JP2008026144A - Image diagnosis support system and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve diagnosis accuracy by eliminating the variances due to individuals in image diagnosis, performed by comparing each tomographic image of the examinee and a normal person. <P>SOLUTION: This system is equipped with storage parts 11, 12 for storing the voxel data of each tomographic image of the subject and a plurality of healthy persons, an AI data calculation part 23 for calculating the index (AI value), showing the asymmetry of each tomographic image of the subject and a plurality of persons to be compared by comparing the right and left sides of the same tomographic image, based on the voxel data of the tomographic image; a Z-value calculation part 25 for calculating Z value by comparing the AI value of the tomographic image of the examinee with the AI values of each tomographic image of the plurality of persons to be compared, and a display control part 27 for displaying the tomographic image of the subject and the Z-value. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a technique for assisting diagnosis by utilizing left-right symmetry or asymmetry in a tomographic image of a subject.

  2. Description of the Related Art In recent years, an image diagnostic machine (X-ray CT (Computed Tomography), MRI (Magnetic Resonance Imaging), ultrasonic diagnostic machine) for capturing a state in the body as an image without imposing a heavy burden on a subject or the like and performing an accurate diagnosis. , Radiation diagnostic machines, etc.) are indispensable in the current medical field.

  In the clinical field of nuclear medicine, single photon emission computed tomography (SPECT) and positron emission tomography using a radioisotope introduced into a subject and utilizing gamma rays emitted from the radioisotope. Photography (Positron Emission Tomography, PET) is used. As a result, it has become clear that blood flow in a specific part of the brain increases or decreases for each disease by taking and analyzing a tomographic image of the brain. Therefore, it is possible to extract the presence / absence of an abnormal site by comparing SPECT and PET images of a subject and a healthy person.

The images obtained so far are evaluated visually by a diagnostician or using a statistical image analysis method in which computer image processing is added to the visual evaluation method. As a statistical image analysis method, an SPM method (Statistical Parametric Mapping), a 3D-SSP method (Three-dimensional Stereotactic Surface Projections), and the like have been developed and used. Thereby, the abnormal part of a subject can be extracted objectively and it is useful for the diagnosis of a disease.
Image diagnosis in clinical psychiatry, volume S10, Nakayama Shoten

  However, the conventional method uses a relative count value of each area with respect to the average count of the whole brain, and a relative count value with respect to a specific area in the brain such as the cerebellum and the thalamus, thereby making a difference from a normal group or a specific disease group. If there is a change in a wide range of the entire brain, or if the region considered as the reference changes, the detection ability of the abnormal part will deteriorate and the degree of abnormality will be accurately evaluated. There was a problem that it was not possible.

  In addition, some brain diseases include lesions that occur only on one side, such as cerebrovascular disorders and epilepsy, and others that progress with left-right asymmetry, such as degenerative diseases. In these cases, even if there is a left-right difference in the tomographic image, there is a case where the left-right difference cannot be detected by the conventional method. Therefore, in addition to the conventional method, development of an image evaluation method based on a new scale has been demanded.

  Therefore, an object of the present invention is to provide a technique for supporting diagnosis based on the symmetry or asymmetry of the left and right brains in a brain tomographic image of a subject.

  The diagnostic imaging support system according to one embodiment of the present invention is to support diagnosis by comparing a tomographic image of a subject with a tomographic image of a plurality of contrasters for comparison with the subject. Based on the tomographic image data, storage means for storing the tomographic image data of the subject and the plurality of contrasters, and performing left-right comparison of the same tomographic image, the tomographic images of the subject and the plurality of contrasters By an asymmetry index calculation means for calculating an index indicating asymmetry, a comparison means for comparing an asymmetry index of the tomographic image of the subject and an asymmetry index of the tomographic images of the plurality of contrasters, and the comparison means Means for outputting a comparison result. Here, the plurality of contrasters may be persons having common characteristics, such as healthy persons and patients with specific diseases.

  In a preferred embodiment, the asymmetry index calculating means compares the left region of interest and the right region of interest in the same tomographic image with respect to the left and right regions of interest on the tomographic image, which are set to be bilaterally symmetric in advance. The asymmetry index may be calculated.

  In a preferred embodiment, representative value calculating means for calculating a value representative of each region of interest in the tomographic images of the subject and the plurality of contrasters based on the tomographic image data of the subject and the plurality of contrasters is further provided. You may have. The asymmetry index calculating means may calculate the asymmetry index of each tomographic image using the representative value calculated by the representative value calculating means.

  In a preferred embodiment, the representative value calculating means may calculate an average value of tomographic image data in each sensation area as a representative value of each region of interest.

  In a preferred embodiment, the comparison means calculates, for each region of interest, a degree of divergence between an average value of the asymmetry index of the plurality of contrasters and the asymmetry index of the subject, and the output means includes a display device. Are output so that the tomographic image of the subject is displayed, and each region of interest on the displayed tomographic image is displayed in a different display mode according to the degree of divergence of each region of interest. May be.

  In a preferred embodiment, the tomographic images of the subject and the plurality of contrasters are both SPECT (Single Photon Emission Tomography) tomographic images of the brain, and the region of interest is preset for a specific disease. The asymmetry index may be an index indicating the difference between the left and right blood flow rates.

  Hereinafter, an image diagnosis support system according to an embodiment of the present invention will be described with reference to the drawings. The image diagnosis support system according to the present embodiment performs image analysis by comparing SPECT tomographic images of the brains of a subject and a healthy person, and supports diagnosis of the subject's disease.

  FIG. 1 shows an overall configuration diagram of an image diagnosis support system 1 according to the present embodiment.

  An image diagnosis support system 1 according to the present embodiment includes an image diagnosis support system main body 10, an input device 2 connected to the image diagnosis support system main body 10, and a display device 3. The diagnostic imaging support system main body 10 is configured by, for example, a general-purpose computer system, and individual components or functions in the diagnostic imaging support system main body 10 described below are realized by, for example, executing a computer program. .

  The image diagnosis support system main body 10 stores a normalization processing unit 20, a subject voxel data storage unit 11 that stores image data of a subject to be diagnosed, and image data of a healthy person that has been collected in advance. A healthy person voxel data storage unit 12, a VOI data storage unit 13 storing VOI data indicating a preset three-dimensional sense of interest (VOI), and a two-dimensional form formed by each VOI in each tomographic image A ROI value calculation unit 21 that calculates a value representative of a region of interest (ROI), an AI data calculation unit 23 that calculates an index indicating asymmetry (AI: Asymmetry Index) using the ROI value, and the like Memorization of subject's asymmetry index data (hereinafter referred to as AI data) The subject AI data storage unit 15, the healthy subject AI data storage unit 16 that stores the AI data of the healthy person, and the ROI of the tomographic image of the subject and the healthy person A Z value calculation unit 25 that calculates a Z value based on the AI value, and a display control unit 27 that displays a comparison result between the tomographic image of the subject and the tomographic image of the healthy person on the display device 3 are provided.

  The normalization processing unit 20 normalizes the voxel value data of the SPECT tomographic image of the subject imaged by the SPECT tomographic image imaging device 5 to a standard brain having a predetermined shape and size. The shape and size of the human brain vary from person to person. Therefore, the raw data captured by the SPECT tomographic image capturing apparatus 5 is normalized so as to match a standard brain whose shape and size are determined in advance for simplification of subsequent processing.

  Note that the level of the voxel value of the captured image differs depending on the individual difference of the SPECT tomographic imaging apparatus 5 and the imaging conditions, but the normalization processing unit 20 does not normalize the level of the voxel value. As will be described later, in this embodiment, the left region and the right region in the same image of the subject and the healthy person are compared and the right and left balances of the subject and the healthy person are compared and analyzed. For this reason, it is sufficient to obtain substantially the same effect as normalization of the level of the voxel value and to accurately grasp the right and left balance at the time when the left and right are compared in the same image.

  The subject voxel data storage unit 11 stores voxel value data (hereinafter referred to as subject voxel data) normalized to the standard brain by the normalization processing unit 20.

  The healthy person voxel data storage unit 12 stores voxel value data (hereinafter referred to as healthy person voxel data) of a brain tomographic image of one or more healthy persons for comparison with the subject. Healthy person voxel data is also normalized to the standard brain.

  FIG. 2 is a diagram illustrating an example of the data structure of the subject voxel data 111 and the healthy person voxel data 121.

  The test subject and healthy subject voxel data 111 and 121 in the same figure have a common data structure. When the horizontal direction of the head is the X axis, the longitudinal direction is the Y axis, and the vertical direction is the Z axis, the Z axis direction And N pieces of XY cross-sectional image data. The voxel value included in each image data corresponds to the pixel value of each image.

  The VOI data storage unit 13 stores data indicating a VOI preset by a doctor or the like. Although the VOI can be set arbitrarily, for example, in the present embodiment, a predetermined region in the brain that the doctor is paying attention to for each disease is set.

  Here, the disease-specific VOI is a region in which attention is paid to diseases such as epilepsy and Alzheimer's disease, stage, and progression (severity). Here, these classifications can be created by dividing a group satisfying a certain standard by a predetermined diagnostic standard or various examinations according to a threshold value. The threshold value may be determined by a ROC analysis method in addition to a statistical method such as test statistics.

  For example, the disease-specific VOI data may be generated as follows. That is, by superimposing a large number of subject data of a certain disease and extracting a region most characteristic of the disease, VOI data having the extracted region and other regions as VOIs is generated.

  The VOI may be determined based on an anatomical or functional classification. Here, the division based on the anatomical classification refers to, for example, a case where a brain region is divided into a cerebrum, a cerebellum, and the like. Since the anatomical classification is hierarchical, this embodiment also has a hierarchical classification. Functional classification refers to, for example, division into functional areas such as motor areas and language areas localized in each part of the brain. The setting of each divided area is performed based on predetermined anatomical or functional knowledge.

  Further, the VOI may be set according to the blood vessel dominating region of the brain. This is useful for diagnosis of vascular diseases and degenerative diseases. That is, in a vascular disease, if an abnormality occurs in a blood vessel, any change in blood flow on the left or right along the blood vessel can be detected. Also, some brain diseases are known to have lesions on only one side, such as cerebrovascular disorders and epilepsy, and to progress with left-right asymmetry, such as degenerative diseases. Diagnosis of disease and pathological evaluation can be performed effectively.

  FIG. 3 schematically shows a VOI set in a three-dimensional manner and an ROI formed on each tomographic image by the VOI.

  First, as shown in FIG. 2A, VOIs 101 and 102 are set symmetrically with respect to the standard brain. This is because in the present embodiment, diagnosis support using the asymmetry of the blood flow between the left and right brains is performed.

  At this time, FIG. 7B shows the state of ROIs 201, 202, 301, 302 formed by slicing VOIs 101, 102 in the tomographic images of z = m and z = n. Here, the ROIs 201, 202, 301, and 302 are also symmetrical.

  Further, as can be seen from the figure, the shape of the ROI (ROI 201, 301) varies depending on the slice position (that is, the value of z) even in the same VOI (for example, VOI 101). Here, the ROI (for example, ROI 201) formed in the tomographic image is further divided into a front side (ROI 201a) and a rear side (ROI 201b).

  Here, each ROI is assigned an ID. Different IDs are assigned to ROIs generated from the same VOI on different tomographic images.

  Next, FIG. 4 is a diagram illustrating an example of VOI data stored in the VOI data storage unit 13. That is, the VOI data storage unit 13 can store a plurality of sets of VOI settings such as “VOI setting 1”, “VOI setting 2”, and so on. For each VOI setting, for each tomographic image, “0” is set for the pixel of the portion (background) that is not the brain image, and “1” is set for the pixel of the region that does not belong to any VOI in the brain image. ing. For the pixels belonging to each VOI, numbers after “2” are set for each VOI.

  Referring again to FIG. 1, the ROI value calculation unit 21 first specifies the position of the ROI set in the image for each tomographic image based on one VOI setting in the VOI data storage unit 13. Pixels belonging to ROI are determined. Then, the ROI value calculation unit 21 calculates an average value of voxel values of pixels belonging to each ROI in each tomographic image as a value representative of the ROI (hereinafter referred to as ROI value).

The AI data calculation unit 23 compares the ROI values of the ROIs corresponding to the ROIs set symmetrically (for example, 201a and 202a in FIG. 3), and calculates an AI value that is an asymmetry index of each ROI. To do. In the present embodiment, for example, the AI value of each ROI is calculated as follows using each ROI value.
AI value of left ROI = (left ROI value−right ROI value) / (left ROI value + right ROI value) * 100
AI value of right ROI = (right ROI value−left ROI value) / (left ROI value + right ROI value) * 100
Left ROI value: ROI value of the corresponding left ROI Right ROI value: ROI value of the corresponding right ROI

  The AI data calculation unit 23 calculates the AI value for the corresponding left and right ROI for each tomographic image for each of the subject and the healthy person. The AI value of each ROI calculated by the AI data calculation unit 23 is stored in the subject AI data storage unit 15 and the healthy subject AI data storage unit 16, respectively.

  That is, in the subject AI data storage unit 15 and the healthy person AI data storage unit 16, as shown in FIGS. 5A and 5B, the IDs 151 and 161 of the subject or the healthy person, the IDs 152 and 162 of the ROI, and the AI value 153 and 163 are stored.

  The AI data calculation unit 23 further calculates an average value and a standard deviation of the AI values of all healthy persons for each ROI when there are data of a plurality of healthy persons, and stores them in the healthy person AI data storage unit 16. .

  In the present embodiment, the AI data calculation unit 23 calculates an AI value for each ROI, but may calculate an AI value for each pixel.

  That is, as shown in FIG. 5C, the healthy person AI data storage unit 16 further stores the ROI ID 166, the average value 167 of the RO values for the healthy persons by ROI, and the standard deviation 168.

The Z value calculation unit 25 calculates the Z value for each subject using the average value and the standard deviation of the healthy person for all ROIs. That is, the Z value calculation unit 25 calculates the Z value for each ROI according to the following equation.
Z value = {(average value of AI value of healthy subject) − (subject AI value)} / standard deviation of AI value of healthy subject

  By calculating the Z value for each subject's ROI, the degree to which the subject's AI value deviates from the average value of the healthy subject's AI value is grasped.

  The display control unit 27 displays a tomographic image of the subject on the display device 3. At the same time, the processing result by the Z value calculation unit 25 is displayed on the display device 3. That is, in the tomographic image, each ROI may be displayed in a display mode corresponding to the Z value of the ROI.

  In addition, the display control unit 27 may display the tomographic image of the subject in a display mode corresponding to the subject's AI value itself. This is because diagnosis support may be possible with the AI value itself based on the subject image without comparing the subject image with the image of the healthy person.

  FIG. 6 is an example of a display mode of the tomographic image of the subject and the analysis result output to the display device 3. That is, in the example of the figure, the tomographic image 500 of the subject is displayed, and is displayed in a predetermined color according to the Z value of each ROI.

  Next, the processing procedure of the diagnostic imaging support system 1 having the above-described configuration will be described with reference to the flowcharts shown in FIGS.

  FIG. 7 shows a procedure of a healthy person process for processing healthy person voxel data to generate healthy person AI data.

  First, the ROI value calculation unit 21 reads the voxel data of the healthy person stored in the healthy person voxel data storage unit 12, and performs the following processing on all the tomographic images of all the healthy persons. That is, all ROIs in one tomographic image are specified, and an ROI value that is an average value of voxel values in each ROI is calculated (S11).

  Next, the AI data calculation unit 23 calculates the AI value of all ROIs based on the ROI value of each ROI in the same tomographic image, and stores it in the healthy subject AI data storage unit 16 (S12).

  The processes of steps S11 and S12 are repeated for all tomographic images of one healthy person (S13).

  Furthermore, the process of step S11-S13 is performed about the voxel data of all the healthy persons stored in the healthy person voxel data storage part 12 (S14).

  And the AI data calculation part 23 reads AI value of all the healthy persons for every ROI-ID from the healthy person AI data storage part 16, calculates the average value and standard deviation of the AI value of each ROI, and is healthy. Stored in the user AI data storage unit 16 (S15).

  By the above-described processing, the average value and standard deviation of the AI value of each ROI based on the data of healthy persons can be obtained in advance. In this state, the subject process described below can be performed.

  FIG. 8 shows a procedure of subject processing for processing subject voxel data and performing comparison with a healthy person.

  First, the tomographic image data (voxel data) of one subject is acquired from the SPECT tomographic imaging apparatus 5, and after normalization is performed by the normalization processing unit 20, it is stored in the subject voxel data storage unit 11 ( S21).

  Next, the ROI value calculation unit 21 reads the voxel data of the subject stored in the subject voxel data storage unit 11, and specifies all the ROIs in the tomographic image for each tomographic image of all the tomographic images of the subject. The ROI value that is the average value of the voxel values in each ROI is calculated (S22).

  Next, the AI data calculation unit 23 calculates the AI values of all the ROIs based on the ROI values of the ROIs in the same tomographic image, and stores them in the subject AI data storage unit 15 (S23).

  Steps S22 and S23 are repeated for all tomographic images of the subject (S24).

  Thereafter, the Z value calculating unit 25 acquires the ROI AI value of the subject subject from the subject AI data storage unit 15. Moreover, the average value and standard deviation of the AI value of all the healthy persons of each ROI are acquired from the healthy person AI data storage unit 16. Then, the Z value calculation unit 25 calculates a Z value for each ROI (S25).

  The display control unit 27 displays a tomographic image of the subject, and displays the ROI on the tomographic image in a display mode corresponding to each Z value (S26).

  According to this system, it is possible to easily know the ROI of a subject who exhibits characteristics different from those of a healthy person. Furthermore, since the display mode is made different according to the Z value, the degree of difference from a healthy person can be understood at a glance, and a doctor's diagnosis can be supported.

  In this embodiment, the whole brain average of the voxel values is not taken, the right and left balance of the blood flow of the individual is grasped, and the subject and the healthy person are compared in this left and right balance. As a result, it is possible to detect a subject whose left and right blood flow balance is lost compared to a healthy person. Therefore, the system according to the present embodiment is suitable for diagnosis support for diseases in which the blood flow tends to be asymmetric between the left and right brains.

  In addition, when normalization is performed by taking the whole-brain average of the test subject's voxel value, even if there is a slight increase or decrease in blood flow at a certain location, if the increase or decrease is within the standard deviation range, it is regarded as a problem. There is nothing. As a result, even if there is some abnormality at that location, it will be overlooked, but according to the present embodiment, if the left and right balance is less than the average of healthy people, such detection There is no leakage.

  The above-described embodiments of the present invention are examples for explaining the present invention, and are not intended to limit the scope of the present invention only to those embodiments. Those skilled in the art can implement the present invention in various other modes without departing from the gist of the present invention.

  For example, in the present embodiment, the average value of the voxel values in the ROI is obtained, and the AI value and the Z value are calculated using the average value. After calculating the AI value and the Z value for each pixel, the ROI is set. May be displayed.

  In the present embodiment, the left and right blood flow volumes are compared using a tomographic image by SPECT, and diagnosis support is performed based on the comparison. However, in addition to SPECT, tomographic images by PET, X-ray CT, and MRI are also used. It can also be used. Here, in the case of SPECT and PET images, in addition to the voxel value indicating blood flow, for example, an integrated value indicating increase / decrease in accumulation on the receptor may be used. Further, CT values may be used for CT images, and T1 values and T2 values (relaxation times) may be used for MRI images.

  Or in the said embodiment, although Z value is used for a test subject and a healthy person's comparison, it is not necessarily limited to this. That is, it suffices if the subject and the healthy person can be compared with each other to extract a region where the subject shows different characteristics from the healthy person. For example, t value or F value can be used.

  Furthermore, in the above-described embodiment, the contraster to be compared with the subject is a healthy person, but besides this, for example, the average value of a large number of patient data of a specific disease is compared with the subject data. Also good. For example, when an early stage Alzheimer's disease patient is used as a contraster, a difference between the subject and the early Alzheimer's disease patient is detected. And if it compares with a test subject using the VOI data according to disease of late stage Alzheimer's disease, the progress degree of a test subject's Alzheimer's disease can also be determined. Furthermore, by comparing the average of multiple patient data for one disease (eg frontal lobe Alzheimer's disease) with the subject of that disease using VOI data for another disease (eg depression), It is also possible to discriminate between subjects who are co-occurring and subjects who are not.

1 is an overall configuration diagram of an image diagnosis support system 1 according to an embodiment of the present invention. An example of the data structure of healthy person voxel data and test subject voxel data is shown. The three-dimensionally set VOI and the state of the ROI formed on each tomographic image are shown. It is a figure which shows an example of VOI data. An example of the data structure of a test subject AI data storage part and a healthy subject AI data storage part is shown. It is an example of a display mode of a tomographic image of a subject to be output to a display device and an analysis result. It is a flowchart which shows the procedure which produces | generates healthy person AI data. It is a flowchart which shows the procedure which contrasts a test subject and a healthy person.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Image diagnosis support system 2 Input device 3 Display apparatus 5 Tomographic imaging apparatus 10 Image diagnosis support system main body 11 Subject voxel data storage part 12 Healthy person voxel data storage part 13 VOI data storage part 15 Subject AI data storage part 16 Healthy person AI Data storage unit 21 ROI value calculation unit 23 AI data calculation unit 25 Z value calculation unit

Claims (8)

A system for performing diagnosis support by comparing a tomographic image of a subject with a tomographic image of a plurality of contrasters for comparison with the subject,
Storage means for storing tomographic image data of the subject and the plurality of contrasters;
Based on the tomographic image data, the left and right comparison of the same tomographic image is performed, an asymmetry index calculating means for calculating an index indicating the asymmetry of each tomographic image of the subject and the plurality of contrasters;
Comparison means for comparing the asymmetry index of the tomographic image of the subject and the asymmetry index of the tomographic images of the plurality of contrasters;
An image diagnosis support system comprising: means for outputting a comparison result by the comparison means;   The asymmetry index calculating means calculates the asymmetry index of each tomographic image by comparing the left region of interest and the right region of interest in the same tomographic image with respect to the left and right regions of interest on the tomographic image that are set to be symmetrical in advance. The diagnosis support system according to claim 1. Based on the tomographic image data of the subject and the plurality of contrasters, further comprising representative value calculation means for calculating a value representing each region of interest in the tomographic images of the subject and the plurality of contrasters,
The image diagnosis support system according to claim 2, wherein the asymmetry index calculating means calculates an asymmetry index of each tomographic image using the representative value calculated by the representative value calculating means. The representative value calculating means includes
The image diagnosis support system according to claim 3, wherein an average value of tomographic image data in each sensitivity area is calculated as a representative value of each region of interest. The comparison means calculates, for each region of interest, the degree of divergence between the average value of the asymmetry index of the plurality of contrasters and the asymmetry index of the subject,
The output means outputs so as to display the tomographic image of the subject on the display device, and displays different regions of interest on the displayed tomographic image according to the degree of divergence of the regions of interest. The image diagnosis support system according to claim 2, wherein the image diagnosis support system outputs the image so as to display the image. The tomographic images of the subject and the plurality of contrasters are all SPECT (Single Photo Emission Computed Tomography) tomographic images of the brain,
The region of interest is a region set in advance for a specific disease,
The image diagnosis support system according to claim 1, wherein the asymmetry index is an index indicating a difference between left and right blood flow rates. Storing tomographic image data of a subject and tomographic image data of a plurality of contrasters for comparison with the subject;
Performing left and right comparison of the same tomographic image based on the tomographic image data, calculating an index indicating the asymmetry of each tomographic image of the subject and the plurality of contrasters;
Comparing the asymmetry index of the tomographic image of the subject with the asymmetry index of the tomographic images of the plurality of contrasters;
Outputting the comparison result, and an image diagnosis support method. When executed on a computer,
Storing tomographic image data of a subject and tomographic image data of a plurality of contrasters for comparison with the subject;
Performing left and right comparison of the same tomographic image based on the tomographic image data, calculating an index indicating the asymmetry of each tomographic image of the subject and the plurality of contrasters;
Comparing the asymmetry index of the tomographic image of the subject with the asymmetry index of the tomographic images of the plurality of contrasters;
A computer program for supporting image diagnosis, which executes the step of outputting the comparison result.

Images