JP7348642B2 - Data processing device, X-ray device equipped with the data processing device, and data processing method - Google Patents

Description

The present invention relates to a data processing apparatus that processes X-ray transmission data of an object imaged using continuous X-rays, an X-ray apparatus equipped with the data processing apparatus, and a data processing method. In particular, the present invention provides a data processing device for processing X-ray transmission data collected using a photon counting type X-ray detector to evaluate the properties of an object, such as bone mineral quantification. , an X-ray apparatus equipped with the data processing apparatus, and a data processing method.

In recent years, there has been a growing desire to identify the type and properties of objects using X-rays. One example is bone densitometry (bone mineral quantification), which measures the density of bones in the human body for the prevention and treatment of osteoporosis. Preventing osteoporosis is an important factor in extending a healthy lifespan.

In osteoporosis diagnosis, in addition to bone density (bone mineral content), which is a typical index for determining bone strength, recently bone quality has also come to be emphasized. Bone density (bone mineral content) indicates the amount of minerals such as calcium in the bone, and bone quality is determined by the microstructure of the bone, the speed of bone turnover, the presence or absence of microfractures, and calcification. It is an indicator that indicates the state of collagen.

Conventionally, the DEXA (dual-energy X-ray absorptiometry) method, the ultrasonic method, and the MD method are often used for this osteoporosis diagnosis. The DEXA method uses two types of X-rays with different energies and measures bone density (bone mineral content) based on the difference information between the X-ray transmission data after the two types of X-rays pass through the bone. It is a method. The ultrasonic method is a method of measuring bone density by irradiating ultrasonic waves onto the bones of the heel or tibia and using the reflected information. The MD method takes X-rays of the bones of the hand and aluminum plates (reference material) of different thicknesses at the same time, and measures bone density by comparing the density of the bones and the aluminum plate on the taken X-ray images. This is a method to do so. In addition, by performing an X-ray examination, it is possible to confirm the presence or absence of fractures, deformities, osteoporosis (reduction in bone density), etc. from the X-ray photographs. On the other hand, bone quality is generally measured by evaluating the speed of bone metabolism through a blood test or urine test using bone metabolism markers.

An example of such an osteoporosis diagnostic method is known as described in Patent Document 1. The method described in Patent Document 1 belongs to the DEXA method, and is disclosed as an X-ray image diagnostic apparatus. According to one aspect of the method, an X-ray beam having two different average energies, high and low, is used a differential image generation unit that generates a differential image from each taken X-ray image; a detection unit that detects a lumbar region from the differential image and also detects a transverse process region and a soft tissue region from a peripheral region of the lumbar vertebrae; and a detected soft tissue region. The apparatus includes a correction section that corrects the pixel value of the lumbar region based on the pixel value of the tissue region, and a bone density calculation section that calculates bone density based on the corrected lumbar region.

In other words, this diagnostic method utilizes the fact that the degree of attenuation of the bone component of X-rays differs depending on the energy.

By the way, the present inventors have already developed a data processing method described in Patent Document 2 (International Publication Number WO 2016/171186 A1 (International Publication Date: October 27, 2016) regarding this photon counting type X-ray detection). proposing a law.

According to this proposal, an image of the object is created from the counts detected by a photon-counting X-ray detector, a region of interest is set on the image, and the background of the material present in the region of interest is Delete the pixel information that becomes . Furthermore, for each of the multiple energy regions of continuous X-rays in the region of interest, the transmission characteristics when the X-rays pass through the material are expressed as vector quantities based on the count values for each pixel, and from this vector quantity Calculate unique unique information.

If the energy regions of continuous X-rays are, for example, three, and the linear attenuation coefficients corresponding to the effective energy of each energy region are μ ₁ , μ ₂ , μ ₃ , then the vector quantity described above is expressed as the linear attenuation coefficient μ ₁ , μ ₂ and μ ₃ as a three-dimensional vector in a three-dimensional space.

Of course, due to the effects of statistical noise and material thickness t, the direction of the three-dimensional vector of each of the multiple pixels forming the region of interest varies, and the length of the vector also varies, so the ending point of the vector starting from the coordinate origin is 3. It is dispersed in a dimensional manner. Therefore, in the method of Patent Document 2, in order to eliminate the factor of the thickness t of the material, all three-dimensional vectors are normalized, and then a spherical dispersion surface whose radius is the normalized length is created. , is displayed as the position (scattered point) of the tip (end point) of the vector connecting it to the coordinate origin. Then, in order to remove the influence of statistical noise, averaging processing is performed, such as calculating the centroid position of the distribution for each set of scattered points, and the vector connecting the centroid position and the coordinate origin is calculated based on the distribution of the set distribution. is set as a representative three-dimensional vector. The direction of this representative three-dimensional vector is unique to the substance or to the properties of the substance from the viewpoint of attenuation of continuous X-rays. In other words, the method of Patent Publication 2 compares the vector direction with a reference value measured and calculated using a known phantom, or a theoretically calculated reference value, to determine the type and properties of the substance. It was said that it could be identified (estimated and evaluated). Further, in the method of Patent Publication Document 2, the dimensions to be processed are not limited to three dimensions, but the method is applicable to two or more dimensions.

Japanese Patent Application Publication No. 2017-131427 International Publication No. 2016/171186

<About Patent Document 1>

However, the conventional DEXA method, typified by the above-mentioned Patent Document 1, only measures bone density (bone mineral content) using the difference in energy information of two types of high and low X-rays. It does not take into account the factors that are attracting attention these days in the diagnosis and treatment of osteoporosis. Therefore, this DEXA method does not completely meet recent demands for the diagnosis and treatment of osteoporosis.
<About Patent Document 2>

On the other hand, in the above-mentioned Patent Document 2, the energy spectrum of photons constituting continuous X-rays is divided into, for example, three energy regions, and the linear attenuation coefficients corresponding to the effective energy of each energy region are μ ₁ and μ ₂ . , μ ₃ as each dimension. Therefore, when the physical meaning of this three-dimensional vector is applied to bone evaluation, the direction and magnitude of this three-dimensional vector are an indicator of bone condition, including bone quality and bone density (bone mineral content). It is thought that this is the case.

The reason is that even if the materials have the same name (for example, bone), the degree of attenuation of X-ray photons passing through the tissue of that material will differ depending on the energy level of the X-ray photons. Even if there are substances, if their constituent compositions are different, they can be considered to be different types of substances from the viewpoint of X-ray attenuation. For example, even if the bone is the same material, the three-dimensional vector described above is The directions are different from each other. In other words, the direction and magnitude of the vector can be said to be one of the indicators indicating not only the bone mineral content but also the difference in the composition and components of the bone (that is, the difference in bone quality).

However, even if a scatter diagram of three-dimensional vectors that does not depend on the material thickness t proposed in Patent Document 2 is used, only length information of three-dimensional vectors that is not normalized, called an absorption vector length image, is used. Even if it is used, it is not sufficient to measure and evaluate bone density and bone quality, which is required for the diagnosis and treatment of osteoporosis. In other words, both are not accurate as they are.

For example, by taking an X-ray of your hand, we can quantify the amount of bone mineral reflected in that area (bone mineral refers to inorganic bone mineral, and the chemical substance is known to be hydroxyapatite). (For the sake of explanation, we will assume that bones are composed only of hard tissue (corresponding to bone mineral).) In this case, a region of interest is set in the bone portion reflected in the X-ray image. In the X-ray path leading to each pixel of the X-ray detector that constitutes that area, both soft tissues such as skin and muscles and hard tissues such as bones exist. Therefore, the X-ray detector collects counts that depend on the attenuation coefficient and thickness of soft and hard tissues. Therefore, even if the aforementioned three-dimensional vector is calculated from this count value, it is not possible to obtain a three-dimensional vector that accurately reflects the attenuation characteristics due to hard tissue (bone mineral). Therefore, it is difficult to perform bone mineral quantification based on this with higher accuracy.

The above-mentioned problems have been discussed with respect to osteoporosis diagnosis in the general X-ray imaging field, taking bone mineral quantification as an example, but we will take a bird's-eye view of the field of so-called material identification, where we want to identify the type and shape of an object. As you can see, this is not necessarily limited to the field of general X-ray imaging. For example, when examining bone mineral density in dentistry or foreign matter in food, if there are two substances on the X-ray path with different attenuation coefficients and unknown thicknesses, it is necessary to It is difficult to easily identify the properties of a single substance of interest.

Therefore, the present invention was made in view of the above-mentioned substance identification in conventional photon counting type X-ray detection, and even when the target consists of essentially two types of known substances A and B, On the other hand, the properties of the substance A of interest can be determined with simpler calculations and with higher accuracy, and information on the properties of the substance of interest can be provided in a more multifaceted manner, such as bone density and bone quality. The object of the present invention is to provide a data processing device and a data processing method suitable for photon counting type X-ray detection, and an X-ray device equipped with or mounted with them.

In order to achieve the above object, a data processing device according to one aspect of the present invention is capable of processing continuous X-rays including n (n is a positive integer of 2 or more) different energy regions with respect to the transmission characteristics of the X-rays. When the X-rays that are irradiated onto an object consisting of substantially two types of known substances A and B and transmitted through the object are detected as detection data by an X-ray detector having a plurality of pixels, the detection Perform processing based on data. This data processing device includes an X-ray transmission image providing means for creating an X-ray image of the object based on the detection data detected by the X-ray detector and displaying the X-ray image on a monitor; ROI setting means for setting an ROI (region of interest) on an X-ray image in a portion where the total thickness of the substances A and B is estimated to be constant in the direction of the path of the X-ray; one n-dimensional line attenuation corresponding to the line attenuation value when the X-ray passes through the object in each of the energy regions, and whose magnitude is averaged over the plurality of pixels forming the ROI; linear attenuation vector calculation means for calculating a vector based on the detection data; a reference vector holding means for estimating or assuming and holding the n-dimensional reference vector corresponding to the linear attenuation value; and a reference vector holding means for estimating or assuming and holding the n-dimensional reference vector corresponding to the linear attenuation value; 1. A data processing device comprising: objective vector calculation means for subtracting the n-dimensional reference vector from the n-dimensional linear attenuation vector calculated by the attenuation vector calculation means. At this time, the data processing device can also be equipped with calculation means for estimating or assuming the reference vector.

Further, as another aspect of the present invention, a data processing method is provided. In this data processing method, continuous X-rays containing n (n is a positive integer of 2 or more) different energy regions are obtained from substantially two types of known substances A and B with respect to the transmission characteristics of the X-rays. When the X-rays that are irradiated onto an object and transmitted through the object are detected as detection data by an X-ray detector having a plurality of pixels, processing is performed based on the detection data. According to this data processing method, an X-ray image of the object is created based on the detection data detected by the X-ray detector and displayed on a monitor, and an X-ray image is displayed on the X-ray image displayed on the monitor. , an ROI (region of interest) is set in a portion where the total thickness of the substances A and B is estimated to be constant in the direction of the path of the X-ray, and the calculating one n-dimensional line attenuation vector, which corresponds to the line attenuation value when transmitting through the object and whose size is averaged over the plurality of pixels forming the ROI, based on the detected data; and estimating or The hypothetically held reference vector is subtracted from the n-dimensional linear attenuation vector calculated by the linear attenuation vector calculation means to obtain the n corresponding to the linear attenuation value equivalent only to the substance A. It is characterized by calculating a dimensional objective vector. At this time, the data processing method preferably includes calculation processing for estimating or assuming a reference vector.

This data processing device and data processing method can be installed in a bone mineral quantification device, for example. As an example of that case, substance A is the bones (hard tissues) of the subject's limbs, and substance B is the skin and muscles (soft tissues) thereof.

According to the present invention, continuous X-rays containing n (n is a positive integer of 2 or more) different energy regions are obtained from substantially two types of known substances A and B with respect to the transmission characteristics of the X-rays. The X-rays irradiated onto an object and transmitted through the object are processed based on the count value detected by an X-ray detector having a plurality of pixels.

More specifically, an X-ray image of the object is created based on the counted value and displayed on a monitor. On the X-ray image displayed on this monitor, a ROI (region of interest) is set at a portion where the total thickness of the substances A and B is estimated to be constant, for example, in the direction of the X-ray path. be done. Furthermore, in each of the n energy regions, the size corresponds to a linear attenuation value when the X-ray passes through the object, and the size is averaged over the plurality of pixels forming the ROI. One n-dimensional (n is a positive integer of 2 or more) linear attenuation vector is calculated based on the count value. Assuming that the X-rays have passed through only one of the two types of substances A and B, the n-dimensional reference vector corresponding to the linear attenuation value of the substance B is defined as the linear attenuation vector. By subtracting from the n-dimensional linear attenuation vector calculated by the calculation means, the n-dimensional objective vector corresponding to the linear attenuation value equivalent only to the substance A is calculated. At this time, the reference vector may be set in advance theoretically or experimentally, or in the case of a subject for which it is possible to set an ROI for reference vector calculation (for example, if the object consists only of substance B on the X-ray path). (for example, if there is a location where the line attenuation vector exists) is calculated as appropriate in real time or from one image using a method similar to the method for deriving the line attenuation vector.

Therefore, on the X-ray image obtained by continuous X-ray irradiation, the ROI is set at a desired portion where the total thickness of the substances A and B is estimated to be constant, for example. Furthermore, one n-dimensional line attenuation vector whose size is averaged over a plurality of pixels forming the ROI is calculated. This n-dimensional line attenuation vector reflects the X-ray transmission characteristics obtained by combining two different types of substances A and B in the ROI.

Therefore, substance A is the bone tissue (hard tissue) included in the object (for example, human limbs), substance B is the muscle and skin tissue (soft tissue) of that part, and the target site is the bone part (substance A ). In this case, muscle and skin tissues correspond to areas that are difficult to diagnose. However, if it is assumed that X-rays have passed through this substance B, is information on the n-dimensional reference vector set in advance and corresponding to the linear attenuation value of the substance B stored in advance, for example, in a memory? Since it can be easily calculated, the information of this reference vector is read out from memory or calculated. Further, this reference vector is subtracted from the n-dimensional linear attenuation vector corresponding to the combined X-ray transmission properties of the two substances A and B. As a result, an n-dimensional objective vector corresponding to a linear attenuation value equivalent only to the substance A is obtained.

Therefore, an objective vector is obtained that indicates the X-ray attenuation of only the material A, for example, a bone portion. This target vector reflects the degree of attenuation of continuous X-ray photons with a continuous energy distribution from low energy to high energy when they pass through the tissue of the bone, so it reflects the density and quality of the bone. Count values that more accurately represent (properties) can be collected. Furthermore, even if both substances A and B are included in the ROI portion, a target vector that reflects the linear attenuation of only the target substance A can be extracted with higher precision by a simple operation called vector subtraction.

In other words, in the conventional case, even if the types of substances A and B (conceptual types such as bones and muscles) are known and their total thickness is constant, the thickness of each in the X-ray path direction is unknown. However, since there are two substances with different X-ray transmission characteristics, it has been difficult to obtain vector information based on the X-ray attenuation of only the target substance A with simple calculations and with high accuracy.

However, according to the present invention, a target vector that reflects the linear attenuation of only the target substance A can be derived easily and accurately by vector calculation for each region of interest.

Further, according to the present invention, a target vector of substance A is obtained for each region of interest, and the length of the vector indicates the bone mass (bone density) per unit volume in the case of bones, for example. Information indicating a more multifaceted property (state) of a substance can be provided for each region of interest, such as the direction indicating bone quality in the case of bone. Unlike conventional processing that provides information based only on bone density, it is possible to enrich the property information provided and meet the demands for diagnosis and treatment of osteoporosis, for example.
This effect can be similarly enjoyed in the data processing method and X-ray apparatus according to the present invention.

Preferably, the reference vector is i) estimated from the external size or weight of the part of the object to be irradiated with X-rays, or ii) statistically collected in advance and compiled into a database. iii) In a partial region of the imaged region of the target in which only the substance B is present, it is equivalent to the linear attenuation vector (since there is only substance B, in this region, the linear attenuation vector = reference vector = line attenuation vector of only material B) and call up a reference vector that has been obtained and stored in advance, or simultaneously image the object when measuring the line attenuation vector of the region of interest that includes material A. It may be set by performing calculations on the assumption that a partial region containing only the substance B in the region is equivalent to the linear attenuation vector. Furthermore, for example, if the direction of the reference vector exhibited by the portion of substance B is known, its magnitude may be estimated from experiments or theoretical calculations and previously held information can be used. Therefore, in that case, the calculations required for vector subtraction can be further simplified.

Furthermore, direction information about the direction of the reference vector itself obtained experimentally or empirically may be stored in advance and called up when necessary. This greatly simplifies the amount of calculation for the reference vector.

In the attached drawings,
FIG. 1 is a block diagram illustrating an overview of an X-ray inspection system equipped with a data processing device according to one embodiment of the present invention. FIG. 2 is a diagram illustrating an example of an energy range and an X-ray energy spectrum set in a photon counting type detector. FIG. 3 is a diagram illustrating the relationship between a single material model and photon counting for each energy region. FIG. 4 is a diagram illustrating the relationship between the multi-material model and photon counting for each energy region. FIG. 5 is a flowchart outlining the substance identification process and its preprocessing executed by the data processor. FIG. 6 is a diagram illustrating preprocessing for substance identification performed by the data processor. FIG. 7 is a schematic flowchart illustrating the central part of substance identification performed by the data processor in this embodiment. FIG. 8 is a diagram illustrating generation of a three-dimensional vector of X-ray absorption amount from each pixel of a region of interest of an image for each energy region. FIG. 9 is a schematic flowchart illustrating the process from creating a three-dimensional scatter diagram to presenting identification information. FIG. 10 is a partial flowchart illustrating in more detail the process of step S134 in FIG. 7 according to the embodiment described above. FIG. 11 is a perspective view schematically illustrating a normalized three-dimensional scatter diagram. FIG. 12 is a diagram illustrating generation of a three-dimensional vector from substance-specific scattering points from a three-dimensional scatter diagram. FIG. 13 is a diagram illustrating the process from setting an ROI to the bone of the back of the hand (finger) as a target object to calculating one target vector representing only the bone portion specified by the ROI. FIG. 14 is a schematic flowchart illustrating the bone mineral quantification process executed by the data processor. FIG. 15 is a diagram illustrating the relationship between the cross section of the finger portion and the X-ray path passing through it.

Hereinafter, embodiments of a data processing device and a data processing method for X-ray inspection according to the present invention will be described with reference to the accompanying drawings, and an X-ray inspection device equipped with the data processing device will be described as a modified example. .
[Embodiment]

First, with reference to FIGS. 1 to 15, a data processing apparatus and data processing method for X-ray inspection according to one aspect of the present invention will be described as one embodiment. FIG. 1 shows a schematic configuration of an X-ray inspection system. This X-ray inspection system includes an X-ray inspection device 10 that functions as an X-ray device.

This X-ray inspection system also includes a data processing device 12 equipped with and installed a data processing device and a data processing method for performing bone density measurement (bone mineral quantification) according to one embodiment of the present invention (see FIG. 1). ). This data processing device 12 may be integrated as an element of a system that collects X-ray data, or may be provided in a stand-alone manner as a general-purpose computer separate from the X-ray data collection system. In the case of a stand-alone system, the X-ray data acquisition system can be connected, for example, via the Internet, and the computer can also be configured to read the X-ray acquisition data and perform bone mineral quantification processing.

Note that this data processing device 12 may have other necessary programs installed in advance so as to execute processes other than those for bone mineral quantification.

Furthermore, excluding the functional configuration and effects of bone mineral quantification in this X-ray examination system, the present applicant has already disclosed the publication "WO 2016/171186 A1" (title of invention: X-ray examination). A data processing device and a data processing method for this purpose, as well as an X-ray inspection device equipped with the device, can be configured in the same manner as the device described in Job Co., Ltd. (Applicant: Job Co., Ltd.). However, in this embodiment, it is not necessarily necessary to scan the X-ray beam together with the collimator, and spot imaging (simple imaging) may be performed by irradiating X-rays once without scanning.

Furthermore, it is effective to perform bone mineral quantification according to this embodiment in combination with so-called substance identification, which has already been proposed by the applicant. Therefore, in the description of this embodiment, the configuration of the X-ray inspection system, the substance identification process, and the bone mineral quantification process will be described in this order.
<Configuration of X-ray inspection system>

As shown in FIG. 1, a data processing device 12 is communicably connected to the X-ray inspection device 10 via a communication line LN, and this data processing device 12 is connected to, for example, a control section of the X-ray inspection device 10. It may be integrated into one piece or may be installed separately.

In this embodiment, the X-ray inspection apparatus 10 is configured to generate X-ray images (reconstructed images based on the laminography method or obtained by spot imaging) of the hands and feet (examination target OB) of a patient (human body) as a subject. It is configured to perform bone mineral quantification based on X-ray transmission images). Of course, this apparatus 10 may be provided as a non-destructive inspection system for inspecting food for foreign substances using X-rays, or as a medical X-ray panoramic imaging system. Bone mineral quantification can also be considered, in a broad sense, as a form of non-destructive testing.

Hereinafter, the X-ray inspection apparatus 10 according to the present embodiment will be described as one that performs bone mineral quantification using a laminography method in which the X-ray beam is scanned by relative movement between the X-ray beam and the subject. .

When performing bone mineral quantification, which is an aspect of non-destructive testing, the data processing device and data processing method for X-ray testing according to this embodiment collects absorption information (or attenuation information) when X-rays pass through a substance. The basic element is to perform processing to identify (also be referred to as specifying, discriminating, discriminating, or determining) the type and properties of the substance based on the above. In the following explanation, this process may be collectively referred to as "substance identification."

As shown in FIG. 1, this X-ray inspection apparatus 10 has an object space OS that can virtually set an orthogonal coordinate system of X, Y, and Z axes. Among these, the Z-axis direction corresponds to the scan direction in the object space OS in the case of non-destructive inspection. This device 10 generates an X-ray beam having a predetermined cone angle θ in the Z-axis direction and a predetermined fan angle β in a direction (Y-axis direction) along a cross section (XY plane) perpendicular to the scanning direction. An X-ray generator 23 including an X-ray tube 21 and a collimator 22 for generating XB is provided. The X-ray tube 21 is, for example, a rotating anode X-ray tube having a point-like X-ray tube focal point F (focal diameter is, for example, 1.0 mmφ). This X-ray tube 21 is supplied with a driving high voltage for X-ray irradiation from an X-ray high voltage device (not shown).

Furthermore, the X-ray inspection apparatus 10 includes an X-ray detector 24 (hereinafter also simply referred to as a detector) disposed so as to be able to face the X-ray tube 21 at a distance from the X-ray tube 21 . The detector 24 is configured by connecting a plurality of modules in a line, so that the detector 24 as a whole has an elongated rectangular X-ray entrance window. Each module consists of a detection layer made of a semiconductor material such as CdTe or CZT (CdZnTe) formed into, for example, 20 x 80 pixels (each pixel has a size of 0.2 mm x 0.2 mm). This is a so-called direct conversion type X-ray detection element that directly converts into a signal. Although not shown, charging electrodes and collection electrodes are actually pasted on both sides of the detection layer forming the plurality of pixels. A bias voltage is applied between these two electrodes.

This detector 24 is a photon counting type detector that regards X-rays as a collection of photons with various energies and can count the number of photons in each energy range. It is a vessel. As this energy region, for example, as shown in FIG. 2, four energy regions Bin ₁ to Bin ₄ are set. Of course, the number of energy regions Bin may be plural.

In this detector 24, the X-ray intensity is detected for each pixel and for each energy region Bin as the number of X-ray photons per unit time (actually, the cumulative number of photons over a certain period of time is measured). ). When one photon enters a certain pixel, an electric pulse signal having a peak value corresponding to the energy value of the photon is generated at the collecting electrode corresponding to that pixel. The peak value, that is, the energy value, of this electric pulse signal is classified into each predetermined energy region Bin by a measuring circuit downstream from the collecting electrode, and its count value is increased by one. This count value is collected as a cumulative value (digital value) at fixed time intervals for each pixel and for each energy region Bin.

This acquisition is performed by a data acquisition circuit 25 built as an ASIC layer on the underside of the detection layer of the detector 24. The detector 24 and the data acquisition circuit 25 constitute a detection unit 26. Therefore, X-ray transmission data (frame data) is sent from the detection unit 26, that is, the data acquisition circuit 25, to the data processing device 12 at a constant image transfer rate (frame rate). Note that a frame is a unit of data transfer, and is, for example, a collection of data collected at each pixel over a certain period of time, like a still image.

An example of an X-ray inspection system having such a configuration is shown in, for example, Japanese Patent Application Publication No. 2007-136163, International Publication No. WO 2007/110465 A1, and International Publication No. WO 2013/047778 A1. Further, an example of the photon counting type detector 24 described above is also shown in, for example, International Publication No. WO 2012/144589 A1.

Note that when this X-ray inspection system is used for, for example, dental X-ray panoramic photography, the inspection object OB is the head of the subject. In that case, the pair of X-ray generator 23 and detector 24 rotate around the head, facing each other with the head in between, for example. A scan structure related to this X-ray panoramic imaging is also shown in, for example, Japanese Patent Laid-Open No. 2007-136163. Bone mineral quantification is not necessarily limited to limb bones, but is performed on bones in various parts of the body. For this reason, the subject's jaw is also one of the targets for bone mineral quantification.

Furthermore, the data processing device 12 receives X-ray transmission data (frame data) from the X-ray inspection device 10 via the communication line LN.

As will be described in detail below, this data processing device 12 processes this X-ray transmission data to obtain information specific to the type or property of the substance that constitutes the inspection object itself or the substance present in the region of interest of the inspection object. (unique information), and furthermore, it is configured to be able to detect whether or not other substances such as foreign substances are present in the test object, and to perform bone mineral quantification.
[Acquisition of substance-specific information and data processing for bone mineral quantification]
The configuration and operation of the data processing device 12 will be described below based on bone mineral quantification performed together with substance identification.

The data processing device 12 is configured by, for example, a computer system CP. As shown in FIG. 1, the computer system CP itself may be a computer system having a known calculation function, and includes an interface (I/O) 31 connected to the detection unit 26 via a communication line LN. This interface 31 is connected to a buffer memory 32, a ROM (read-only memory) 33 (functioning as a "Non-transitory computer readable medium"), a RAM (random access memory) 34, and a CPU (central processing unit) via an internal bus B. ) (this device may be simply called a processor or a computer), an image memory 36, an input device 37, and a display device 38 are connected to each other so as to be able to communicate with each other.

The ROM 33 stores in advance a computer-readable substance identification and bone mineral quantification program, and the data processor 35 reads the program into its own work area and executes it. Buffer memory 32 is used to temporarily store frame data sent from detection unit 26. The RAM 34 is used to temporarily store data necessary for the calculation when the data processor 35 performs the calculation.

The image memory 36 is composed of, for example, an SSD (solid state device) or an HDD (hard disk drive), and is used to store various image data and information processed by the data processor 35. The input device 37 and the display 38 function as a man-machine interface with the user, and the input device 37 receives input information from the user. The display 38 can display images and the like under the control of the data processor 35. The interface 31, input device 37, and display 38 constitute an interface unit that obtains information from the outside (for example, information from a user).
・Relationship between data collection using photon counting method and material model

Next, the X-rays irradiated from the X-ray tube 21 (fan-shaped beam X-rays) transmit through the object OB, and the transmitted X-rays are collected by the detector 24 under the photon counting method ( The concept of data collection for each material model when counting) will be explained using FIGS. 2 to 4.

Figure 2 shows X-rays measured by an X-ray detector, with the horizontal axis representing the X-ray energy [keV] and the vertical axis representing the incident frequency (count) of photons that form X-rays. The general profile of the spectrum is shown. In the case of photon counting, as is well known, a threshold value TH is set to divide the energy on the horizontal axis into a plurality of energy regions Bin. In the example of FIG. 2, four threshold values TH ₁ , TH ₂ , TH ₃ , TH ₄ are provided as appropriate reference voltage values to a comparator (not shown), thereby making the first to fourth thresholds available. Three energy regions Bin ₁ , Bin ₂ , and Bin ₃ are set. Note that the energy below the first energy region Bin ₁ belongs to the energy region where there is a lot of noise and cannot be measured, while the fourth energy region Bin ₄ located above the highest threshold TH ₄ belongs to the energy region that is not measurable. It is not used as it is not involved. Therefore, in this example, the first to third energy regions Bin ₁ , Bin ₂ , and Bin ₃ , excluding the highest and lowest energy regions, are used for photon counting.

The shape of the frequency profile shown in FIG. 2 is also determined by the type of anode material of the X-ray tube 21 and the tube voltage, and normally, as shown in the figure, the count in the second energy region Bin ₂ is the largest. For this reason, the threshold value TH is appropriately determined in consideration of the balance of the count values (frequency, counts) for each energy region. These four thresholds TH ₁ to TH ₄ are set as voltage thresholds to the comparator for each pixel of the detector 24 in the ASIC forming the data acquisition circuit 25. Therefore, X-ray photons are counted for each pixel and for each energy region. Of course, the number of threshold values TH for each pixel may be any number as long as it is three or more. If the number of threshold values TH is three, the number of energy regions used is two. Furthermore, if the count component included in Bin ₄ is considered to be zero, the value of Bin ₃ + Bin ₄ can be _used instead of the count of Bin 3 without setting TH ₄ . In this case, the number of threshold values TH for each pixel may be any positive integer as long as it is two or more. Therefore, in this case, if the number of threshold values TH is three, the number n (positive integer) of energy regions used is two.

When creating an X-ray transmission image (density image) from this count value, various forms can be taken. Count information is obtained for each pixel forming the X-ray entrance surface of the detector 24 and for each energy region Bin. Therefore, when moving the OB to be inspected relatively, it is recommended to multiply the count value of each pixel for each energy region Bin by an appropriate weighting coefficient and perform shift & add, or to keep the OB to be inspected still. In this case, X-ray transmission data (frame data) for each energy region Bin can be obtained by simply adding the count value of each pixel for each energy region Bin to each pixel for each energy region Bin. Also, among these three energy regions Bin ₁ to Bin ₃ , the count values of any two or all energy regions Bin are multiplied by an appropriate weighting coefficient and added to the pixels at the same position, and the result is It may also be X-ray transmission data.

In this way, the number of X-ray photons is collected for each pixel in each energy region Bin, and it can be used for image creation, etc., taking into account the contribution of photon energy to the pixel, so it is possible to freely create energy-enhanced images according to the purpose. can be created, which is superior to conventional integral type X-ray transmission data collection.

Now, when this photon counting method is applied to substance identification, a substance (a substance located in the part of the OB to be inspected; it may be a substance forming the inspection subject itself or a substance other than the inspection subject). It is appropriate to consider whether it is made of a single tissue or multiple tissues, and to consider the X-ray absorption of each tissue. For example, when quantifying bone minerals from bones of limbs, X-rays are used for skin/muscle (portion B in FIG. 15), bone (cancellous bone, cortical bone) (Fig. 15 A part), muscle/skin (Fig. 15 B part), so there are two general names for the substance: soft tissue (skin/muscle) and hard tissue (bone). It can be broadly divided into
(i) When a substance consists of a single tissue (single substance model)

In the case of this single material model, for example in FIG. 15, the materials in part A and part B are the same, and as shown in FIG. 3(A), the first, second, and third energy regions Bin Linear attenuation coefficients representing each of _{Bin 1} , Bin ₂ , and Bin ₃ can be expressed as μ ₁ , μ ₂ , and μ ₃ (cm ⁻¹ ). This linear attenuation coefficient is an index indicating the inherent transmission characteristics of a substance for X-rays.

A model when X-rays are incident on a material having different linear attenuation coefficients μ ₁ , μ ₂ , μ ₃ for each energy region Bin and having a thickness t (cm) is expressed as shown in the figure. In other words, the incident X-ray dose (photon number) C _l1 , C _l2 , C _l3 undergoes attenuation depending on the linear attenuation coefficient μ ₁ , μ ₂ , μ ₃ and the thickness t, respectively, and the output X-ray dose (photon number) number) C _o1 , C _o2 , C _o3 are C _o1 = C _l1 ×e ^{- μ1t}
C _o2 = C _l2 ×e ^{- μ2t}
C _o3 = C _l3 ×e ^{- μ3t}
... (1)
It can be expressed as.

Therefore, in the case of a single material model consisting of a single tissue, as shown in FIG. 3(B), when the X-ray dose (photon number) C _li is incident on a material with a linear attenuation coefficient μ _i and a thickness t, The output X-ray dose (number of photons) _Coi is _Coi = C _li ×e ^{- μit}
(i=1~3)
... (2)
It can be expressed as.
(ii) When a substance consists of multiple tissues (multiple substance model)

In the case of this multi-material model (a model in which n tissues overlap in layers), from the perspective of X-ray attenuation, as shown in FIG. 4(B), the material has a thickness t _a and a linear attenuation coefficient. It can be said that it has a layered structure in which a layer μ _ia , a layer having a thickness t _b and a linear attenuation coefficient μ _ib , . . . a layer having a thickness t _n and a linear attenuation coefficient μ _in are laminated. (The subscript i takes a value of 1 to 3 and corresponds to the subscripts of Bin ₁ to Bin _3. ) Therefore, as shown in FIG. 4(A), the first to third energy regions Bin Considering that the linear attenuation coefficients representing each of _{Bin 1} to Bin ₃ are different for each layer, μ _1a ,…, μ _1n ; μ _2a ,…, μ _2n ; μ _3a ,…, μ _3n (cm ^-1 ) can be written. Each energy region Bin has different linear attenuation coefficients μ _1a ,…, μ _1n ; μ _2a ,…, μ _2n ; μ _3a ,…, μ _3n in a layered manner, and has a layer thickness t _a ,…,t _n (cm ) is expressed as shown in the figure. That is, the incident X-ray dose (number of photons) C _l1 , C _l2 , C _l3 is the linear attenuation coefficient μ _1a , ..., μ _1n ; μ _2a , ..., μ _2n ; μ _3a , ..., μ _3n and the thickness, respectively. After receiving attenuation depending on t _a ,...,t _n , the output X-ray doses (number of photons) C _o1 , C _o2 , C _o3 are C _o1 = C _l1 ×e ^-μ1ata × ... ×e ^-μ1ntn
C _o2 = C _l2 ×e ^-μ2ata × … ×e ^-μ2ntn
C _o3 = C _l3 ×e ^-μ3ata × … ×e ^-μ3ntn
... (3)
It can be expressed as.

Therefore, in the case of a multi-material model consisting of multiple compositions, as shown in FIG. 4(B), for an incident X-ray dose (photon number) C _li , the output X-ray dose (photon number) _Coi is C _oi =C _li ×e ^-μiata × ... ×e ^-μintn
(i=1-3)
… (4)
It can be expressed as.
[Processing procedure]

The material identification and bone mineral quantification processing performed by the data processing device 12 will be described based on the relationship between photon measurement using the material model described above and the linear attenuation value μt. In the data processing device 12, the data processor 35 executes a predetermined program to perform substance identification and bone mineral quantification in accordance with the procedure shown in FIG. Of course, the data processor 35 may be configured to only perform bone mineral determination.
[Preprocessing]

First, the data processor 35 determines whether to perform image acquisition interactively or automatically with the user, for example (step S1), and waits until the timing of image acquisition. If it is determined to acquire an image (step S1, YES), whether the frame data is transferred from the detector unit 26 to the buffer memory 32 and stored, or whether the frame data is automatically transferred to the buffer memory 32 regardless of whether or not it is determined to acquire an image. The frame data that has been transferred and already stored is read into, for example, the RAM 34 (step S2). As schematically shown in FIG. 6, this frame data includes frame data FD ₁ , FD ₂ , FD 3 of counts of X-ray photons having energies belonging to three energy regions Bin ₁ , Bin ₂ , Bin _{3 ,} respectively. , and frame data FD _all of the counts of X-ray photons in the entire energy range Bin _all (Bin ₁ + Bin ₂ + Bin ₃ ).

Next, the data processor 35 determines whether to perform substance identification and/or bone mineral quantification interactively with the user or in response to automatic instructions (step S3). The process waits until such an instruction is received, and ends the process when there is an instruction to end the process (step S4).
[Creating a focused image]

If it is determined in step S3 to perform substance identification and/or bone mineral quantification (step S3, YES), the data processor 35 interacts with the user or automatically, for example, crosses the OB to be examined. A cross section to be used is specified (step S5).

As an example, when specifying the position of the cross section interactively, the user may specify the height Hc from the detector 24 via the input device 37, as shown in FIG. For example, in FIG. 1, the height in the height direction (Y-axis direction) of the inspection target OB (for example, the back of a hand or foot) made of an X-ray transparent material and placed on a bed BD is H _OB or higher. If it is known that OB to be inspected is, it is sufficient to specify a cross section with a height H _C corresponding to approximately the center in the height direction of the OB to be inspected. Of course, in this example, since the detection unit 26 is located below the bed BD, the height Hc is determined by taking into account the height of the gap between the bed BD and the detection surface of the detector 24 of the detection unit 26. is specified. That is, it can be specified as H _C =H _BD +H _OB /2. When the height of the inspection target OB is unknown or varies, the height Hc may be set to the height of the surface of the bed BD. In this case, it can be specified as H _C =H _BD .

On the other hand, if you want to automatically specify the cross section of the OB to be inspected, in step S5, as the cross section specification information, specify that the all-pixel focusing plane for optimal focusing for each pixel is to be set, instead of the height Hc. will be done. In this case, the height of this all-pixel focusing plane is not necessarily constant, and in order to achieve optimal focus for each pixel, it often has unevenness that intersects the OB to be inspected but has a different height for each pixel. . Such a method of creating an all-pixel focusing plane is exemplified in, for example, US Patent No. 8,433,033 and PCT/JP2010/62842. A laminography method (or tomosynthesis method) is used to create these examples.

After specifying the cross section in this manner, the data processor 35 creates a tomographic image of the specified cross section using, for example, a plurality of frame data FD _all for the entire energy region Bin _all (step S6).

When a certain height Hc is specified, a laminography method is used in which multiple frame data FD _all are overlapped and pixels are added while shifting by a shift amount corresponding to the height Hc. You can create it below. As a result, a tomographic image (laminography image) IM _all in which the optimal focusing position is aligned with the designated height Hc is created (see FIG. 6). This tomographic image IM _all is also one of the focused images, although the focused position is limited to the height Hc.

On the other hand, if it is specified that the all-pixel focusing plane of the inspection _{target OB is to be set, all pixels} are A focused image IM _all ′ is created under the laminography method (see FIG. 6). The inspection target OB is reflected in this all-pixel focused image IM _all ', and the all-pixel focused image IM _all ' is a tomographic image that is optimally focused for each pixel in the height direction or X-ray irradiation direction. It is a statue. Examples of this tomographic image include those described in US Pat. No. 8,433,033 and PCT/JP2010/62842, similar to the above-described creation method. The creation method described in this publication is exemplified for dental use, but by converting the curved pseudo three-dimensional image created by this creation method into a two-dimensional image, the image that can be used in this embodiment is A two-dimensional tomographic image is created as a focused image. Since this two-dimensional tomographic image has undergone a process called all-pixel focusing, the degree of focusing is more precise than the above-described focused image IM _all having a constant height. In this embodiment, any focused image may be used, so the following description will be made simply as the focused image IM _all .

Subsequently, the data processor 35 uses the frame data FD ₁ , FD ₂ , FD _{3 collected from the three energy regions Bin 1 , Bin 2 , Bin 3 to obtain the} specified height Hc or, for example, the total pixel total. Tomographic images are sequentially created using the laminography method according to the average height of the focal plane (steps S7, S8, S9). As a result, as schematically shown in FIG. 6, a total of three focused images IM ₁ , IM ₂ , and IM ₃ are created for each energy region. The order in which these three focused images IM ₁ , IM ₂ , and IM ₃ are created is arbitrary. Of course, these three focused images IM ₁ , IM ₂ , IM ₃ may also be created as all-pixel focused images in the same manner as described above.
[Setting region of interest]

Next, the data processor 35 interactively or automatically sets a region of interest ROI on the all-pixel focused image IM _all (step S10). For example, when identifying the type of material forming the inspection object OB, this region of interest ROI is assumed to be composed of the same material in the inspection object OB reflected in the focused image IM _all . A region of interest ROI of appropriate size surrounding the portion (which may vary in thickness) is established. In the case of detecting a foreign object or specifying a lesion, a region of interest ROI of an appropriate size is set to surround a suspected or pathologically suspicious location (see FIG. 6).

Specifically, in the case of bone mineral quantification, minute particles are placed in areas that are presumed to be the same "skin/muscle" and "bone" (for example, the second joint of the index finger) in the direction of the X-ray path. A rectangular region of interest ROI is set (see FIG. 13(A) described later). This region of interest ROI has a size of, for example, about 5 mm x 5 mm, and if the pixel size is 200 μm x 200 μm, the size is equivalent to 5 x 5 = 25 pixels. This region of interest ROI does not necessarily have to be rectangular, and may have an irregular shape.

Once the region of interest ROI is determined on this all-pixel focused image IM _all , the region of interest ROI is similarly set on the three focused images IM ₁ , IM ₂ , IM ₃ for each energy region using this region information. (See Figure 6).
[Background estimation and background deletion]

Next, the data processor 35 estimates a pixel component (background component) serving as a background in the region of interest ROI on the focused image IM _all (step S11). As described above, this background component is determined by what identification information is desired. When identifying or specifying the type or property of a substance, or when quantifying bone mineral density using an X-ray scan as in this embodiment (X-ray spot photography may also be used), background components are often caused by the bed and air. It is a known ingredient. When identifying (estimating) the type of foreign object or the state of a lesion, the component of the OB itself to be inspected is added to these known components as a background component of the foreign object or lesion. If this background component information is known, it is set as a fixed value and subtracted from the region of interest ROI of each of the three focused images IM ₁ , IM ₂ , IM ₃ for each energy region (step S12).

On the other hand, if the amount of the background component is unknown, it is necessary to estimate the background component. This estimation method may be performed using an appropriate method, such as an interpolation method from pixel values at a plurality of mutually spaced arbitrary positions outside the region of interest and including only background components in the X-ray path.

Note that the main purpose of the above-mentioned preprocessing is to set a region of interest ROI in each of the focused images IM ₁ , IM ₂ , and IM ₃ for each of the three energy regions and to remove the background components thereof. Therefore, without creating the all-pixel focused image IM _all in all energy regions, it is possible to replace it with an image sufficient to estimate the background component, that is, any one of the focused images IM ₁ , IM ₂ , and IM ₃ . good. In this case, a database of estimated background components or the like can be used to indirectly estimate background components such as a focused image that is not directly used for background component estimation.
[Main processing of substance identification]

When the preprocessing is completed as described above, the data processor 35 performs main processing for substance identification (step S13). This main process is also used as part of the bone mineral quantification process described later, and is performed as shown in FIG.
<Calculation of linear attenuation value μt>

First, the data processor 35 calculates the linear attenuation value μt using pixel values surrounded by the region of interest ROI and from which background components have been removed in each of the three focused images IM ₁ , IM ₂ , IM ₃ (FIG. 7 , step S131). Here, μ is the linear attenuation coefficient (also simply called attenuation coefficient) of the substance, and t is the thickness of the substance along the X-ray irradiation direction.

（ｉ＝１～３、ｊ＝ａ～ｎ：複数物質モデルの場合）
… （６）
ここで、ｌｎは自然対数を採ることを意味している。 In the single material model and multiple material model described above, for each energy region Bin _i (i=1 to 3), the linear attenuation value μt for each pixel is calculated as follows by modifying equations (2) and (4). Calculated from expressions.
μ _i t=lnC _li −lnC _oi
(i=1 to 3: in case of single substance model)
... (5)

(i = 1 to 3, j = a to n: in case of multiple substance model)
... (6)
Here, ln means taking a natural logarithm.

Therefore, if the number of photons that entered (input) a substance and the number of photons that exited (output) from the substance are known, the linear attenuation value μt can be calculated. The number of outgoing photons _Coi is the number of photons detected by the detector 24 for each energy region and for each pixel. C _li is the number of photons of X-rays incident under the same conditions as actual X-ray inspection, and is, for example, a known value set in advance. Of course, it may be a value estimated at the time of substance identification in consideration of variations in actual X-ray inspection conditions each time.

Note that since the linear attenuation value μt is usually affected by the beam hardening phenomenon, it is desirable to perform the vector calculation described below after performing the correction process. Beam hardening is a phenomenon in which when continuous X-rays pass through a material, more low-energy photons are absorbed than high-energy photons, resulting in the average (effective) energy shifting to the higher energy side. be. When this beam hardening occurs, artifacts occur or the pixel values of the reconstructed image become inaccurate. Beam hardening occurs to varying degrees and depends on the thickness of the material (the thicker the material, the greater the beam hardening). For this reason, it is desirable to correct the linear attenuation value μt for each energy region and for each pixel, based on the correction processing method described in, for example, International Publication No. WO 2017/069286 A1, which has already been filed by the present applicant.

In particular, as a method for correcting the beam hardening, the applicant may already adopt the method described in International Publication No. WO 2019/083014 A1. The purpose of this method is to perform more accurate beam hardening correction on a wider range of targets with elements with effective atomic numbers Z _eff with a lower computational load, contributing to the presentation of more quantitative X-ray images. Proposed.

This correction method has been proposed in various ways, but according to one example, a desired effective atomic number (for example, Zm=7) is specified from the effective atomic number in a desired range (Zmin to Zmax), and this specified A straight line on the two-dimensional coordinates is set as a target function when it is assumed that monochromatic X-rays are irradiated onto a substance consisting of an element having an effective atomic number (for example, Zm=7). Furthermore, on the two-dimensional coordinates, multiply the slope (μ/ρ) of the objective function in the horizontal axis direction to obtain multiple curves for each of multiple effective atomic numbers (for example, Z = 5 to 14) of the effective atomic number. Generalize as a variable. Furthermore, a curve for an element with a designated effective atomic number (for example, Zm = 7) is designated from among the plurality of generalized curves, and the beam is determined based on the residual difference between this designated curve and other curves. A beam hardening correction function for correcting hardening is stored in advance in the storage unit as the correction information. Therefore, if the information regarding the generalized objective function and the residual of the effective atomic number specified in the predetermined effective atomic number range is held, the beam hardening correction function can be calculated using the above-described procedure. Therefore, even if the preset range of effective atomic numbers is wider, the amount of calculation does not need to be proportional to the range when calculating the beam hardening correction function. In other words, beam hardening can be corrected for objects having elements with a wider range of effective atomic numbers Z _eff with less computational load.

In addition, in the case of medical examinations, the soft tissue of the breast and limbs can be considered to be composed of simpler substances, and instruments are used to compress or fix the area to be imaged. Even in this case, since the flat plate structure and the material are known, the accuracy of background removal is high, and this linear attenuation value μt can be calculated with higher accuracy. Furthermore, in non-destructive testing of foods and the like, if the background component can be appropriately estimated as described above, the linear attenuation value μt can be calculated with high accuracy from pixel information after the background component has been removed.

Next, the data processor 35 picks up the linear attenuation value μt of each pixel forming the region of interest ROI in the focused images IM ₁ to IM ₃ of the three energy regions Bin ₁ to Bin ₃ described above, and vectorizes it. (Step S132: See FIG. 8).

That is, a three-dimensional line attenuation vector (μ ₁ t, μ ₂ t, μ ₃ t) is created for each pixel (see FIG. 8). This three-dimensional line attenuation vector (μ ₁ t, μ ₂ t, μ ₃ t) still includes the factors of thickness t and density, so this vector itself is It only indicates the amount of attenuation and cannot be used as a substance-specific index. This is because, as in the case of an X-ray scanogram or plain X-ray photography, since the thickness t is unknown, the material-specific linear attenuation coefficients μ ₁ , μ ₂ , μ ₃ for X-rays cannot be determined. Furthermore, in the application of examinations and diagnosis using X-rays, the moving speed of the belt conveyor used for the examination is fast and it can quickly pass the X-ray irradiation area, which also reduces patient exposure for diagnosis. For this purpose, the X-ray dose is limited. In such cases, the number of photon counts collected at each pixel for inspection or diagnosis will be small, so one three-dimensional linear attenuation vector (μ ₁ t, μ ₂ t, μ ₃ t) alone is insufficient. , it is difficult to obtain material-specific information because it is buried in other noise components.

Therefore, material identification is performed by normalizing this three-dimensional linear attenuation vector (μ ₁ t, μ ₂ t, μ ₃ t) and treating it as a set.

First, each three-dimensional line attenuation vector (μ ₁ t, μ ₂ t, μ ₃ t) is normalized (or normalized) to a unit length (length 1) by the following equation (7), and the thickness t and a three-dimensional mass attenuation vector (μ ₁ ^′ , μ ₂ ^′ , μ ₃ ^′ ) that does not include the material density factor (step S133).
(μ ₁ ^′ , μ ₂ ^′ , μ ₃ ^′ )
= (μ ₁ t, μ ₂ t, μ ₃ t) / ((μ ₁ t) ² + (μ ₂ t) ² + (μ ₃ t) ² ) ^1/2
=(μ ₁ , μ ₂ , μ ₃ )/(μ ₁ ² +μ ₂ ² +μ ₃ ² ) ^1/2 …(7)

Of course, this normalization is to make the lengths of each three-dimensional mass attenuation vector (μ ₁ ^′ , μ ₂ ^′ , μ ₃ ^′ ) the same, and the length does not necessarily have to be 1. It can be any length you like.

By normalizing in this way, the thickness t and density factors disappear, so on the three-dimensional coordinates with the linear attenuation coefficients μ ₁ , μ ₂ , μ ₃ as mutually orthogonal axes, each three-dimensional mass is placed at the coordinate origin. Once the starting point of the attenuation vector (μ ₁ ^′ , μ ₂ ^′ , μ ₃ ^′ ) is placed (step S134), the position coordinates of the end point can be determined based on material-specific information (type of material, properties, etc.) that causes a change in μ ^′ . information).

In this way, in this embodiment, the vector amount indicating X-ray attenuation is treated as a three-dimensional line attenuation vector (μ ₁ t, μ ₂ t, μ ₃ t) before normalization, and is treated as a three-dimensional mass after normalization. They are treated as attenuation vectors (μ ₁ ^′ , μ ₂ ^′ , μ ₃ ^′ ). In this embodiment, all processing is performed in three dimensions, but the same applies even in two dimensions.

_The ^process in ^{step S134} , for example, as _shown in FIG _. For example, if the length of the mass attenuation vector is 1, a partial sphere passing through the length of the three orthogonal axes = 1 is set in the memory space (Step S134-2). ), the tips of the three-dimensional mass attenuation vectors (μ ₁ ^′ , μ ₂ ^′ , μ ₃ ^′ ) are placed (also called dots or mapping) on this partial sphere from the origin O (step S134-3).

Note that the three-dimensional slope information of the three-dimensional mass attenuation vector (μ ₁ ^′ , μ ₂ ^′ , μ ₃ ^′ ) for each pixel changes depending on the type and property of the material in the three-dimensional space set in step S134-1. It can also be said to be scatter data that pseudo (virtually) represents information specific to a substance. Therefore, the position pointed to by the tip of the three-dimensional mass attenuation vector (μ ₁ ^′ , μ ₂ ^′ , μ ₃ ^′ ), that is, the position indicated by the tip of the three-dimensional mass attenuation vector (μ ₁ ^′ , μ ₂ ^′ , μ ₃ ^′ ) A set of dimensional slope information (that is, scatter points) is also called a "three-dimensional scatter diagram." In other words, if the substance changes, the slope of the three-dimensional mass attenuation vector (μ ₁ ^′ , μ ₂ ^′ , μ ₃ ^′ ) will change, and the three-dimensional position pointed to by its tip (the position of the dispersion point) will also change. The three-dimensional position information reflects the energy distribution of X-ray photons before and after passing through the inspection target OB.

Furthermore, the data processor 35 calculates the length of each three-dimensional line attenuation vector (μ ₁ t, μ ₂ t, μ ₃ t) for each pixel by ((μ ₁ t) ² + (μ ₂ t) ² +(μ ₃ t) ² ) ^1/2 …(8)

Calculate as The amount expressed by this formula (8) is the X-ray absorption amount (or X-ray attenuation amount; in the X-ray imaging area, X-ray absorption has the largest effect on attenuation, so it will be referred to as "absorption" from now on. ), which is also useful as complementary information for substance identification, and also forms pixel values as a substitute image for conventional absorption images. Therefore, an image is created whose pixel values are values obtained by gradating this absorption amount (step S135). The length of this three-dimensional line attenuation vector is called the "absorption vector length" which pseudo (virtually) corresponds to the X-ray attenuation value, and an image with this as a pixel value is an "absorption vector length image (or pseudo It is also called ``absorption image''. This absorption vector length image is a stable image because it does not easily depend on the shape of the incident energy spectrum of X-rays, and comprehensively reflects each line attenuation value μt. As a result, this absorption vector length image becomes an image with strong contrast. This absorption vector length image may be stored in the image memory 36 and displayed on the display 38 when necessary. In particular, it is possible to obtain a characteristic image for a large mass substance with strong X-ray beam hardening.

Finally, the data processor 35 stores the data of the three-dimensional scatter diagram described above as material-specific information and the absorption vector length image as complementary information for material identification in the image memory 36 (FIG. 5, step S14), and stores it as necessary. Accordingly, they are presented to the user via the display 38, for example (step S15).

As described above, based on the X-ray photon counts for each energy region collected by the photon counting type detector 24, material-specific information can be acquired regardless of the thickness of the OB to be inspected. This exhibits great advantages when combined with the display and analysis of substance-specific information described below.
[Example of display and analysis of substance-specific information]

This process of displaying and analyzing the substance-specific information is executed, for example, as part of step S15 described above. The data processor 35 displays the above-mentioned substance-specific information, for example, in response to an instruction from a user. Specifically, a spherical surface with a radius of 1 centered at the origin is set in a three-dimensional coordinate space whose three axes are the elements μ _{1 ′} ^, μ ₂ ^′ , and μ ₃ ^′ of the 3-dimensional mass attenuation vector (Fig. 9. Step S31).

Next, a three-dimensional scatter diagram consisting of the three-dimensional mass attenuation vectors (μ ₁ ^′ , μ ₂ ^′ , μ ₃ ^′ ) of each pixel is plotted with the origin of the 3-dimensional coordinate space as the starting point, and the ending point, as an example, on the same plane. It is mapped (dots) and displayed on the spherical surface (spherical surface normalized to radius = 1) that functions as (step S32). This mapped set of end points on the spherical surface is a set of material-specific scattered points based on material-specific information. Therefore, even if the object to be inspected is a material whose thickness t differs between pixels, a set of scattered points that does not depend on the factor of the thickness t can be obtained.

FIG. 11 schematically shows an example in which the set of scatter points is plotted on a part of the normalized spherical surface (a part of the same surface) as a three-dimensional scatter diagram.

Next, as shown in FIG. 12A, the data processor 35 groups the scattered points derived for all or part of each pixel forming the region of interest ROI (see dotted line box: step S33). , as shown in FIG. 3B, the center of gravity GR of the grouped scattering points is calculated (step S34). Next, as shown in FIG. 3C, a vector V _obj connecting the gravity center position GR of each scattered point group and the origin is calculated (step S35).

Regarding grouping, the range can be changed as appropriate depending on the analysis content. In other words, all the scattered points derived for all pixels forming the region of interest ROI may be grouped together, or once all the scattered points are grouped, statistically irregular (largely spaced from the center of gravity) scattered points may be grouped. It is also possible to remove points and then regroup them. Alternatively, if the range of the ROI is not appropriate and multiple substances are spread in the in-plane direction of the pixel, the scattering points will naturally spread out or be separated, so it is possible to group them using nearby scattering points. Grouping may be performed by resetting the ROI, or by having the user or automatic determination software designate a grouping target range on the scattering points.

Next, this vector V _obj is compared with previously held reference data to identify or specify the type and properties of the substance (step S36). The reference data includes, for example, a storage table in which the three-dimensional slope of the vector V _obj measured in advance according to the type and properties of the substance is stored together with the allowable width. Therefore, material identification can be performed based on whether the slope of the calculated vector V _obj falls within the permissible range, and vector information that becomes noise can be eliminated. The identified information is saved (step S37).

This vector V _obj corresponds to an average vector obtained by averaging the directions of the three-dimensional line attenuation vectors (μ ₁ t, μ ₂ t, μ ₃ t) of each of the plurality of pixels forming the region of interest ROI (however, its length is normalized).

In addition, in step S15 described above, the three-dimensional scatter diagram and the absorption vector length image can be presented and provided in various ways. For example, the data processor 35 may divide and display the three-dimensional scatter diagram and the absorption vector length image on the display 38, or first display the three-dimensional scatter diagram and supplementally display the absorption vector length image in response to a user's request. A long image may be displayed, or vice versa. Further, in response to a request from the user, the range of grouping of scattering points or the range of ROI may be re-specified on the once displayed screen.
<Bone mineral determination>

For example, when performing bone mineral quantification on the back of the hand OB (finger) of a subject, the process shown in FIG. 14 is executed by the data processor 35.

If YES in step S51 of FIG. 14, that is, bone mineral quantification is to be performed, the finger FG of the hand (for example, the area between the second and third joints) has already been determined in step S10 described above. A region of interest ROI is set in the tomographic image IM _ALL of a specified cross section of a location matching the multi-substance model of "muscle/skin" (see FIGS. 13A and 13B). Here, the three-dimensional line attenuation vectors (μ ₁ t, μ ₂ t, μ ₃ t) at each pixel PX of this region of interest ROI have already been calculated in step 132 described above (see FIG. 13(C)). . This three-dimensional line attenuation vector (μ ₁ t, μ ₂ t, μ ₃ t) is an n-dimensional (here, three-dimensional ) corresponds to the space vector of

Therefore, when performing bone mineral quantification, the data processor 35 calculates the three-dimensional linear attenuation vector (μ ₁ t, μ ₂ t, μ ₃ t) of each pixel PX of the region of interest ROI by the calculations in steps S31 to S35 described above. Information indicating the vector V _obj (however, the length has been normalized), which is the average vector of each scattering point group obtained using (See (D)). In the case of this bone mineral quantification, as shown in the X-ray path IBn (Figure 15), the X-rays pass through "skin/muscle part B, bone A part, muscle/skin part B" in this order. The vector V _obj forms one scattered point group determined by the composite characteristics of all substances existing on the X-ray path IBn (however, it has a spread due to noise). The direction of a representative vector (three-dimensional mass attenuation vector) representing the region of interest ROI can be determined from the vector V _obj .

Furthermore, the data processor 35 reads out data indicating the absorption vector length forming the absorption vector length image, which has already been calculated for each pixel PX of the region of interest ROI in step S135, from the image memory 36 into its own work area (step S53). ). In the case of bone mineral quantification, this absorption vector length corresponds to the length of the three-dimensional line attenuation vector according to the definition mentioned above, and is equivalent to the length of the three-dimensional linear attenuation vector in the case of bone mineral quantification. This approximately corresponds to a pseudo (virtual) X-ray attenuation value.

Therefore, the data processor 35 calculates the average value of the absorption vector lengths for each of the read pixels PX (step S54). As a result, the average length of the three-dimensional linear attenuation vectors of the plurality of pixels PX forming the region of interest ROI can be found.

Therefore, the data processor 35 converts the three-dimensional vector having the direction of the vector obtained in step S52 and the average value of the absorption vector length obtained in step S54 into a three-dimensional representative vector (three-dimensional vector) representing the region of interest ROI. The average of the line attenuation vectors (hereinafter referred to as three-dimensional representative vector) is calculated as V _obj-d (step S55: see FIG. 13(E)).

Note that as a modification example of calculating the three-dimensional representative vector (average of three-dimensional line attenuation vectors) V _obj-d that can be implemented in this step S55, a three-dimensional line attenuation vector (μ ₁ t, μ ₂ t, μ ₃ It is also possible to use a vector averaging method in which t) is averaged for each element of the three-dimensional line attenuation vector itself for a plurality of pixels forming the grouped region described above.

As can be seen from FIG. 15, this three-dimensional representative vector V _obj-d is the total value when passing through the soft tissue of "skin/muscle B", the hard tissue of "bone A", and the "skin/muscle B" in order. Contains X-ray attenuation information. On the other hand, the information desired in bone mineral quantification is information on the bone itself (bone density, bone mass), which is surrounded by skin and muscles. From this point of view, this three-dimensional representative vector V _obj-d includes redundant X-ray attenuation information due to skin and muscles.

Therefore, the data processor 35 reads out the three-dimensional reference vector V _ref , which is set and stored in advance in the ROM 33, for example (see FIG. 13(F): step S56). This three-dimensional reference vector V _ref is a three-dimensional reference vector equivalent to the line attenuation value of the skin and muscle, assuming that X-rays have passed through only the skin and muscle (see path IP indicated by the virtual line in FIG. 15). This is the reference vector for , and has been estimated and measured in advance. In addition, instead of this prior estimation and measurement, when calculating the three-dimensional representative vector V _obj-d, at the same time, ROI is placed on the part where the X-rays are thought to have passed only through "skin/muscle B": It is also possible to set the ROI _ref (see FIG. 13(A)) and calculate the three-dimensional reference vector V _ref in the same manner as the calculation of the three-dimensional representative vector described above.

Specifically, this three-dimensional reference vector V _ref can be either i) estimated from the external size or weight of the target area to be irradiated with X-rays, or ii) statistically collected in advance and stored in a database. iii) A three-dimensional reference vector V that is considered to be equivalent to a linear attenuation vector and is previously obtained and stored in a partial region that is only skin and muscle within the target imaging region. _ref (three dimensions have the dimensions μ ₁ t, μ ₂ t, μ ₃ t), or, as mentioned above, iv) partial region (only skin/muscle) of the target imaging region ( For example, it is set by calculating a three-dimensional reference vector V _ref in the region indicated by ROI _ref (see FIGS. 13A and 15), assuming that it is equivalent to a linear attenuation vector. Among these, When performing the calculation shown in item iv, it is necessary to read and write image data, and the data processor 35 performs the calculation using not only the data read from the ROM 33 but also the RAM 34.

Note that the size of this three-dimensional reference vector V _ref is usually considerably smaller than that of the three-dimensional representative vector V _obj-d , but it is an amount that must be definitely excluded in order to perform accurate bone mineral quantification. It is.

Therefore, the data processor 35 subtracts the reference vector V _ref on the three-dimensional coordinates from the representative vector V _obj-d representing the already determined region of interest ROI (step S57). Through this vector subtraction, a three-dimensional vector V _obj-d ′ that substantially reflects the X-ray attenuation information of only the target finger bone A is obtained. This three-dimensional vector V _obj-d ′ is called a target vector. Therefore, this objective vector V _obj-d ′ represents the entire region of interest ROI set on the tomographic image of the specified cross section of the finger FG, has the average linear attenuation value as the vector length, and has the mass attenuation vector It has a three-dimensional vector direction that reflects information. Therefore, this objective vector V _obj-d ′ is information that reflects the state of bone mass and bone quality when bone mineral quantification is performed through the finger bone. Specifically, according to the simulation conducted by the inventors, the length of the objective vector V _obj-d ′ strongly reflects the bone mass (bone mineral content) only in the bone region, and the vector direction It is presumed that this strongly reflects the bone quality of only the site.

Next, the data processor 35 performs processing to visualize the vector length and vector direction of the above-mentioned target vector V _obj-d ′, and displays and saves the data, or presents only the data of the bone mineral quantification results. (Step S58). This visualization provides bone mineral quantification information that is easy to see for the interpreting doctor and the patient, and typical examples include color imaging, its display, and digitization.

Note that, particularly in foreign matter detection, by displaying the image in color, it is possible to display it as an effective atomic number image. For example, when a foreign substance is mixed into a product that is relatively homogeneous (e.g., the same substance, such as chocolate, cereal, or milk powder), the substance other than the substance of the product may be displayed in color. Foreign matter detection becomes possible.
<Effect>

As described above, according to the X-ray inspection system according to the present embodiment, the three-dimensional line attenuation vector (μ ₁ t , μ ₂ t, μ ₃ t), it is possible to more accurately quantify bone information (bone amount (bone mineral content), bone quality) for bone diagnosis, which is also one aspect of substance property exploration.

According to the present embodiment, an object OB (for example, a patient's Processing based on count values in each of three X-ray energy regions Bin ₁ to Bin ₃ , in which the X-rays irradiated to the back of the hand or foot and transmitted through the object are detected by the detector 24 having a plurality of pixels. will be held.

Specifically, an X-ray image of the target OB is created based on the counted value and displayed on the display 38. On the X-ray image displayed on the display 38, a ROI (region of interest) is set at the same bone portion in the direction of the X-ray path.

Furthermore, a three-dimensional space vector (three-dimensional line attenuation vector) of each of the plurality of pixels corresponds to a line attenuation value when the X-ray passes through the object in each of three energy regions, and a photon Calculated based on count data. Furthermore, a three-dimensional representative vector V _obj-d representing the ROI is calculated by averaging the directions and magnitudes of the spatial vectors of each of a plurality of pixels. Assuming that X-rays have passed through one of the two materials A and B, the three-dimensional reference vector Vref corresponding to the linear attenuation value of material B, which is equivalent to the linear attenuation value, is the object of interest. It is subtracted from the representative vector V _obj-d of the region ROI to obtain an objective vector V _obj-d ′ which is equivalent only to the substance A corrected by the subtraction. At this time, the reference vector V _ref is theoretically or It is set (estimated/evaluated) in advance through experiments, etc., and is stored in a readable manner.

The reference vector V _ref is determined by setting a portion consisting only of the substance B in the X-ray image of the target OB displayed on the display 38 as the region of interest ROI _ref for determining the reference vector. After calculating the three-dimensional line attenuation vector of the part, the correlation information between the thickness of material B in the ROI part and the ROI _ref part (may be actually measured, or may be statistically determined from a prior experiment and retained) From this, the reference vector (the three-dimensional line attenuation vector V _ref of only the substance B in the ROI part) may be estimated or calculated by adjusting the size of the three-dimensional line attenuation vector in the ROI _ref portion. In particular, if we know the thickness of the target OB in the part where the three-dimensional representative vector V _obj-d was actually calculated and the thickness of the target OB in the part where the reference vector V _ref was calculated, we can calculate the target vector V _obj-d '. Can be calculated analytically.

The above-mentioned objective vector V _obj-d ' reflects the degree of attenuation when continuous X-ray photons with a continuous energy distribution from low energy to high energy pass through the tissue of the bone part. It is possible to collect numerical values that more accurately represent the density and condition (characteristics) of bone quality. Moreover, even if substances A and B are included in the ROI portion, a target vector that reflects the linear attenuation of only the target substance A can be extracted with higher precision by a simple operation called vector subtraction.

In other words, in the conventional case, even if the types of substances A and B are known and their total thickness is constant, the thickness of each in the X-ray path direction is unknown and two different substances exist. Therefore, it has been difficult to obtain vector information based on the X-ray attenuation of only the target substance A with simple calculations and with high accuracy.

However, according to the present embodiment, a target vector that reflects the linear attenuation of only the target substance A can be easily and accurately calculated by vector calculation for each region of interest.

Furthermore, according to the present embodiment, when bone diagnosis (including bone mineral quantification) is performed, a target vector indicating at least the properties of substance A can be obtained for each region of interest, so that the length and direction of the vector can be determined based on bone mass and It is possible to provide more multifaceted information regarding bone quality. Unlike conventional processing that provides information based only on bone density, it is possible to enrich the property information provided and meet the demands for diagnosis and treatment of osteoporosis, for example.

Furthermore, information on three-dimensional reference vectors can be stored in advance using a relatively simple method. In other words, as mentioned above, i) estimated from the external size including the thickness of the part of the target to be irradiated with X-rays, or weight, or ii) read from a reference table that has been statistically collected in advance and compiled into a database. iii) Setting by calling a reference vector that is considered to be equivalent to the line attenuation vector and has been obtained and stored in advance in a partial region where only the substance B is present in the imaged region of the target. That's fine. In this case, the computation required for vector subtraction (that is, the computation for determining the reference vector) can be further simplified.

Further, direction information about the direction of the three-dimensional reference vector itself obtained empirically may be stored in advance and called up when necessary. This greatly simplifies the amount of calculation for the reference vector.

Furthermore, if the thickness of the OB to be inspected is appropriately measured and used for estimating and calculating the three-dimensional line attenuation vector (representative vector) V _obj-d and the three-dimensional reference vector V _ref , depending on the application situation, Although the amount of calculation increases, on the other hand, quantitative accuracy can be improved.
<Other effects>

In addition to the effects of bone mineral quantification described above, the X-ray inspection system according to this embodiment provides various effects.

First, a region of interest is set in a focused tomographic image (image) of a cross section (or uneven cross section) of the OB to be inspected, and from that image, it is determined whether there is a substance of interest present in the region of interest (such as an object to be inspected or a foreign object). pixel information (background component) that forms the background of objects) is removed. Based on the data of the tomographic image after this removal and the count value for each X-ray energy region and each pixel in the region of interest, the unique transmission characteristics (for example, linear attenuation coefficient μ) of the substance of interest for X-rays are determined for each pixel. It is calculated as information. Since this unique information does not depend on the thickness t of the material, the type and properties of the material of interest can be identified or specified based on this information. For example, by comparing the calculated unique information with known unique information held in advance (known unique vector information (information with a certain tolerance range) unique to the substance), material identification becomes possible. .

Further, depending on the setting of the region of interest, it is possible to adjust the range in which substances are to be identified, including the whole or part of the object to be inspected. At this time, this unique information is obtained as information specific to the type and properties (state) of the material that are not affected by the thickness t, so as long as the region of interest is selected where the type and properties of the material do not change, the thickness It is sufficient to set the width to an appropriate size regardless of the change in width. Unlike conventional substance identification, unique information is obtained after removing background components that are unnecessary information for substance identification, so substance identification can be performed with higher accuracy and its reliability is also improved.

More specifically, a three-dimensional scatter diagram can be presented on a spherical surface by normalizing a vector made up of linear attenuation coefficients μ for each energy region. This scattering point (spectrum) represents three-dimensional slope information of this vector (ie, material specific information). Therefore, just by looking at the state of this scattering point, you can determine whether the OB to be inspected is made of metal or something else, and whether there is something else (such as a foreign object) in the OB to be inspected. By determining the center of gravity of the dispersion points and adding information such as the state of the body (such as the ratio of muscle to fat), it is easy to understand both visually and quantitatively.

Further, in the process of obtaining a three-dimensional scatter diagram, data of an absorption vector length image is also obtained. The present inventors demonstrated that this absorption vector length image is less dependent on the shape of the energy spectrum of the irradiated X-rays than conventional X-ray absorption images. I am checking using. The spectrum shape is, for example, as illustrated in FIG. 2, a spectrum shape in which the counting frequency of X-ray photons in the middle energy region Bin ₂ is higher than those on both sides thereof. In the case of this embodiment, since the processing of equation (8) according to the embodiment described above is performed, differences in X-ray absorption depending on the spectrum shape are unlikely to occur, and the coefficients on the low energy side where the linear attenuation coefficient is largest are This is thought to be because the influence of frequency on images becomes stable.

Therefore, this absorption vector long image is more robust against X-ray irradiation conditions such as X-ray tube voltage because the energy spectrum has less shape dependence, has good image contrast, and is proportional to the linear attenuation value μt. Then, the linear attenuation values μt of all energy bands are averaged (noise is superimposed on the linear attenuation values of each pixel and each energy band, reflecting the quantization noise present in the count value of each energy band). The effect is to reduce noise.
<Modified example>
Furthermore, another modification of the method of actually measuring the thickness of the target OB to improve the evaluation accuracy of bone diagnosis of substance B will be described.

As schematically shown in FIG. 15, among the target OB, t is the actual measured thickness of multiple substances in the region of interest ROI for which the linear attenuation vector is calculated, and t is the thickness of the material A, which is the target for which the target vector is calculated. ₁ and the thickness of material B corresponding to the reference vector is t ₂ (=t ₂₁ +t ₂₂ ). Then, each three-dimensional line attenuation vector can be expressed as follows.
3D line attenuation vector of ROI:
μt=(μ ₁ t, μ ₂ t, μ ₃ t) ‥‥(A.1)
3D reference vector:
V _ref = μ _ref t ₂ = (μ _{1 ref} t ₂ , μ _{2 ref} t ₂ , μ _{3 ref} t ₂ ) ‥‥(A.2)
3D objective vector:

V _obj-d '=μ _obj-d 't ₁ = (μ _1obj-d 't ₁ , μ _2obj-d 't ₁ , μ _3obj-d 't ₁ ) ‥(A.3)

Here, by definition, V _obj-d '=μt- V _ref , so using equations (A.1) to (A.3), each element of the vector can be expressed as follows. can.
μ _1obj-d 't ₁ = μ ₁ t - μ _1ref t ₂ ‥‥(A.4)
μ _2obj-d 't ₁ = μ ₂ t - μ _2ref t ₂ ‥‥(A.5)
μ _3obj-d 't ₁ = μ ₃ t - μ _3ref t ₂ ‥‥(A.6)
Furthermore, due to the assumptions of the multi-substance model, the following relationship exists.
t = t ₁ + t ₂ ‥‥(A.7)

Here, when determining the linear attenuation coefficient of the three-dimensional reference vector, the relationship between the thickness _t3 of material B in the region of interest ROI _ref and the thickness _t2 of material B in the region of interest ROI is clear. In this case, since t ₃ is actually measured, t ₂ is known. Furthermore, since μ _1ref , μ _2ref , and μ _3ref are known from the calculation of the region of interest ROI _ref , in that case, independent equations (A.4) to (A.7) exist, and the unknown There are four variables: μ _1obj-d ′, μ _2obj-d ′, μ _3obj-d ′, and t ₁ , which can be solved a priori.

Therefore, for example, when photographing a person's hand (ROI part is a finger), the ROI and ROI _ref should be set so that the same part is photographed even if the patient changes, data is accumulated, and the ROI is It is desirable to statistically process the relationship between the thickness t ₃ of the material B in the ROI _ref portion and the thickness t ₂ of the material B in the region of interest ROI so that it can be estimated accurately. More directly, the direction of X-ray photography may be changed to directly measure the thickness t ₁ of the finger bone (substance A). In this case, there are four unknown variables that can be solved a priori without estimating _t2 .

The present invention may employ, for example, only a configuration for performing bone mineral quantification. That is, the X-ray inspection system of the embodiment described above is provided as a system specialized for bone mineral quantification, and the data processing device 12 is provided as a data processing device specialized for bone mineral quantification.

Although various aspects of the data processing device, data processing method, and X-ray inspection device according to the present invention have been described above, the present invention is of course not limited to the above-mentioned examples, and the gist of the claims The invention can be further modified in various ways without departing from the above.

10 X-ray inspection system (X-ray inspection device equipped with a data processing device: X-ray inspection device that implements a data processing method)
12 Computer system (data processing device)
21 X-ray tube 24 Detector 25 Data acquisition circuit 26 Detection unit 12 Data processing device 32 Buffer memory (storage means)
33 ROM
34 RAM
35 Data processor (CPU)
36 Image memory (storage means)
37 Input device 38 Display device OB Inspection target (object)

Claims (11)

Continuous X-rays containing n (n is a positive integer of 2 or more) different energy regions are irradiated onto an object consisting of substantially two types of materials A and B in terms of the transmission characteristics of the X-rays, and the object is In a data processing device that performs processing based on the photon count data when the X-ray that has passed through is detected as photon count data by an X-ray detector having a plurality of pixels,
X-ray image display means for creating an X-ray image of the object based on the photon count data detected by the X-ray detector and displaying it on a monitor;
ROI setting means for setting an ROI (region of interest) on the X-ray image displayed on the monitor in a portion where the substances A and B are estimated to exist in the direction of the X-ray path; ,
The photon count data corresponds to the linear attenuation value when the X-ray passes through the object in each of the n energy regions, and the n-dimensional space vector of each of the plurality of pixels is pixel-by-pixel vector calculation means that calculates based on
representative vector calculating means for calculating a representative vector representative of the ROI by averaging the direction and magnitude of the spatial vector of each of the plurality of pixels;
When it is assumed that the X-rays have passed through one of the two types of substances A and B, the n dimensions correspond to the linear attenuation value of the substance B that is equivalent to the linear attenuation value. Reference vector holding means for estimating or assuming and holding a reference vector of
an objective vector that subtracts the n -dimensional reference vector from the n - dimensional representative vector corresponding to the linear attenuation value equivalent only to the substance A, and obtains an objective vector corrected by the subtraction; acquisition means,
A data processing device comprising: 2. The data processing device according to claim 1, further comprising reference vector calculation means for estimating or assuming the n - dimensional reference vector. The representative vector calculation means is
Direction calculation means for calculating a distribution in which the size of the spatial vector of each of the plurality of pixels is normalized, and calculating the direction of the representative vector from the position of the center of gravity of the distribution;
size setting means for setting the size of the representative vector by averaging vector length pixel values or average vector length pixel values that are the sizes of the spatial vectors of each of the plurality of pixels;
3. The data processing device according to claim 1, further comprising: image data creation means for creating image data based on the corrected target vector information;
4. The data processing apparatus according to claim 1, further comprising image display means for displaying the image data created by the image data creation means as an image. The object is a hand or foot of a person or an animal,
The substance A is tissue of the bony part of the finger or toe,
The substance B is muscle and skin tissue of the hand or fingers,
5. The image data creation means is configured to create image data representing information on bone density of the bone portion from information on the target vector of the bone portion. Data processing equipment. comprising reference vector calculation means for estimating or assuming the n-dimensional reference vector,
The reference vector calculation means estimates the reference vector from i) the external size or weight including the thickness of the X-ray irradiation site of the object, and ii) a reference table that is statistically collected in advance and compiled into a database. iii) calling a reference vector that is determined and stored in advance and is considered to be equivalent to the linear attenuation vector in a partial area where only the substance B is present in the imaging region of the target; or iv 6. The method according to claim 2, wherein the calculation is performed by assuming that a partial region containing only the substance B in the imaging region of interest is equivalent to a linear attenuation vector. Data processing equipment. 7. The data processing device according to claim 1, wherein the n dimensions are three dimensions. An X-ray apparatus, characterized in that the data processing apparatus according to any one of claims 1 to 7 is incorporated integrally or so as to be able to cooperate through communication. Continuous X-rays containing n (n is a positive integer of 2 or more) different energy regions are irradiated onto an object consisting of substantially two types of substances A and B, which are known regarding the transmission characteristics of the X-rays, In a data processing method, when the X-rays that have passed through the object are detected as detection data by an X-ray detector having a plurality of pixels, processing is performed based on the detection data,
creating an X-ray image of the object based on photon count data detected by the X-ray detector and displaying it on a monitor;
Setting an ROI (region of interest) on the X-ray image displayed on the monitor at a portion where the substances A and B are estimated to exist in the direction of the X-ray path;
The photon count data corresponds to the linear attenuation value when the X-ray passes through the object in each of the n energy regions, and the n-dimensional space vector of each of the plurality of pixels is Calculate based on
calculating a representative vector representative of the ROI by averaging the direction and magnitude of the spatial vector of each of the plurality of pixels;
Assuming that the X-rays have passed through one of the two types of substances A and B, the value of the substance B equivalent to the linear attenuation value estimated or evaluated in advance and held The n - dimensional reference vector corresponding to the linear attenuation value is subtracted from the n - dimensional representative vector corresponding to the linear attenuation value equivalent only to the substance A, and the target vector is corrected by the subtraction. is obtained,
A data processing method characterized by: The reference vector is the n - dimensional vector information prepared in advance, and
i) Estimated from the external size or weight including the thickness of the X-ray irradiated area of the target, or ii) Read from a reference table that has been statistically collected in advance and compiled into a database, or iii) The imaged area of the target or iv) reading from information that is determined and stored in advance and is considered to be equivalent to a linear attenuation vector in a partial region where only the substance B exists, or iv) the substance B in the target imaging region. 10. The data processing method according to claim 9, wherein the data processing method is performed by assuming that a partial region where only a linear attenuation vector is equal to a linear attenuation vector. 11. The data processing method according to claim 9, wherein the n dimensions are three dimensions.

Images