JP6036009B2 - Medical image processing apparatus and program - Google Patents

Description

  The present invention relates to a technique for performing diagnosis support for a brain disease by performing image processing on a brain image obtained by MRI (Magnetic Resonance Imaging) and the like, and in particular, a brain image obtained by MRI or the like in a state suitable for diagnosis support. It relates to processing technology.

  With the arrival of an aging society, the number of patients with dementia is increasing year by year. There are various types of dementia diseases, and in diagnosis, it is necessary to distinguish between them and perform appropriate treatment according to the disease.

  On the other hand, in order to meet such demands, in recent years, information on the state of the brain can be acquired by nuclear medicine examinations such as SPECT (Single Photon Emission Computed Tomography) and PET (Positron Emission Tomography), CT (Computererized Tomography) and MRI. It is becoming.

  As a result, it has become clear that blood flow and metabolism in specific parts of the brain decrease and tissue atrophy changes depending on the disease, and a quantitative evaluation method for these is required. .

  For example, a decrease in blood flow or metabolism in a local region of the brain can be tested by comparing with SPECT or PET images.

  As for tissue atrophy, the volume of a specific part can be obtained from an MRI image, and the relative size can be compared to determine the presence or absence of an abnormality.

  As a method of evaluating brain atrophy using such a brain image, a brain image obtained by imaging the head of a subject is image-processed in units of voxels that are three-dimensional pixels. Based Morphometry) is known (for example, see Patent Document 1).

  This VBM method is an effective evaluation method for identifying Alzheimer's disease, and it has been reported that there was 87.8% diagnostic ability in distinguishing between healthy subjects and Alzheimer's disease (see Non-Patent Document 1). .

  In addition, in order to improve the conventional VBM method that the accuracy for white matter is low, a technique for accurately evaluating the degree of atrophy of white matter extracted by tissue separation from an MRI brain image of an input subject has been developed ( Patent Document 2).

JP 2005-237441 A WO2011 / 040473

Yoko Hirata, Hiroshi Matsuda, Kiyotaka Nemoto, Takashi Ohnishi, Kentaro Hirao, Fumio Yamashita, Takashi Asada, Satoshi Iwabuchi, Hirotsugu Samejima, Voxel-based morphometry to discriminate early Alzheimer's disease from controls.Neurosci Lett 382: 269-274, 2005 J. Ashburner and K. J. Friston. Unified segmentation. NeuroImage. 2005; 26: 839-851. Ashburner J, A fast diffeomorphic image registration algorithm.Neuroimage.2007 Oct 15; 38 (1): 95-113. Hiroshi Matsuda: Statistical image analysis of SPECT. Imaging diagnosis of Alzheimer-type dementia, Medical View: pp. 76-86 (2001).

  However, with the above-described conventional technology, changes in tissue in the brain over time could not be correctly grasped.

  Therefore, an object of the present invention is to provide a medical image processing apparatus and a program capable of correctly grasping changes in tissues in the brain over time and supporting diagnosis.

  In order to solve the above-described problem, in the first aspect of the present invention, after performing tissue separation processing on two brain images of the same healthy person with different imaging timings, changes in the healthy person expressing the amount of change in both images are performed. A healthy person change amount image database storing a quantity image, tissue separation means for obtaining a tissue image by performing tissue separation processing on each of two brain images of the same subject at different imaging times, and the two obtained tissue images Between the voxels constituting the tissue image, the change amount calculation means for obtaining the change amount image, the change amount image obtained by the change amount calculation means, and the healthy person change amount image database Statistical ratio for obtaining a statistically processed image by performing a statistical comparison with the change image of the healthy person, calculating a predetermined evaluation value, and replacing the value of each voxel of the change image with the calculated predetermined evaluation value Providing means, and display means for displaying the statistical processing images, the medical image processing apparatus characterized by having a.

  The predetermined evaluation value calculated by the statistical comparison is a value obtained by scaling the difference between the voxel value of the variation image of the subject and the average voxel value of the corresponding voxel of the healthy subject variation image group by the standard deviation. Z score and t score used in other general tests can be adopted.

  According to the first aspect of the present invention, between two tissue separated images of the same subject whose imaging time is different, a difference image between the voxels is obtained to obtain a variation image, and the obtained variation image and a healthy person Statistical comparison with the healthy person's change amount image, which is the change amount image of each person, and each voxel is displayed as a statistically processed image expressed with a predetermined evaluation value, so the secular change is large compared to the healthy person It is possible to easily grasp the location.

  According to a second aspect of the present invention, in the medical image processing apparatus according to the first aspect of the present invention, the healthy person variation image database stores an elapsed period in association with each healthy person variation image. , Means for inputting information for specifying an elapsed period between two brain images of the subject, an elapsed period associated with the healthy person variation image, and an elapsed period between the brain images of the subject And a normal person change amount image correcting means for correcting the value of each voxel of the normal person change amount image.

  According to the second aspect of the present invention, the elapsed period is stored in association with the healthy person variation image, and the elapsed period of the healthy person variation image and the elapsed time between the two brain images of the input subject. Since the value of each voxel in the healthy person change amount image was corrected using the period, in the state where the elapsed period having a large influence on the change amount is aligned, the point where the secular change is large compared with the healthy person It becomes possible to grasp easily.

  Further, in the third aspect of the present invention, the tissue separation means for obtaining a tissue image by performing tissue separation processing on each of two brain images of the same subject with different imaging times, and the two obtained tissue images are configured. A difference calculation means for obtaining a change amount image by calculating a difference between the values of each voxel, calculating a standard deviation of the voxels in the change amount image, and dividing the value of each voxel by the calculated standard deviation. There is provided a medical image processing apparatus comprising: a normalizing unit that obtains a normalized variation image, and a display unit that displays the normalized variation image.

  According to the third aspect of the present invention, between two tissue separated images of the same subject whose imaging time is different, a difference calculation is performed between the voxels to obtain a change amount image, and the voxels in the obtained change amount image are obtained. The standard variation image obtained by dividing the value of each voxel by the calculated standard deviation is displayed, reducing the effects of differences in shooting models and It is possible to easily grasp a place where the change is large.

  According to a fourth aspect of the present invention, in the medical image processing apparatus according to any one of the first to third aspects of the present invention, the tissue separation unit separates white matter or gray matter from a brain image as the tissue separation process. Processing is performed to obtain a white matter image or gray matter image as a tissue image.

  According to the fourth aspect of the present invention, the change amount of the tissue image of white matter or gray matter, which is the tissue in which the change amount is most easily grasped, is created as the change amount image. It becomes possible to grasp.

  According to the present invention, there is an effect that it is possible to correctly grasp changes in brain tissue over time and to support diagnosis.

It is a figure which shows the medical image processing apparatus in one Embodiment of this invention. It is a flowchart which shows the process outline | summary of the medical image processing apparatus which concerns on reference embodiment of this invention. It is a figure which shows the mode of the change of the image by gray matter change amount calculation. It is a figure which shows the example which displayed the variation | change_quantity by the numerical value according to site | part. It is a flowchart which shows the process outline | summary of the medical image processing apparatus which concerns on 2nd Embodiment. It is a figure for demonstrating the statistical comparison which concerns on 2nd Embodiment. It is a conceptual diagram which shows the characteristic of the analysis by ROI. It is a conceptual diagram which shows the characteristic at the time of creating ROI. It is a flowchart which shows the process outline | summary of the medical image processing apparatus which concerns on 3rd Embodiment. It is a figure for demonstrating the diagnostic assistance using the atrophy degree image which recorded the atrophy degree of the brain.

<1. Device configuration>
DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, preferred embodiments of the invention will be described in detail with reference to the drawings. FIG. 1A is a hardware configuration diagram of a medical image processing apparatus according to an embodiment of the present invention. The medical image processing apparatus can be realized by a general-purpose server computer. As shown in FIG. 1A, a CPU (Central Processing Unit) 1a and a RAM (Random Access Memory) 1b which is a main memory of the computer , A large-capacity storage device 1c (for example, a hard disk, a flash memory, etc.) for storing programs and data executed by the CPU, an input unit 1d realized by an input device such as a keyboard and a mouse, and an external device (data A data input / output I / F (interface) 1e for data communication with a storage medium or the like and a display unit 1f such as a liquid crystal display are connected to each other via a bus.

  FIG. 1B is a functional block diagram of a medical image processing apparatus according to an embodiment of the present invention. In the hardware configuration shown in FIG. 1A, the CPU 1a reads the program stored in the storage device 1c into the RAM 1b and executes it, so that the medical image processing apparatus performs each unit shown in FIG. It becomes possible to make it function. As shown in FIG. 1B, the medical image processing apparatus includes a tissue separation unit 11, a change amount calculation unit 12, a statistical comparison unit 13, a normal person change amount image correction unit 14, a normalization unit 15, and a normal person change amount. It has an image database D1 and a sick person variation image database D2.

  The healthy person change amount image database D1 is a database that stores a healthy person change amount image that is an image indicating a change amount of the healthy person. The sick person change amount image database D2 is a database in which a sick person change amount image which is an image showing a change amount of the sick person is stored. The healthy person change amount image database D1 and the sick person change amount image database D2 are realized by securing predetermined areas for storing the healthy person change amount image data and the sick person change amount image in the storage device 1c. . The healthy person change amount image is generated by the procedure of steps S1 to S5 described later on the basis of two brain images of the same healthy person with different imaging times. In addition, the diseased person change amount image is generated by the procedure of steps S1 to S5 described later, based on two brain images of the same diseased person with different imaging times. In the case of a healthy person variation image, a photographed image that is determined to have no disease is used at any photographing time, but in the case of a sick person variation image, a photograph that has been determined to have no disease as the previously captured image. An image is used, and a captured image determined to have a disease is used as an image captured later.

  The tissue separation unit 11 is a unit that obtains a tissue image by performing a tissue separation process on the brain image to be processed. The change amount calculation means 12 is a means for obtaining a change amount image by performing a difference calculation between corresponding voxels between two tissue images. The statistical comparison means 13 is a means for performing statistical comparison between the change amount image and the healthy person change amount image in the healthy person change amount image database D1. The healthy person change amount image correction means 14 is a means for correcting the voxel value of the healthy person change amount image based on the elapsed time between the change amount image and the healthy person change amount image. The normalizing unit 15 is a unit that obtains a normalized variation image by calculating a standard deviation of the voxel values in the variation image and dividing each voxel value by the standard deviation. The tissue separation unit 11, the change amount calculation unit 12, the statistical comparison unit 13, the healthy person change amount image correction unit 14, and the normalization unit 15 are all realized by the CPU 1a executing a dedicated program.

<2. Processing action>
<2.1. Reference Embodiment: Simple Difference Evaluation>
FIG. 2 is a flowchart showing an outline of processing of the medical image processing apparatus according to the reference embodiment of the present invention. In FIG. 2, two brain images with different imaging times for the same subject are input, and processing is performed using these two brain images. In FIG. 2, as [Input 1], the brain image at the time of the previous shooting was input as [Input 1], and as [Input 2], the brain image at the time of the shooting after that (most recently) was input. To do. A T1-weighted MRI brain image is input as the brain image. This T1-weighted MRI brain image is a three-dimensional image that is a set of voxels. A voxel is a coordinate unit of an image having “thickness”, and corresponds to a pixel in a two-dimensional image. In steps S1 to S4, the same processing is performed on the brain images input from [Input 1] and [Input 2].

  Next, the medical image processing apparatus performs alignment for correcting the spatial deviation with respect to the input brain image (step S1). Subsequently, the tissue separation unit 11 creates a tissue image in which a predetermined tissue is extracted from the brain image after alignment by tissue separation processing (step S2). As a tissue image, gray matter may be extracted to create a gray matter image, or white matter may be extracted to create a white matter image. In the processing after step S3, an example in which a gray matter image is used will be described. However, a white matter image can also be used.

  Next, the medical image processing apparatus performs spatial standardization that applies the DARTEL algorithm described in Non-Patent Document 3 to the gray matter image created as the tissue image (step S3) and standardization. The gray matter image thus smoothed is smoothed (step S4). As a result, a smooth gray matter image (tissue image) is obtained. The gray matter image is expressed as a three-dimensional image in which the voxels of the gray matter portion have a high value. Each process of step S1-S4 is implement | achieved by the well-known technique described in the patent document 2 and the nonpatent literatures 2-4. Since the value of each voxel becomes the existence probability of the tissue by the tissue separation process in step S2, the value of the voxel after the smoothing process in step S4 is a value in the range of 0.0 to 1.0 at this time. It becomes.

  When the brain images input from [Input 1] and [Input 2] are processed in steps S1 to S4 to obtain a smooth gray matter image, the change amount calculation means 12 then The amount of change between the two gray matter images (tissue images) is calculated (step S5). Specifically, a difference between corresponding voxels of both images is calculated. As a result, a change amount image is obtained. Then, the obtained change amount image is displayed in different colors according to the value of each voxel. In the color classification, the voxel values are divided into predetermined ranges and corresponded to the respective voxel values, but it is preferable to express a large change portion with a prominent color such as “red”. At this time, a part map that is an image specifying a part in the brain region may be displayed in an overlapping manner. The region map is a three-dimensional image that records in which region in the brain region each voxel is included. When the part maps are overlapped, it is possible to grasp at a glance which part is a part having a large change amount.

  FIG. 3 is a diagram showing how the image changes due to the processing in step S5. FIGS. 3A and 3B are gray matter images at the time of pre-photographing and post-photographing before the processing of step S5, respectively. FIG. 3C is a diagram showing a state in which the part map is superimposed on the change amount image after the process of step S5.

  In addition, when displaying using a change amount image, numerical display may be performed. In this case, the average value of the change amount for each voxel is calculated for each region, and this average value is displayed as the change amount of the region. FIG. 4 is a diagram illustrating an example in which the amount of change is displayed by a numerical value for each part. Although the reference of the numerical value to be displayed can be set as appropriate, in the present embodiment, as described above, it is obtained by the difference between the values in the range of 0.0 to 1.0. When the amount of change is larger than the normal value, the numerical value is highlighted. The example of FIG. 4 shows that the amount of change between the hippocampus and the cerebellum is larger than the normal value.

  Moreover, the normal value in FIG. 4 calculates the average value of the amount of change for each voxel for each region, similar to the amount of change image of the subject, with respect to the amount of change image of the healthy person stored in the amount of change image database D1 of healthy people The average value is a normal value of the amount of change in the part. The healthy person change amount image stored in the healthy person change amount image database D1 is obtained by executing the same process as the process of steps S1 to S5 on the two brain images of the healthy person. . About a normal value, it is preferable to output the average value using data of not only one healthy person but also as many healthy persons as possible. In addition, it is preferable to select an equivalent time interval between the pre-photographing time and the post-photographing time of each healthy person as much as possible as the time interval between the pre-photographing time and the post-photographing time of the subject. Since the time interval is taken at the time of diagnosis, it is usually from a monthly unit to a year unit.

<2.2. Second Embodiment: Statistics Evaluation>
FIG. 5 is a flowchart showing an outline of processing of the medical image processing apparatus according to the second embodiment. In FIG. 5, as in FIG. 2, two brain images of the same subject with different imaging times are input, and processing is performed using these two brain images. In steps S11 to S14, the same processing is performed on the brain images input from [Input 1] and [Input 2].

  The processes in steps S11 to S15 in FIG. 5 are the same as the processes in steps S1 to S5 in FIG. Therefore, a gray matter change amount image is obtained as a result of the change amount calculation processing in step S15.

  When the change amount image is obtained, a statistical comparison with the healthy person change amount image in the healthy person change amount image database is performed (step S16). In the present embodiment, a comparison test is performed between the variation image of the subject that has been standardized through the processes of steps S11 to S15 and the healthy subject variation image group stored in the healthy subject variation image database D1. It is desirable to select a healthy person variation image group to be used that is close to the age of the subject.

  Specifically, as shown in the image of FIG. 6, a comparison test of 1: N (N is the total number of healthy subject change amount images) is performed on a voxel basis with such a healthy subject change amount image group, and statistically performed. Detect voxels with significant differences (presumed to be abnormal). In FIG. 6, “year X” is a year (for example, 2005) indicating the time of previous photographing of a healthy person, and “X + a year” is a year (for example, 2010) indicating the time of subsequent photographing of a healthy person. is there. In addition, “Y year” is a year (for example, 2007) indicating the time of previous photographing of a healthy person, and “Y + a year” is a year (for example, 2012) indicating the time of subsequent photographing of a healthy person. The time interval is “a year” in both cases, and is the same.

First, for all voxels, Z scores expressed by the following equations are calculated.

  Thus, the Z score is a value obtained by scaling the difference between the voxel value of the variation image of the subject and the average voxel value of the corresponding voxel of the healthy subject variation image group by the standard deviation, which is gray matter It shows the degree of relative decrease in volume.

Next, an appropriate critical value Z ′ is determined, and the Z score is Z ′ <Z (2)
The voxels that satisfy the above are obtained, and the voxels exhibiting a statistically significant difference are obtained. As the critical value, Z ′ = 2 that can be estimated to be abnormal with a probability of about 95% or more is used. In addition, the following formula is also used as a method for designating a critical value including all sites whose volume is lower than that of a healthy person.
0 <Z '(3)

  Then, voxels having a Z score that is estimated to be abnormal are color-coded with other voxels and displayed as a Z score map. Similar to Method 1, it is preferable to express a voxel having a Z score that is estimated to be abnormal by a prominent color such as “red”. Further, as in the method 1, the region maps may be displayed in an overlapping manner.

  In the present embodiment, analysis by ROI may be further performed.

  This is a method of setting a region of interest (ROI) of a predetermined size on an image when determining the presence or absence of an abnormality using a brain image (see, for example, Non-Patent Document 4) On the brain image, a ROI having a predetermined size is set at a specific part that is attracting attention as related to a specific disease, and comparison is performed.

  In this analysis method, as described in Patent Document 1, an ROI corresponding to a disease is obtained for a voxel at a coordinate position where a significant difference is found from a healthy person by the statistical processing and its Z score (evaluation value). By applying (disease-specific ROI), the degree of affliction is obtained. The features are the following two points.

  (1) Prepare ROI (disease specific ROI) as standardized image data corresponding to each disease such as Alzheimer's, and for each disease that can be considered from the symptoms of the subject, Apply (set), and use the highest significance based on the Z score in the ROI as the diagnosis result.

  (2) Not only is the disease determined based on the Z score of the ROI portion alone, but also the Z score map of the entire brain when the ROI is not applied is compared with the Z score map of only the portion to which the ROI is applied. The purpose is to see the ratio of the atrophy of the site of interest to the atrophy of the entire brain.

  Here, first, as shown in the image of FIG. 7, it is determined whether or not the subject suffers from a certain disease A, taking as an example the case where a specific ROI for each of the diseases A to C is prepared. A method will be described. Each ROI applied to this method will be described later.

Using the ROI corresponding to the disease A, the following five parameters are obtained using the above equations (2) and (3) with respect to the Z score map of the subject obtained by the statistical comparison in step S16. calculate.
P1 = Sum of Z scores of voxels satisfying formula (3) in the ROI portion / number of voxels satisfying formula (3) in the ROI portion P2 = number of voxels satisfying formula (2) in the entire brain / number of voxels in the entire brain P3 = number of voxels satisfying the formula (2) in the ROI portion / number of voxels in the ROI portion P4 = P3 / P2
P5 = Maximum Z score among all voxels in the ROI part

  For the five parameters P1 to P5, characteristics in a patient group having the disease A are obtained in advance, and when the parameter value of the subject matches that, the subject is determined to be the disease A.

  For example, a threshold value (pathological condition identification value) that is regarded as disease A is defined for five parameters, and the subject is assumed to be disease A when the parameter value obtained from the change amount image of the subject exceeds the threshold value. . That is, when the thresholds for identifying the pathological conditions of P1 to P4 are set to thP1 to thP5, respectively, when at least one of P1> thP1, P2> thP2, P3> thP3, P4> thP4, P5> thP5 is satisfied. Is determined as disease A. Specifically, for example, a case where the determination is made by paying attention to only one parameter, such as P1, or a case where the determination is made by referring to some or all of P1 to P5 as necessary can be cited as examples.

In addition to the parameters P1 to P5 listed here, for the five parameters, values limited to the right and left hemispheres may be obtained and added as parameters. Furthermore, the right / left ratio obtained by the equation (4) or the right / left difference obtained by the equation (5) may be added to the parameter with respect to the values on the right hemisphere side and the left hemisphere side.
Right / left ratio = (R−L) / (R + L) * 200 (4)
Left-right difference = R−L (5)
However, the value on the right hemisphere side is R, and the value on the left hemisphere side is L.

  Next, a method for creating an ROI (disease specific ROI) set for each disease will be described.

  The ROI is determined based on statistical processing as follows. For example, in order to determine the ROI of a specific disease A, as shown in the image of FIG. 8, a group of MRI images of a patient with disease A (a sick person variation image group in the sick patient variation image database D2) 2 samples t-test for statistically testing the significant difference between the two groups in units of voxels with respect to the other person's image group (the healthy person variation image group in the healthy person variation image database D1). Ask for it. Voxels that are significantly different by this test are regarded as characteristic voxels in the disease, and the set of coordinates is set as the ROI corresponding to the disease.

<2.3. Third Embodiment: Normalization by Whole Brain Standard Deviation>
FIG. 9 is a flowchart illustrating an outline of processing of the medical image processing apparatus according to the third embodiment. Also in FIG. 9, as in FIGS. 2 and 5, two brain images with different imaging times for the same subject are input, and processing is performed using these two brain images. In steps S21 to S24, the same processing is performed on the brain images input from [Input 1] and [Input 2].

  The processes in steps S21 to S25 in FIG. 9 are the same as the processes in steps S1 to S5 in FIG. 2 and steps S11 to S15 in FIG. Therefore, a change amount image is obtained as a result of the change amount calculation process in step S25.

When the change amount image is obtained, the obtained change amount image is normalized by standard deviation (step S26). Specifically, first, the standard deviation of all voxels of the obtained variation image is calculated. Then, the value of each voxel is normalized by dividing the value of each voxel of the variation image by the calculated standard deviation. Then, the obtained change amount image after normalization is displayed by being color-coded according to the value of each voxel. As in the methods 1 and 2, it is preferable to express a portion having a large change with a conspicuous color such as “red”. Further, as in the methods 1 and 2, the region maps may be displayed in an overlapping manner. In the third embodiment, since normalization is performed using standard deviations, it is possible to reduce the influence between models as compared to using a simple change amount value as in the reference embodiment.

<2.4. Correction of healthy person variation image>
In 2nd Embodiment, the healthy person variation | change_quantity image memorize | stored in the healthy person variation | change_quantity image database D1 was utilized. The elapsed time (elapsed years, months, etc.) at the time of the previous shooting and the subsequent shooting that is the basis of the change image of the healthy person is equal to the time period of the previous shooting and the time of the subsequent shooting that is the basis of the change image. In this case, it is possible to perform an appropriate comparison even if the normal person change amount image stored in the normal person change amount image database D1 is used as it is. However, since the amount of change greatly depends on the elapsed time, the elapsed time (elapsed years, etc.) at the time of the previous shooting and the time of the subsequent shooting that is the basis of the change image of the healthy person is the basis of the change image. If the elapsed time between the previous shooting and the later shooting is not equal, appropriate comparison cannot be performed. Therefore, when the elapsed periods of the two are not equal, the healthy person change amount image is corrected.

  In this case, each healthy person change amount image in the healthy person change amount image database D1 is stored in association with the elapsed period. Then, at the time of inputting the brain image of the subject, the elapsed period at the time of the previous photographing and the time of the subsequent photographing is inputted from the input unit 1d. At this time, the date and time of the previous shooting and the date and time of the subsequent shooting may be input from the input unit 1d, and the CPU 1a may obtain the elapsed period. When the elapsed period of the subject's brain image is obtained, the healthy person change amount image correction means 14 obtains the change amount of each voxel of the normal person change amount image per reference period (for example, one year), and linearly changes in the elapsed period. The value of each voxel in the healthy subject change amount image is corrected to a value obtained by multiplying the change amount per reference period by the elapsed period / reference period. For example, when the elapsed period of the subject's brain image is 5 years and the elapsed time of the healthy person variation image is 2 years, the value of each voxel of the healthy person variation image is corrected to 2.5 times. .

<3. Modified example>
Although the present invention has been specifically described above, the present invention is not limited to the above embodiment, and various modifications can be made. For example, although the first to third embodiments have been described as embodiments, the elements included in the first to third embodiments may be appropriately combined. For example, the healthy person change amount image stored in the healthy person change amount image database D1 is also normalized by the standard deviation and the change amount image normalized by the standard deviation in the third embodiment is used. Statistical comparison may be performed as in the second embodiment.

  In the second embodiment, the test method using the Z score as the evaluation value is shown. However, the present invention is not limited to this, and a t score or the like used in other general tests may be used.

In the reference embodiment, diagnosis can be further supported by using an atrophy degree image in which the degree of atrophy of the brain is recorded. In this case, the procedure is as shown in FIG. As described in the above-described reference embodiment, a change amount image is obtained from the brain image at the time of the previous photographing input as [Input 1] and the brain image at the time of the subsequent photographing input as [Input 2]. On the other hand, the degree of atrophy is calculated from the brain image at the time of imaging after being input as [Input 2] by the method described in Patent Document 2, and an atrophy degree image in which the value of each voxel is replaced with the degree of atrophy is obtained. Then, a threshold value is set for each of the change amount image and the atrophy degree image, and voxels having values larger than the threshold value are extracted (threshold processing). Then, only the voxels extracted in both the change amount image and the atrophy degree image are extracted, and the voxel is emphasized and expressed with a conspicuous color such as “red”. At this time, analysis by ROI as shown in the second embodiment may be further performed.

1a ... CPU
1b ... RAM
DESCRIPTION OF SYMBOLS 1c ... Memory | storage device 1d ... Input part 1e ... Data input / output I / F
1f: Display unit 11: Tissue separation means 12 ... Change amount calculation means 13 ... Statistical comparison means 14 ... Healthy person change amount image correction means 15 ... Normalization means D1 ... Healthy person variation image database D2 ... Disease patient variation image database

Claims (3)

Changes in healthy subjects are stored in association with the elapsed period after performing tissue separation processing on two brain images of the same healthy subject with different imaging times and then representing the amount of change in both images. A quantity image database;
Means for inputting information for specifying the elapsed period between two brain images of the subject;
Tissue separation means for obtaining a tissue image by performing tissue separation processing on each of two brain images having different imaging times of the same subject;
Between the obtained two tissue images, a change amount calculating means for performing a difference calculation between the voxels constituting the tissue image and obtaining a change amount image;
Healthy person variation image correction means for correcting the value of each voxel of the healthy person variation image using the elapsed period associated with the healthy person variation image and the elapsed period between the brain images of the subject. ,
The change amount image obtained by the change amount calculating means and the healthy person change amount image corrected by the healthy person change amount image are statistically compared to calculate a predetermined evaluation value, and the change amount image Statistical comparison means for obtaining a statistically processed image in which the value of each voxel is replaced with the calculated predetermined evaluation value;
Display means for displaying the statistically processed image;
A medical image processing apparatus comprising: The tissue separating means according to claim 1, wherein the tissue separation processing performs processing of separating the white matter or gray matter brain image as is to obtain a white matter image or gray matter image as tissue image Medical image processing apparatus. As medical image processing apparatus according to claim 1 or claim 2, a program for causing a computer to function.