images E and F plot the intensity of a single band, the peak of DNA phosphate (1094 cm ⁻¹ ) and protein based amide I (1664 cm ⁻¹ ), respectively. G shows the proof of fitting of the morphologically based model (•) to the spectrum of individual pixels (located in the region containing cellular content) in the Raman image (−). The residual of the fit is plotted below the spectrum is plotted (zero line is drawn). A comparison of four different analysis methods of a Raman image of a region including a plurality of breast duct units separated by collagen according to a preferred embodiment of the present invention will be specifically described. An image produced by the first two principal component fit coefficients is shown in A. B shows two corresponding images produced by multivariate curve resolution (MCR). C shows an image based on Euclidean distance using the collagen (left) and nucleus (right) spectra from the morphological model. The images in D were produced using the fit factor generated by normal least squares fitting with the morphological model, showing only collagen (left) and cell nuclei (right) but using the complete model . E is the basis vector used to create the image from top to bottom: the first two principal components, the corresponding spectrum produced by multivariate curve resolution, the morphologically derived spectrum of collagen and the morphology of the cell nucleus Shows a spectrum derived automatically. The last two spectra were used for both Euclidean distance measurement and morphological modeling. In one preferred embodiment of the present invention, (A) a Raman image in which the third column is indicated by a white line (FIG. 81D, the same on the left), and (B) four different models: PCA (Δ), MCR (□), Euclidean Describe specifically the height of the corresponding fit factor of the indicated column obtained using the distance (O) and the morphological model (x).

Claims (14)

A fiber optic probe having a proximal end and a distal end;
A delivery optical fiber in a probe connected to the light source at the proximal end and having a first filter at the distal end;
A plurality of collection optical fibers in a probe for collecting Raman scattered light from tissue having a second filter at the distal end and connected to a detector at the proximal end; and a first filter on the proximal surface of the rod probe comprising an optical system including a connected rod delivery optical fiber placed module, and a second filter connected to a plurality of collection optical fiber placed on the proximal surface of the cylindrical tube is a cylindrical tube, the The distal end of the optical system , wherein the rod is placed inside the tube and separated from the tube by a sleeve that optically isolates the tube from the rod; Separating the optical fiber at a distance from the distal optical element, the distal optical element including a reflective or refractive optical element,
The including ing in the organization measuring probe. The tissue measurement probe of claim 1, wherein the distal optical element comprises a ball lens optically connected to the delivery fiber and the collection fiber. Around the collection fiber feeding our fibers several are arranged concentrically with the first radius, the concentrically arranged and around the delivery optical fiber at a first radius greater than the second radius having a second plurality of collection optical fibers, tissue measuring probe according to claim 1. The tissue measurement probe of claim 1, further comprising a controller that gates a collection time that is less than 2 seconds. The tissue measurement probe according to claim 1, wherein the optical system has a length of less than 10 mm. The tissue measurement probe according to claim 1, wherein the optical system has a length of less than 4 mm. Further comprising a light source connected to the delivery optical fiber having a wavelength longer than 750 nm, the tissue measuring probe according to claim 1. The tissue measurement probe of claim 1, wherein the distal optical element delivers and collects light in a radial direction. The tissue measurement probe according to claim 1, wherein the distal end has a diameter of 2 mm or less. The tissue measurement probe of claim 1, further comprising a laser light source optically connected to a proximal end of the delivery optical fiber. The tissue measurement probe according to claim 1, wherein the distal optical element comprises a refractive optical lens. The tissue measurement probe of claim 1, wherein the distal optical element comprises a reflective optical element that collects light in a radial direction. The tissue measurement probe of claim 1, wherein the distal optical element comprises a portion of a ball lens. An endoscope system comprising the probe according to claim 1 and an endoscope having a channel into which the probe is inserted.