JPH0618570B2 - Two-dimensional tomographic imaging device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Use of the Invention] The present invention is a two-dimensional method for back-projecting data collected by a computed tomography (CT) method using radiation or the like as a radiation source. The present invention relates to a two-dimensional tomographic imaging apparatus capable of obtaining a clear image by reducing noise appearing in a reconstructed image as a projected image.

[Background of the Invention]

Conventional CT devices include, for example, Iinuma et al .; new medical X-ray diagnostic device (“Measurement and control” Vo115, No. 2 (Showa 51, 2
As described in (Mon.), various calculation methods including a superposition integration method are considered as a method for calculating a tomographic image of a subject. In any operation method, the original data or correction data projected is redistributed to the spatial calculation mesh corresponding to the imaging area to reconstruct the image. However, according to this method, CT
In order to project data to the calculation mesh (usually at least 2 sample points around the subject) corresponding to the transmission point of the air layer, which is always necessary when reconstructing the image, at the same ratio as the subject,
Especially when the transmission data is small or when the difference in the linear absorption coefficient between the object and the air layer is large, the influence of measurement noise appears on the object as it is and the image becomes unclear. Therefore, there is a problem that the distribution of the linear absorption coefficient obtained in the CT reconstructed image cannot be clearly determined.

Although there are various CT reconstruction calculation methods as described above, the characteristic here is that the calculation can be executed every time one projection data is collected, and it is also widely used in the medical CT method. The following is a brief description of the existing superposition integration method.

That is, FIGS. 13A and 13B show a transmission data acquisition method and an image reconstruction method. According to this method, the object 123 is disposed between the radiation source 121 and the detector 122. Thus, transmission data of the object 123 can be obtained from any direction. Image reconstruction is supposed to be performed by redistributing and superimposing these data on a spatial mesh. Thereby, the distribution of the linear absorption coefficient inside the object 123 can be obtained. If this is expressed mathematically, if the transmission data imaged at a certain rotation angle θ in FIG. 5A is P (X, θ), this is given by the following equation.

Here, μ is the linear absorption coefficient at the coordinates (x, y) of the subject, and the coordinates (X, Y) are the new coordinates of the coordinates (x, y) when rotated by the angle θ.

The CT image reconstruction corresponding to FIG. 13 (a) is obtained from the linear absorption coefficient shown by the following equation.

However, since the image obtained from the equation (2) is blurred in inverse proportion to the distance, processing for applying the filter function h (X) to the original data has been adopted. Mathematically, it is expressed by the following equation.

The formula (3) shows a continuous calculation with respect to X and θ. However, since the calculation is usually executed by a digital computer, the formula (3) is modified as the following formula. .

However, k is k = 0, ± 1, ..., ± (M-1), M and N respectively indicate a translation number and a rotation number in the photographing region, and θ _j is θ _j = jπ / N. It is said.

FIG. 14 shows the mesh constructed by the equation (4).
The calculation mesh is an M × M matrix. The correction value of the transmission data of the column X at a certain N (a certain rotation angle) is calculated from the equation (4). However, when rearranging the data in the M × M mesh, the transparent data is multiplied by 1 / M and allocated. That is, regardless of the distribution of the linear absorption coefficient in the column direction of the mesh corresponding to X, the data is uniformly assigned in the column direction. This suggests that the noise appearing in the image increases when the transmission data P is small or when the linear absorption coefficient of the subject increases.

[Object of the Invention]

An object of the present invention is to capture a two-dimensional tomographic image capable of obtaining a clear image by reducing the noise appearing in the reconstructed image even when the transmission data is small or the linear absorption coefficient of the subject is large. The equipment is provided.

[Outline of Invention]

For this purpose, the present invention does not allocate the data evenly to the mesh number M when backprojecting the correction value of the transmission data, but does not allocate the projection data to the mesh known as an air layer in advance. It was done that way. The position and shape of the subject in the imaging region are known in advance by a television camera or the like, or by imaging at the time of collecting projection data.

Example of Invention

 The present invention will be described below with reference to FIGS. 1 to 12.

First, the configuration of an example of a two-dimensional tomographic image pickup apparatus according to the present invention will be described. FIG. 1 shows the configuration together with the subject. According to this, the source box 11 and the detector box 12 mounted on the rotary table 17 are arranged so that the subject 13 is sandwiched therebetween, and a scanning motor (not shown) controlled by the motor control device 14 is provided. )
It is possible to perform the translational scanning via the rotary table 17, and the rotation table 17 allows the rotational scanning. A computer 18 as a control calculation means controls the positions of the source box 11 and the detector box 12 via the motor controller 14, and the signal related to this position is an address signal in the RAM (1).
It is supposed to be input to 20. On the other hand, the amount of transmission of radiation received by the detector in the detector box 12 is converted into a digital signal by a data acquisition circuit 10 including an amplifier and an analog-digital converter, and the same as the position-related signal. RA in computer 18
It is input to M (1) 20. The computer 18 is RAM (1)
Although the transmission data is stored in the address 20 corresponding to the positions of the source box 11 and the detector box 12, the operation up to this point is the same as the data storage method of the CT method up to now.

However, according to the present invention, for example, the television camera 16 is provided at the position where the turntable 17 is seen, and the subject 13
The position and shape of is measured. In this example, it is assumed that the TV camera 16 incorporates MOS photodiodes arranged two-dimensionally in order to facilitate the processing of the output signal of the TV camera 16.

FIG. 2 shows the positional relationship among the source box, the detector box, and the subject as seen from the television camera before the start of scanning. In this case, the source box 11 is imaged.
And the detector boxes 12, but their exact position cannot be easily determined in this state. Therefore, in this example, the light emitting diodes 30 and 31 are installed on the upper surface of the box corresponding to the positions of the radiation source and the detector, respectively. In FIG. 1, the television camera 1
The image formed on the MOS type photodiode in FIG. 6 is converted into an electric signal by the photodiode reading clock from the reading clock generating circuit 15. FIG. 3 shows the signal reading from the MOS type photodiode. It is a thing. As shown in the drawing, the cells in the photodiode are sequentially accessed by the reading clock from the reading clock generating circuit 15 so that the voltage corresponding to the video information is output line by line. For example, if the line number of the photodiode corresponding to the radiation source position and the detector position is l as shown in FIG. 2, the output voltage of the photodiode of line l will be as shown in FIG. Since the light emitting diodes 30 and 31 are installed at the radiation source position and the detector position, respectively, the brightness is particularly high as compared with other portions. Therefore, as shown in FIG. If the comparator 41 in the binarization circuit 25 binarizes with the threshold value TH ₁ , the binarized signal corresponding to the line 1 becomes as shown in FIG. Similarly, if the line number of the photodiode corresponding to the subject 13 is m, the image output of the photodiode corresponding to the line number m is as shown in FIG. In this example, since the reflectance of the subject 13 is lower than that of the other portion (the portion of the air layer), the image output from the subject 13 is low, but also when the reflectance is high, If they are clearly different from the image output value of the air layer, it is possible to easily distinguish them. The read signal corresponding to the line m is converted into a binarized signal corresponding to the line m by being compared with the threshold value TH _2, and this processing is performed by the comparator 42 shown in FIG. Comparator 4
The output pallet trains 1 and 42 are combined in the OR circuit 43 and the RAMs in the computer 8 are processed.
(1) is stored in 20.
In this case, the address of the RAM (1) 20 is generated by counting the read clocks from the read clock generation circuit 15 with the counters 44, 45, and the counters 44, 4
A column signal and a row signal are obtained from each of the five.

FIG. 5 shows the memory contents stored in the RAM (1) 20. As can be seen from FIG. 3, the light emitting diodes 30, 31 and the portion 30 'corresponding to the subject 13 are
"1" is written in 31 'and 13', and "0" is written in the other portions. The computer 18 is RAM (1) 2
After writing the video signal of the MOS type photodiode to 0, the position of the readout line source and the detector and the distance between them are calculated again (calculated by the number of cells of the photodiode), as shown in FIG. In one, the value of the distance between them calculated is n. After this, the computer 18 is a RAM
(1) The data stored in 20 is stored in the RAM as the radiation source position.
(2) The address (0, 0) of 21 and the detector position are stored in the RAM (2) 21 in a state that they correspond to the address (0, M-1) of the RAM (2) 21. I'm running. As a result, the RAM (2) 21 becomes a memory map of size M × M, which facilitates the subsequent processing, and is designed to match the number of meshes M × M required for calculation of CT image reconstruction. Generally, the number of cells of a MOS type photodiode is about 1024 × 1024, whereas the number of meshes for reconstruction is M × M, and the condition of 1024> M is satisfied. 1024 / for one mesh
Information of the MOS type photodiode having M addresses is written. In the example shown in FIG. 5, n is 10
It is less than 24, but its value is close to 1024, so RAM
(1) Data in 20 is n / M in both row and column directions.
It is stored in the RAM (2) 21 in a double compressed state. FIG. 6 shows the RAM (2) 2 for the data shown in FIG.
In the storage state of 1, the value "1" is shown as black and the value "0" is shown as white. Although no particular explanation is necessary for this, CT scanning is started at the time when this transfer and storage is completed.

FIG. 7 shows a calculation mesh for CT image reconstruction. As is well known, this calculation mesh is generally created on a core memory of a computer. By the way, although the rotation term N is involved in the image reconstruction of CT as shown in the equation (4), this is a variable independent of the translation term M, and hence (N = constant at a certain rotation position). ) About the reconstruction method. This is because, in the convolutional integration method, the change in N simply accumulates the transmission data in the rotation direction.
However, in the calculation of the equation (4), data for one translation (i = 0 to 0)
It is clear that the number M of M-1) must be the same.

Now, referring to FIG. 7, the operation according to the present invention will be described with reference to FIG.
This will be described with reference to FIG. First, referring to FIG. 8, this shows a flow of reading the RAM (2) 21.
(2) What part of 21 exists is to be recognized. At the start of reading, column i and row j are initialized. After that, reading is performed while fixing column i and updating row j. In this reading, if the mesh content is "0" (air layer), the next mesh content is immediately read out. If the mesh content is "1", the number of j whose mesh content is "1" is It is calculated and i at that time is described in the core memory. After the number of j, Ma (= the number of meshes of the subject) is calculated, and after that, the process returns to the original loop. j
Is read every time M is updated M times. The process shown in FIG. 8 is 1 even before scanning.
It may be performed at any time when the translational scanning is completed.

FIG. 9 shows a processing flow for performing image reconstruction after the processing shown in FIG. 8 is performed. This processing is supposed to be performed after the completion of one translation scan. In this processing, first, the data for one translation is read, and then it corresponds to the equation (4). Is calculated, and the result (correction data) is stored in the core memory. After this, the initialization of k and n (rotational speed) is performed, and then the already stored i
It is determined whether or not and k do not match, and if they do not match (k ≠ i), the data of the air layer is used, so that the data of “0” is written in the mesh (k, l). is there. If k = i, there is an object to be examined, and therefore data Q shown below is written in this portion.

This process is performed M times while updating the value of k, and the value of n is updated every time this process is performed M times. The loop for n has a value of n-1
Until, that is, the number of rotations is repeated N times, and the whole process regarding k is repeated N times. Therefore, finally, the arithmetic expression relating to the reconstruction of the present invention is represented by the following expression.

As a result of the above calculation, the data "0" is found in the air layer.
The data calculated by the equation (6) is stored in the part of the subject.

In FIG. 1, reference numerals 19, 22, 23 to 25 denote a microprocessor, a program storage ROM, and a program storage ROM, respectively.
CRT (for image table), external storage device (magnetic disk device)
Indicates.

FIG. 10 shows a second embodiment of the present invention. In the first embodiment, the subject is photographed by a television camera under natural light, but in the present embodiment, a light projector including a light source 51 and a lens 52 is used so that light is projected in the direction of the subject. There is. The information of the subject 13 is obtained by photographing with a television camera as in the first aspect,
Of course, in addition to the television camera, an image pickup device in which a condenser lens and a photodiode array are two-dimensionally arranged can also be used. Since the image of the subject can be captured with better contrast in the present aspect than in the first aspect, the surrounding information of the subject is clarified, and a good result is given to the reconstruction of the CT image.

FIG. 11 shows the arrangement of the CT scanner according to the third embodiment. In this mode, the shape of the subject is captured by the television camera or other image pickup device in the first and second modes, whereas the laser or the point light source 61 is either the source box 11 or the detector box 12. The photodiode 62 is arranged on one side and the photodiode 62 on the other side. In the case of this mode, the CT apparatus must be pre-scanned as in the case of data collection, and the shape of the subject must be created on the memory map corresponding to FIG. This causes a problem that the scanning time becomes long, but the number of elements such as photodiodes and the number of channels for signal processing (the number of channels for signal processing in the first and second modes is the square root of the number of elements of photodiodes). (Necessary) is greatly reduced, so that it can be economically implemented and can be obtained at low cost.

〔The invention's effect〕

As described above, according to the present invention, since the transmission data evenly assigned to the mesh corresponding to the air layer is eliminated at all, and a large amount is assigned to the subject portion, the image reconstruction is performed. It is possible to reduce the generated noise (fluctuation of the linear absorption distribution), and it is particularly useful as the difference between the linear absorption coefficient of the subject and the air layer is large. FIG. 12 is an example showing the effect of the present invention, and qualitatively shows the histogram of the central portion of the fantom when the circular fantom is used as the subject shown in FIGS. The figure (a) relates to the conventional method, and the figure (b) relates to the method of the present invention. The noise 71 appearing in FIG. 7A has a certain magnitude regardless of the position (when the count value of the radiation assigned to the mesh is n. Have.

However, the counting noise in the present invention can be reduced because a large amount of data is allocated in the subject. The quantification constant for example, there count value 10 ⁷ counts the transmission data at the translational position, count per mesh screen when the mesh screen number and 100 becomes 10 ^{6/10 2} = 10 ⁴ counts noise 1% . Noise in the present invention is the same as in the conventional method, and if a (the number of meshes of the air layer) is 50, that is, if half of the imaging area is the air layer,
The percentage of noise that appears in the image is Therefore, it is greatly reduced compared to the conventional method. This effect increases as the number of meshes of a from Equation (5) or Equation (6) increases.

Still another effect is that the difference between the level of the air layer (the level of the linear absorption coefficient) and the level of the subject becomes large as seen in the CT reconstructed images in FIGS. 12 (a) and 12 (b). , Image processing techniques usually used for medical purposes, for example, facilitating level setting of ROI (region of interest) display, or edge of a subject caused by a steep inclination of the level between the air layer and the subject It can be mentioned that processing such as emphasis becomes easier to execute.

It is to be noted that the software as shown in FIGS. 8 and 9 is required to correspond the image captured by the television camera and the calculation mesh, which requires a lot of processing time.
When this portion is implemented as, for example, a ROM or firmware, the processing time is greatly shortened and the effect of the present invention can be further improved.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration of an example relating to a basic aspect of a two-dimensional tomographic imaging apparatus according to the present invention together with a subject, and FIG. 2 is a radiation source box seen from a television camera in FIG. FIG. 3 is a diagram showing the positional relationship between the detector box and the subject, FIG. 3 is a diagram showing the image reading from the TV type MOS photodiode in the television camera and the subsequent processing, and FIG. 4 is a diagram showing the address generation in FIG. FIGS. 5 and 6 are diagrams showing a specific circuit configuration of the binarization circuit, and FIGS. 7 and 7 are views for explaining storage of a video image including an object in a memory and conversion thereof into a calculation mesh. Is a diagram generally showing a calculation mesh for CT image reconstruction, and FIGS. 8 and 9 are diagrams showing a flow of processing for recognizing an object existing position and a flow of image reconstruction processing. 10 and 11 are views showing another embodiment of the present invention, and FIG. (A), (b) are diagrams for explaining the extent of effects of the present invention, FIG. 13 (a), (b)
FIG. 14 is a diagram showing a transmission data acquisition method and an image reconstruction method, and FIG. 14 is a diagram showing an M × M matrix calculation mesh. 10 ... Data collection circuit, 11 ... Source box, 12 ... Detector box, 13 ... Subject, 14 ... Motor control device,
16 ... Television camera, 17 ... Rotating table, 18 ... Calculator, 2
5 ... Address generation / binarization circuit.

Claims (1)

[Claims] 1. A two-dimensional tomographic image is obtained by collecting transmission data as transmitted radiation from an object and the periphery of the object with respect to irradiation radiation and back-projecting a calculation processing result of the data onto a calculation mesh. In a two-dimensional tomographic imaging apparatus for reconstructing, a means for previously specifying the existence area of the subject on the calculation mesh of the reconstruction calculation, and a mesh area without assigning transmission data to the mesh area corresponding to the air layer A backprojection calculation means for setting a value of "0" and performing the backprojection calculation of the image reconstruction only on the calculation mesh corresponding to the existence region of the specified subject on the basis of the transmission data that is allocated correspondingly. A two-dimensional tomographic imaging apparatus equipped with and.