JP6399487B2 - Information processing apparatus, method executed by information processing apparatus, and computer program - Google Patents

Description

  The present invention relates to a technique for extracting a white matter lesion site from a FLAIR image photographed by an MRI apparatus.

  As analysis software for MRI images, SPM (Statistical Parametric Mapping), which is general-purpose software, is widely used.

SPM (Statistical Parametric Mapping), University of London, http://www.fil.ion.ucl.ac.uk/spm/

  Incidentally, in recent years, the relationship between white matter lesions and Alzheimer's disease has been pointed out, but white matter lesions have not been accurately extracted from MRI images.

  An object of the present invention is to easily and accurately extract a white matter lesion site from a FLAIR image.

  A computer program according to an embodiment of the present invention is a computer for causing an information processing apparatus having a storage unit to store an MRI image of a subject's head and a mask image for excluding a ventricle in a standard brain to be executed. A program comprising: separating a first region including a ventricle and white matter from the MRI image of the subject; obtaining a conversion function for converting the MRI image of the subject into a standard brain; Converting a mask image for excluding a ventricle in the standard brain into a mask image for excluding a ventricle in the individual brain of the subject using an inverse function of the function; and the first region and Extracting a second region including white matter excluding the ventricle from the FLAIR image of the subject based on the mask image of the subject; From said second region, to execute the steps of extracting a feature region, to the information processing apparatus.

  A computer program according to another embodiment of the present invention includes an MRI image of a subject's head, a first mask image indicating an entire brain region in a standard brain, and an edge of the entire brain region in the standard brain that is substantially circular. A computer program for causing an information processing apparatus having a storage unit to store a second mask image corresponding to an area to execute the first area including white matter from the first image that is an MRI image of the subject A first mask image showing an entire brain region in the standard brain using an inverse function of the conversion function, a step of obtaining a conversion function for converting the first image into a standard brain And a second mask image corresponding to an edge region of the whole brain region in the standard brain, a first mask image indicating the whole brain region in the individual brain of the subject, and the front The MRI image of the subject based on the step of converting into a second mask image corresponding to an edge region of the whole brain region in the subject's individual brain, and the first region and the first mask image of the subject. Extracting a second region containing white matter from the second image; extracting a feature region from the second region; and a brain of the first mask image of the subject from the second image. Extracting a region that overlaps the entire region, and extracting a region inside the second image of the subject from the overlapping region of the second image as a third region containing white matter; and And extracting the feature region from the third region.

  According to the present invention, a white matter lesion site can be easily and accurately extracted from a FLAIR image.

1 shows an overall configuration of an information processing apparatus 1 according to a first embodiment of the present invention. The state which isolate | separated the 1st area | region 51 containing a ventricle and white matter from the T2 image 50 after alignment is shown. Based on the standard brain template 60, the T2 image 50 after alignment is converted into a T2 image 61 after anatomical standardization. A state in which a mask image 70 set on the standard brain is converted into a mask image 80 of an individual brain is shown. The state which extracted the 2nd area | region 91 containing white matter from the FLAIR image 90 is shown. An explanatory diagram for extracting a feature region from the second region 91 and calculating the volume of the feature region is shown. An image 100 based on the second region data and the image 101 obtained by superimposing the image based on the second region data of the FLAIR image data and the SPECT image based on the SPECT image data are displayed on the window W2. It is a flowchart which shows the process which extracts the characteristic area | region from a FLAIR image. It is a flowchart which shows the process which calculates the volume of a characteristic area. It is a flowchart which shows a display process. 2 shows an overall configuration of an information processing apparatus 102 according to a second embodiment of the present invention. (A) A state in which the first mask image 170 set on the standard brain is converted into a first mask image 180 of the personal brain, (b) a second mask image 172 set on the standard brain is converted into the personal brain. (C) shows a state in which the z-score image 174 set on the standard brain is converted into a z-score image 184 of the individual brain. The state which extracted the 2nd area | region 92 containing white matter from the FLAIR image 90 is shown. The state which extracted the 3rd area | region 94 containing white matter from the FLAIR image 90 is shown. An explanatory view for setting the boundary and calculating the volume of the feature region on the front corner side and the rear corner side is shown. It is a flowchart which shows the process which extracts the characteristic area | region from a FLAIR image. It is a flowchart which shows the process which calculates the volume of a characteristic area. It is a flowchart which shows the process which calculates a z-score parameter | index.

  Hereinafter, an information processing apparatus and a computer program executed by the information processing apparatus according to the first embodiment of the present invention will be described with reference to the drawings.

  The information processing apparatus according to the present embodiment performs processing for extracting a feature region from a FLAIR image photographed by the MRI apparatus.

  FIG. 1 shows an overall configuration of an information processing apparatus 1 according to the present embodiment. As shown in the figure, in the information processing apparatus 1, one or more of an input device 2, an output device 3, and an input / output device 4 are connected to the information processing apparatus main body 10. The information processing apparatus body 10 is configured by, for example, a general-purpose computer system including a processor and a memory, and each component or function in the information processing apparatus body 10 described below executes, for example, a computer program Is realized. This computer program can be stored in a computer-readable recording medium. The input device 2 may be, for example, a keyboard or a pointing device. The output device 3 may be, for example, a display device or a printer. The input / output device 4 may be, for example, a mass storage device or a network interface device.

  The information processing apparatus main body 10 includes an MRI (Magnetic Resonance Imaging) data storage unit 11, a SPECT (Single Photon Emission Computed Tomography) data storage unit 13, a registration unit 15, a post-positioning image data storage unit 17, and a separation unit. Processing unit 19, first region data storage unit 21, standard brain data storage unit 23, conversion parameter calculation unit 25, conversion parameter storage unit 27, mask image conversion unit 29, and personal brain mask image data storage Unit 31, white matter extraction unit 33, second region data storage unit 35, feature region extraction unit 37, feature region data storage unit 39, volume calculation unit 41, volume data storage unit 43, and display processing Part 45.

  The MRI data storage unit 11 stores, as MRI data, image data of a plurality of types of MRI images obtained by photographing the head of the same subject using different methods. For example, each image data (T1 image data, T2 image data, FLAIR image data) of a T1 image, a T2 image, and a FLAIR image is stored. The T1 image is an image in which fat is extracted with a high signal (white) and water is extracted with a low signal (black). The T2 image is an image in which water is extracted with a high signal (white) and fat is extracted with a low signal (black), and is useful for extracting lesions. The FLAIR image is a T2 image in which a water signal is suppressed, and is useful for extracting a lesion adjacent to the ventricle. The MRI data storage unit 11 stores MRI data of one or more subjects.

  T1 image data, T2 image data, and FLAIR image data are three-dimensional data each composed of a plurality of tomographic image data. In the present embodiment, the matrix size (number of voxels) in T1 image data and T2 image data is 720 × 720 × 22 (22 tomographic images of 720 × 720 pixels), and the pixel size is 0.31 × 0.31 ×. The matrix size in the FLAIR image data is 384 × 384 × 155 (155 tomographic images of 384 × 384 pixels), and the pixel size is 0.55 × 0.55 × 1.00 mm. The pixel values of T1 image data, T2 image data, and FLAIR image data are not absolute values but relative values, and the numerical values depend on MRI image acquisition conditions and the type of camera.

  The SPECT data storage unit 13 stores, as SPECT data, SPECT image data obtained by imaging the same subject's head as the MRI data stored in the MRI data storage unit 11 with a SPECT imaging apparatus. The SPECT image data is three-dimensional data composed of a plurality of tomographic image data. Further, the MRI data and SPECT data may be acquired, for example, via the input device 2 or the input / output device 4, or may be directly acquired from the MRI device and the SPECT imaging device.

  The alignment unit 15 acquires FLAIR image data and T2 image data of the same person stored in the MRI data storage unit 11, and performs alignment between the FLAIR image data and the T2 image data. For example, the matrix size, pixel size, and position information of FLAIR image data and T2 image data are matched by linear transformation. For example, the matrix size, pixel size, and position information of T2 image data are matched with the matrix size, pixel size, and position information of FLAIR image data. As a result, the matrix size of the T2 image data is converted to 384 × 384 × 155 and the pixel size is converted to 0.55 × 0.55 × 1.00 mm. Thereby, FLAIR image data and T2 image data can be handled with the same coordinates. Note that T1 image data may be obtained instead of T2 image data, and the matrix size and pixel size of T1 image data may be made to match the matrix size and pixel size of FLAIR image data. The alignment process can be executed by a co-registration function of SPM (Statistical Parametric Mapping), which is general-purpose software for MRI image analysis.

  The alignment unit 15 also aligns the SPECT image data stored in the SPECT data storage unit 13 with the FLAIR image data in the same manner as the T2 image data by the above SPM. As a result, the matrix size of the SPECT image is converted to 384 × 384 × 155 and the pixel size is converted to 0.55 × 0.55 × 1.00 mm.

  The post-alignment image data storage unit 17 stores the post-positioning T2 image data and SPECT image data converted by the alignment unit 15.

  The separation processing unit 19 separates first region data including the ventricle and white matter from the T2 image data after alignment. For example, as shown in FIG. 2, the first region 51 including the ventricle and white matter is separated from the T2 image 50 after alignment. This separation process can be executed by the SPM segmentation function described above. Note that with the SPM segmentation function described above, it is not possible to separate only the white matter from the T2 image data after alignment, and the separated data includes a partial region of the ventricle. Therefore, the separation processing unit 19 separates the first region data including the ventricle and the white matter from the T2 image data after the alignment.

  Further, when white matter is separated from the aligned T2 image 50 by SPM, gray matter is included in the region separated as white matter. Therefore, the process of making the separated area completely white matter may be performed. For example, the pixel value of the boundary between white matter and gray matter is set, and the pixel value of the first region data is binarized based on the set pixel value, thereby removing the gray matter from the region separated as the white matter. The first area 51 shown in FIG. 2 is a white matter image excluding the binarized gray matter.

  Returning to FIG. 1, the first area data storage unit 21 stores first area data corresponding to the first area 51 separated by the separation processing unit 19.

  The standard brain data storage unit 23 stores standard brain template data and mask image data that is set on the standard brain and excludes the ventricle. Standard brain template data and mask image data are acquired via the input device 2 or the input / output device 4, for example.

  As shown in FIG. 3, the conversion parameter calculation unit 25 converts conversion parameters (conversion functions) for converting the T2 image 50 after alignment into the T2 image 61 after anatomical standardization based on the standard brain template 60. calculate. Specifically, the conversion parameters for converting the coordinates (individual brain coordinates) of the T2 image data after alignment into standard brain coordinates are calculated by linear conversion and non-linear conversion. This conversion parameter calculation can be performed by the above-described SPM normalization function.

  The conversion parameter storage unit 27 stores the conversion parameter calculated by the conversion parameter calculation unit 25.

  As shown in FIG. 4, the standard brain mask image 70 based on the mask image data has a plurality (five) of ventricular regions 71 and brain regions 72 other than the ventricles. In the mask image 70 shown in FIG. 4, the number of ventricles is five, but the number of ventricles increases or decreases in another cross section.

  The mask image conversion unit 29 converts the mask image data of the standard brain into mask image data corresponding to the shape of the individual brain of the subject using the reciprocal (inverse function) of the calculated conversion parameter. That is, the mask image 70 for excluding the ventricle in the standard brain is converted into the mask image 80 for excluding the ventricle in the subject's individual brain. Thereby, a mask image 80 based on the personal brain coordinates as shown in FIG. 4 is created. The personal brain mask image 80 has a plurality (five) of ventricular regions 81 and a brain region 82 other than the ventricles.

  Returning to FIG. 1, the personal brain mask image data storage unit 31 stores the mask image data of the personal brain converted by the mask image conversion unit 29.

  The white matter extraction unit 33 uses the first region data and the personal brain mask image data to extract second region data including white matter from the FLAIR image data. As shown in FIG. 5, in the FLAIR image 90, a portion where both the first region 51 of the T2 image and the brain region 82 other than the ventricle of the mask image 80 of the personal brain overlap with the white matter. 91 is extracted. Therefore, in the extracted second region 91, the ventricle region 81 is excluded.

  The second area data storage unit 35 stores the second area data extracted by the white matter extraction unit 33.

  The feature region extraction unit 37 extracts pixels having pixel values whose pixel values belong to a predetermined value range from the extracted second region 91 as a feature region. For example, a pixel having a pixel value larger than a predetermined threshold is set as the feature region. In the FLAIR image, the white matter lesion area has a larger pixel value than the normal white matter area. Therefore, it is possible to extract a white matter lesion area as a feature area by setting an appropriate threshold.

  FIG. 6 is an explanatory diagram for extracting a feature region from the second region 91 and calculating the volume of the feature region. In the window W1 in FIG. 6B, the pixel value to be extracted is 1600, the allowable value is 700 (± 350 centering on the reference value), and pixel values of 1250 to 1950 are extracted as feature regions. In FIG. 6A, a white portion indicated by an arrow is extracted as a feature region.

  The feature region data storage unit 39 stores the feature region (coordinates and pixel values) extracted by the feature region extraction unit 37 as feature region data.

  Note that the above processing in the alignment unit 15, the separation processing unit 19, the conversion parameter calculation unit 25, the mask image conversion unit 29, the white matter extraction unit 33, and the feature region extraction unit 37 is performed on all the tomographic images of the processing target image data. The processing is executed for image data, and each of the storage units 17, 21, 27, 23, 35, and 39 stores a processing result for each tomographic image data.

  The volume calculation unit 41 calculates the volume of the feature region. For example, the number of pixels constituting the feature area data is obtained for each tomographic image data, and the volume of the feature area is calculated by multiplying the total number of pixels by the pixel size. As shown in the underlined portion of the Result column of the window W1 in FIG. 6B, when the number of pixels (Pixel) of the feature region (WM (D)) in all tomographic image data is 2291, the pixel size: By multiplying by 0.55 × 0.55 × 1.00 mm, the volume (0.7 cc) of the feature region is calculated. Note that the ratio of the feature region to the entire head or the entire white matter region may be calculated based on the calculated volume of the feature region.

  The volume data storage unit 43 stores the volume of the feature area calculated by the volume calculation unit 41.

  FIG. 7 shows a state in which an image 100 based on the second region data and an image 101 on which the image based on the second region data of FLAIR image data and the SPECT image based on the SPECT image data are superimposed are displayed in the window W2. ing.

  The display processing unit 45 displays the image 100 based on the second region data on the display device (output device 3) as shown in the window W2 in FIG. 7, and changes the color of the feature region in the second region data. The display mode is different from that of the area. In FIG. 7, as indicated by the arrows, the feature regions are all displayed in white in order to emphasize more. Further, as shown in FIG. 7, the display processing unit 45 displays an image 101 in which an image based on the second area data of FLAIR image data and a SPECT image based on the SPECT image data are overlaid on the display device (output device 3). May be displayed. At this time, in the image based on the second region data displayed superimposed on the SPECT image, the feature region may be displayed in a display mode different from other regions.

  The information processing apparatus 1 according to the present embodiment having the above-described configuration performs a process of extracting a feature region from a FLAIR image photographed by the MRI apparatus according to the procedure described below.

  FIG. 8 is a flowchart showing a procedure for extracting a feature region from a FLAIR image photographed by MRI.

  First, MRI data (FLAIR image data, T1 image data, T2 image data) is stored in the MRI data storage unit 11 (S10), and SPECT data of the same subject as the stored MRI data is stored in the SPECT data storage unit 13 ( S12). The MRI data and SPECT data may be acquired from the outside via the input / output device 4. When the data is already prepared, this step may be omitted.

  The alignment unit 15 performs alignment between the FLAIR image data stored in step S10 and the T2 image data, and stores the aligned T2 image data in the post-alignment image data storage unit 17 (S14). Similarly, the registration unit 15 performs registration between the SPECT image data stored in step S12 and the FLAIR image data, and stores the SPECT image data after registration in the post-alignment image data storage unit 17 (S14). .

  The separation processing unit 19 separates the first region data including the ventricle and the white matter from the T2 image data after the alignment performed in step S14, and the first region data is converted into the first region data. It preserve | saves at the memory | storage part 21 (S16).

  Next, standard brain data including standard brain template data and mask image data is stored in the standard brain data storage unit 23 (S18). The standard brain data may be acquired from the outside via the input / output device 4. When the standard brain data has already been prepared, this step may be omitted, and this step may be executed any time before the next step S20 is executed.

  The conversion parameter calculation unit 25 obtains a conversion parameter (conversion function) for converting the T2 image data after the alignment performed in step S16 into a standard brain template, and the obtained conversion parameter is converted into a conversion parameter storage unit. 27 (S20).

  The mask image conversion unit 29 converts the coordinates of the mask image of the standard brain stored in step S18 into the personal brain coordinates using the reciprocal number (inverse function) of the conversion parameter calculated in step S20. Mask image data corresponding to the shape is stored in the personal brain mask image data storage unit 31 (S22).

  Based on the first region data separated in step S16 and the personal brain mask image data created in step S22, the white matter extraction unit 33 generates a second region containing white matter from the FLAIR image data stored in S10. Data is extracted, and the extracted second region data is stored in the second region data storage unit 35 (S24).

  The feature region extraction unit 37 extracts feature region data from the second region data extracted in step S24 and stores it in the feature region data storage unit 39 (S26). Thereby, the process of extracting the feature region is completed.

  Next, processing for calculating the volume of the feature region in the entire subject's head in the information processing apparatus 1 will be described. The processing for calculating the volume of the feature region is executed after the processing for extracting the feature region is executed. FIG. 9 is a flowchart illustrating processing for calculating the volume of the feature region.

  First, the volume calculation unit 41 acquires feature region data from the feature region data storage unit 39 (S30).

  The volume calculation unit 41 obtains the number of pixels constituting the feature region data for each tomographic image data, calculates the volume of the feature region by multiplying the obtained total number of pixels by the pixel size, and stores it in the volume data storage unit. (S32). Thereby, the process of calculating the volume of the feature region is completed. Note that the volume of the calculated feature region may be output by the output device 3.

  Next, display processing in the information processing apparatus 1 will be described. This display process is executed after executing the process of extracting the feature region. FIG. 10 is a flowchart showing the display process.

  First, the display processing unit 45 stores the SPECT image data after alignment from the post-alignment image data storage unit 21, the second region data from the second region data storage unit 35, and the feature region data as feature region data. Each is acquired from the unit 39 (S40).

  As shown in FIG. 7, the display processing unit 45 causes the display device (output device 3) to display an image based on the second region data acquired in S40. At this time, the color of the characteristic area in the second area data is displayed in a display mode different from the other areas. Further, as shown in FIG. 7, the display processing unit 45 superimposes the image based on the second area data of the FLAIR image data and the SPECT image based on the SPECT image data, and displays them on the display device (output device 3). (S41).

  According to the present embodiment, the first region including the ventricle and white matter is separated from the T2 image of the subject, the conversion parameter (conversion function) for converting the T2 image into the standard brain template is obtained, and the conversion parameter is obtained. The mask image data in which the ventricular region is defined in the standard brain is converted into mask image data in which the ventricular region is defined in the subject's individual brain using the inverse (inverse function) of the first region and the subject. Based on the mask image, a second region including white matter excluding the ventricle is extracted from the subject FLAIR image, and a white matter lesion that is a feature region is extracted from the second region. Thereby, a feature region (white matter lesion) can be easily and accurately extracted from the FLAIR image. Further, since the second region including the white matter excluding the ventricle is extracted, the feature region (white matter lesion) adjacent to the ventricle can be extracted clearly.

  Further, since the volume of the feature region (white matter lesion) of the entire head of the subject is calculated, the volume of the feature region (white matter lesion) with respect to the entire head or white matter can be grasped.

  Further, with the MRI apparatus, the T2 image can be taken with a higher spatial resolution than the FLAIR image. Then, based on the first region of the T2 image, a second region including white matter is extracted from the FLAIR image, and a feature region is extracted from the second region. Therefore, it is possible to improve the feature region extraction accuracy.

  In addition, since the white matter lesion as the feature region and the SPECT image of the subject are displayed in a superimposed manner, the white matter lesion site and the position of the blood flow reduction can be compared.

  An information processing apparatus according to a second embodiment of the present invention and a computer program executed by the information processing apparatus will be described with reference to the drawings. In addition, about the same member and process as 1st Embodiment, the same number is attached | subjected and description is abbreviate | omitted and only a different part is demonstrated.

  Similar to the information processing apparatus according to the first embodiment, the information processing apparatus according to the present embodiment performs a process of extracting a feature region from a FLAIR image photographed by the MRI apparatus.

  FIG. 11 shows the overall configuration of the information processing apparatus 102 according to the present embodiment. As shown in the figure, in the information processing apparatus 102, one or more of an input device 2, an output device 3, and an input / output device 4 are connected to the information processing apparatus main body 110. The information processing apparatus main body 110 is configured by, for example, a general-purpose computer system including a processor and a memory. Individual components or functions in the information processing apparatus main body 110 described below execute, for example, a computer program. Is realized. This computer program can be stored in a computer-readable recording medium.

  In addition to the configuration of the information processing apparatus 10 of the first embodiment, the information processing apparatus main body 110 includes a standard brain z-score data storage unit 12, a personal brain z-score data storage unit 14, a re-extraction region extraction unit 16, a third Region data storage unit 18, feature region re-extraction unit 20, z-score processing unit 22, and processing data storage unit 24. The standard brain data storage unit 23 includes a first mask image data set on the standard brain that represents the entire brain region and excludes the scalp and the skull, and a region that is set on the standard brain and has no white matter lesions. Second mask image data to be excluded is stored. The first mask image data and the second mask image data are obtained via the input device 2 or the input / output device 4, for example.

  As shown in FIG. 12A, the first mask image 170 based on the first mask image data is composed of the entire brain region. The entire brain region (first mask image 170) corresponds to a region obtained by combining the ventricle region 71 and the brain region 72 in FIG. Further, as shown in FIG. 12B, the second mask image 172 based on the second mask image data is constituted by an annular region having a substantially annular shape. The annular region (second mask image 172) corresponds to the edge region of the first mask image 170, which is the entire brain region of FIG. The second mask image 172 is, for example, an area where there is no white matter lesion set based on an average image of several tens of subjects.

  The standard brain z-score data storage unit 12 obtains, as standard brain z-score data, z-score image data obtained based on the SPECT image data stored in the SPECT data storage unit 13 and converted into the standard brain. Remember. The z-score image data is obtained by, for example, comparing the cerebral blood flow of the subject with the normal group and quantifying the decrease in cerebral blood flow using SPM or the like. To get through. Note that z-score is a value calculated based on (average blood flow of normal group−blood flow of subject) / normal standard deviation.

  As shown in FIG. 12 (c), the z-score image 174 is an image based on numerical data of the degree of decrease in cerebral blood flow, and the white area shows a decrease in blood flow relative to the normal group. The black area indicates a relatively high blood flow area.

  Returning to FIG. 11, the mask image conversion unit 29 according to the present embodiment uses the reciprocal (inverse function) of the calculated conversion parameter to set the first mask image data and the second mask set on the standard brain. The image data is converted into first mask image data and second mask image data of the personal brain corresponding to the shape of the subject's personal brain. That is, the first mask image 170 corresponding to the entire brain region in the standard brain is converted into the first mask image 180 corresponding to the entire brain region in the individual brain of the subject, and the second mask image 172 in the standard brain is converted. The second mask image 182 in the subject's individual brain is converted. Thus, a first mask image 180 based on the personal brain coordinates as shown in FIG. 12A and a second mask image 182 based on the personal brain coordinates as shown in FIG. 12B are created.

  Furthermore, the mask image conversion unit 29 uses the inverse number (inverse function) of the calculated conversion parameter to convert the z-score image data converted into the standard brain into a z-score of the personal brain corresponding to the shape of the subject's personal brain. -Convert to score image data. That is, the z-score image 174 based on the standard brain coordinates is converted into a z-score image 184 based on the personal brain coordinates. As a result, a z-score image 184 based on the personal brain coordinates as shown in FIG.

  Returning to FIG. 11, the personal brain mask image data storage unit 31 stores the first mask image data and the second mask image data of the personal brain converted by the mask image conversion unit 29.

  The personal brain z-score image data storage unit 14 stores the z-score image data of the personal brain converted by the mask image conversion unit 29.

  The white matter extraction unit 33 in the present embodiment extracts second region data including white matter from the FLAIR image data using the first region data and the first mask image data of the personal brain. As shown in FIG. 13, a portion of the FLAIR image 90 that overlaps both the first region 51 of the T2 image and the first mask image 180 corresponding to the entire brain region of the individual brain includes the second matter containing white matter. An area 92 is extracted. Then, in the feature region extraction unit 37, as in the first embodiment, pixels having pixel values whose pixel values belong to a predetermined value range are extracted as feature regions from the extracted second region 92.

  The re-extraction region extraction unit 16 extracts third region data including white matter from the FLAIR image data using the first mask image data and the second mask image data of the personal brain. As shown in FIG. 14, a region 93 that overlaps the first mask image 180 corresponding to the entire brain region of the individual brain is extracted from the FLAIR image 90, and the scalp and skull are removed from the FLAIR image 90. Then, of the overlapping region 93, a region that does not overlap with the second mask image 182 is extracted as the third region 94. That is, in the overlapping area 93, the area inside the second mask image 182 having a substantially annular shape is extracted as the third area 94.

  The third area data storage unit 18 stores the third area data extracted by the re-extraction area extraction unit 16.

  Similar to the feature region extraction unit 37, the feature region re-extraction unit 20 extracts pixels having pixel values whose pixel values belong to a predetermined value range from the extracted third region 94 as the feature region. The feature regions extracted here include feature regions already extracted by the feature region extraction unit 37 and other feature regions. That is, in the separation processing unit 19, the first region 51 including the white matter separated from the T2 image 50 tends not to include all the white matter regions of the T2 image 50. There may be a slight presence. Therefore, by extracting the third region 94 from the FLAIR image 90 using the first mask image 180 corresponding to the entire brain region of the individual brain, and reextracting the feature region from the third region 94, all the regions are extracted. A feature region (white matter lesion region) is reliably extracted from the white matter region.

  Of the feature regions (coordinates and pixel values) extracted by the feature region re-extraction unit 20, a difference with respect to the feature region extracted by the feature region extraction unit 37 is stored as feature region data in the feature region data storage unit 39. .

  In addition to calculating the volume (cc) of the feature region (the white part indicated by the arrow in FIG. 15A) described in the first embodiment, the volume calculation unit 41, as shown in FIG. As shown in the Result column of the window W3 in FIG. 15B, the volume of the feature region (WMD) on the front corner side and the rear corner side is set with reference to the boundary set in FIG. 15B (vertical plane passing through the center in this embodiment). (Cc) is calculated. Note that the boundary is not limited to this, and two points on the parietal side and the basilar side may be designated and divided into an anterior angle and a posterior angle based on this, or set in advance on the first mask image 180. The front and rear corners may be divided based on the border (coordinates) placed. The boundary set in advance may be a central groove or a Sylvian groove (outer groove). Moreover, you may divide into right and left instead of front and back. In addition, a boundary between periventricular lesion (PVH) and deep subcortical white matter lesion (DSWMH) is set on the standard brain and converted into a personal brain. Based on the converted boundary, the volume of the feature region may be calculated separately for the periventricular lesion and the deep subcortical white matter lesion.

  The volume calculation unit 41 calculates the ratio of the volume of the feature region (white matter lesion area) to the volume of the entire white matter region, and calculates the ratio of the volume of the front part of the feature region to the volume of the rear part of the feature region. Further, the volume calculation unit 41 calculates the volume of the entire brain region based on the first mask image data corresponding to the entire brain region of the individual brain, and calculates the ratio of the entire white matter region to the entire brain region. The volume of the entire white matter region is calculated based on the first region data shown in the first region 51 and the feature region data extracted by the feature region re-extraction unit 20 and stored in the feature region data storage unit 39.

  The result calculated by the volume calculation unit 41 is stored in the volume data storage unit 43.

  Returning to FIG. 11, the z-score processing unit 22 determines the z-score of a pixel whose z-score value in the feature region is z> 0 based on the z-score image data of the individual brain and the feature region data. The average value, the ratio of voxels having a z-score value of 2 or more in the feature region, the ratio of the blood flow reduction region (z-score value of 2 or more) in the whole brain, and the blood flow reduction region (z− The ratio of the score value is 2 or more) is calculated. Note that the threshold value of the z-score value is not limited to “0” and “2”, and may be other values.

  The processing data storage unit 24 stores the processing result calculated by the z-score processing unit 22.

  The information processing apparatus 102 according to the present embodiment having the above-described configuration performs processing for extracting a feature region from a FLAIR image photographed by the MRI apparatus according to the procedure described below.

  FIG. 16 is a flowchart showing a procedure for extracting a feature region from a FLAIR image taken by MRI in the present embodiment. Note that the same processes as those shown in FIG. 8 are denoted by the same reference numerals, description thereof is omitted, and only different processes are described.

  In step S18 in the present embodiment, standard brain data including standard brain template data, first mask image data, and second mask image data is stored in the standard brain data storage unit 23.

  Next, the z-score image data converted into the standard brain is stored in the standard brain z-score data storage unit 12 (S19). The z-score image data converted into the standard brain may be acquired from the outside via the input / output device 4. When z-score image data converted into the standard brain has already been prepared, this step may be omitted, and this step may be executed any time before step S22 is executed.

  In step S22 in the present embodiment, the mask image conversion unit 29 uses the reciprocal number (inverse function) of the conversion parameter calculated in step S20, and the first and second mask images of the standard brain stored in step S18. The data coordinates are converted into personal brain coordinates, and the first and second mask image data corresponding to the shape of the personal brain are stored in the personal brain mask image data storage unit 31. Further, the mask image conversion unit 29 uses the reciprocal number (inverse function) of the conversion parameter calculated in step S20 to convert the z-score image data converted into the standard brain stored in step S19 into the subject's individual brain. Is converted into z-score image data of the personal brain corresponding to the shape of the human brain and stored in the personal brain z-score image data storage unit 14.

  In step S24 in the present embodiment, the white matter extraction unit 33 is stored in step S10 based on the first region data separated in step S16 and the first mask image data of the personal brain created in step S22. Second area data including white matter is extracted from the FLAIR image data, and the extracted second area data is stored in the second area data storage unit 35.

  The re-extraction region extraction unit 16 uses the first mask image data and the second mask image data of the personal brain created in step S22, and uses the FLAIR image data stored in step S10 to include a third matter including white matter. The area data is extracted, and the extracted third area data is stored in the third area data storage unit 18 (S27).

  The feature region re-extraction unit 20 extracts feature region data from the third region data extracted in step S27, and stores feature region data corresponding to the difference with respect to the feature region data extracted in step S26. 39 (S28).

  Next, processing for calculating the volume of the feature region in the entire head of the subject in the information processing apparatus 102 will be described. The processing for calculating the volume of the feature region is executed after the processing for extracting the feature region is executed. FIG. 17 is a flowchart illustrating processing for calculating the volume of the feature region. Note that the same processes as those shown in FIG. 9 are denoted by the same reference numerals, description thereof is omitted, and only different processes are described.

  In step S32 in the present embodiment, the volume calculation unit 41 calculates the number of pixels constituting the feature area data for each tomographic image data, and calculates the volume of the feature area by multiplying the obtained total number of pixels by the pixel size. And stored in the volume data storage unit 43. The volume calculation unit 41 calculates the volume of the feature region on the front corner side and the rear corner side based on the boundary set by the user or a predetermined boundary, and stores the volume in the volume data storage unit 43. Further, the volume calculation unit 41 calculates the ratio of the volume of the feature region to the volume of the entire white matter region, calculates the ratio of the volume of the front portion of the feature region to the volume of the rear portion of the feature region, and calculates the total white matter with respect to the entire brain region. The ratio of the area is calculated and stored in the volume data storage unit 43. The calculated result may be output by the output device 3.

  In the information processing apparatus 102 of the present embodiment, the display process is performed in the same manner as the display process (FIG. 10) in the information processing apparatus 1 of the first embodiment. In the display process, the volume of the feature area calculated in the above process, the volume of the feature area on the front and rear corners, the ratio of the volume of the feature area to the volume of the entire white matter area, and the volume of the rear part of the feature area The ratio of the volume of the front part of the feature region to the ratio of the whole white matter region to the whole brain region may be displayed.

  Next, processing for calculating various z-score indices based on z-score image data in the information processing apparatus 102 will be described. This processing is executed after executing the processing for extracting the feature region. FIG. 18 is a flowchart illustrating processing for calculating a z-score index.

  The z-score processing unit 22 acquires the feature region data from the feature region data storage unit 39 and the z-score image data of the personal brain from the personal brain z-score image data storage unit 14 (S43).

  The z-score processing unit 22 is based on the z-score image data of the individual brain and the feature region data, and the z-score average value of the pixels whose z-score value in the feature region is z> 0, The ratio of voxels with a z-score value of 2 or more in the region, the ratio of the blood flow reduction region (z-score value of 2 or more) in the whole brain, and the blood flow reduction region in the feature region (z-score value of 2 or more) The ratio is calculated and stored in the processing data storage unit 24 (S44). Thereby, the process of calculating the z-score index is completed. The calculated z-score index may be output by the output device 3. As described above, the threshold value of the z-score value is not limited to “0” and “2”, and may be another value.

  According to the present embodiment, the first region 51 including white matter is separated from the T2 image 50 of the subject. A conversion parameter (conversion function) for converting the T2 image 50 into a standard brain template is obtained, and the first mask image 170 indicating the entire brain region in the standard brain and the standard brain are obtained using an inverse number (inverse function) of the conversion parameter. The second mask image 172 corresponding to the edge region of the whole brain region in FIG. 5 corresponds to the first mask image 180 showing the whole brain region in the subject's individual brain and the edge region of the whole brain region in the subject's individual brain. The second mask image 182 is converted. Then, based on the first region 51 and the first mask image 180 of the subject, a second region 92 including white matter is extracted from the FLAIR image 90 of the subject, and a white matter lesion that is a feature region is extracted from the second region. To do. Further, a region 93 that overlaps the entire brain region of the subject's first mask image 180 is extracted from the subject's FLAIR image 90, and the region inside the subject's second mask image 182 in the overlapping region 93 is white matter. And a white matter lesion which is a feature region is extracted from the third region 94.

  The first region 51 including the white matter separated from the T2 image 50 tends not to include all the white matter regions of the T2 image 50, and there may be a slight white matter region that is not completely separated. . However, the third region 94 is extracted from the FLAIR image 90 using the first mask image 180 and the second mask image 182 corresponding to the entire brain region of the individual brain, and the feature region is re-created from the third region 94. By extracting, a feature region (white matter lesion region) can be reliably extracted from all white matter regions. Therefore, a feature region (white matter lesion) can be easily and accurately extracted from the FLAIR image. In addition, since the region where no white matter lesion is present is deleted from the overlapping region 93 and the feature region is re-extracted from the third region 94 by the second mask image 182 of the subject, the re-extraction can be executed in a short time. it can.

  Next, the results of a disease state analysis of the disease based on the volume of the feature region (white matter lesion region) obtained in the above embodiment will be described.

  Brain compartment and cerebral blood flow for 54 patients (22 males, 32 females, average age 73.6 years ± 84), including new patients who visited a forgetfulness outpatient clinic and those who were admitted for dementia Based on the quantitative analysis, the difference of the disease was compared retrospectively. As a result, among 54 patients, Alzheimer's disease (AD) was 19 patients, Vascular dementia (VaD) was 4 patients, and Lewy Body Dementia (DLB: Dementia with Lewy). (Bodies) was 7 and mild cognitive impairment (MCI) was 12 people.

  Then, the ratio of the volume of the feature area to the volume of the entire white matter area (white matter lesion area / total white matter area) was determined for each disease. As a result, AD was 4.7%, VaD was 9.5%, DLB was 2.4%, and MCI was 1.8%. The ratio of VaD was significantly higher than other diseases, suggesting the possibility of distinguishing VaD from other diseases by determining the white matter lesion area / total white matter area.

  Further, the ratio of the front volume of the feature area to the rear volume of the feature area (the front of the white matter lesion area / the rear part of the white matter lesion area) was obtained with the central groove as a boundary. As a result, AD was 0.84, VaD was 1.03, DLB was 1.03, and MCI was 0.77. AD and MCI showed many posterior white matter lesions centered on the posterior corner of the lateral ventricle, while VaD and DLB showed a tendency to have many anterior white matter lesions centered on the anterior horn of the side ventricle. It was suggested that AD and VaD could be differentiated by determining the front-to-back ratio of white matter lesions.

ここで、表中の「↓」は低下の程度を表しており、「↓」が多くなるほど、低下の程度が大きいことを示している。また、「−」は、血流低下がほとんどないことを示している。また、表１において、ＶａＤは、その５０％をしめる皮質下血管性認知症に限り、ＭＣＩは、将来ＡＤにコンバートする確率が高い傾向にあるａｍｎｅｓｔｉｃ ＭＣＩに限っている。また、ＦＴＬＤ（Ｆｒｏｎｔｏｔｅｍｐｏｒａｌ Ｌｏｂａｒ Ｄｅｇｅｎｅｒａｔｉｏｎ）は、前頭側頭葉変性性認知症である。 In addition, from the SPECT image of the subject, there is a tendency that a decrease in blood flow as shown in Table 1 regarding the cerebral blood flow is observed.

Here, “↓” in the table represents the degree of decrease, and the more “↓”, the greater the degree of decrease. Further, “−” indicates that there is almost no decrease in blood flow. In Table 1, VaD is limited to subcortical vascular dementia, which accounts for 50%, and MCI is limited to amnestic MCI that tends to be converted to AD in the future. Further, FTLD (Frontotemporal Robot Degeneration) is frontotemporal lobar degenerative dementia.

  The relationship between white matter lesions and cerebral blood flow is obtained by superimposing the image based on the third region data of FLAIR image data and the SPECT image based on SPECT image data, and displaying the characteristic region in a display mode different from other regions. Therefore, there is a possibility that pathological analysis of diseases can be performed in a short time. For example, on the basis of the displayed image, there are many white matter lesions in the posterior horn, and when blood flow decrease is observed mainly in the posterior zonal gyrus and the front of the wedge, there are many white matter lesions mainly in the anterior horn and the front horn. VaD when there is no decrease in blood flow centered on the posterior gyrus and the front of the wedge, and there is not much white matter lesions, and when there is a decrease in blood flow inside the occipital lobe, DLB and MCI are mostly white matter lesions. If there is no decrease in blood flow and there is almost no decrease in blood flow, there is a possibility that image diagnosis of MCI can be performed. Although FTLD does not describe image diagnosis, the ratio of the front volume of the feature area to the rear volume of the feature area of FTLD (the front of the white matter lesion area / the rear of the white matter lesion area) is obtained. May cause image diagnosis.

  Moreover, although it is the data of 1 person, the white matter lesion area / whole white matter area | region of the subject who converted to AD about half a year in the subject of MCI is 2%, and the subject who did not convert to AD is 0.3%. The former part of the white matter lesion area / the rear part of the white matter lesion area was 0.47, and the latter was 1.15. Although it is necessary to study with increasing cases, this result suggests the possibility of obtaining the probability of conversion to AD from the volume of white matter lesions and the ratio before and after.

  In the above embodiment, the white matter lesion is extracted as the feature region from the FLAIR image 90. Alternatively, the gray matter (cortex) lesion may be extracted at the same time. That is, a region containing the gray matter (fourth region) is separated from the T2 image 50, and the region containing the gray matter from the FLAIR image 90 based on the separated region and the mask image 80 or the first mask image 180 (first region). 5 region) may be extracted, and a gray matter lesion may be extracted as a feature region from the region including the extracted gray matter. Note that the re-extraction described in the second embodiment (S27 and S28 in FIG. 16) is not performed in the gray matter lesion extraction.

  Further, the volume calculation unit 41 calculates the ratio of the volume of the feature area (gray matter lesion area) to the volume of the entire gray matter area, and calculates the ratio of the front volume of the feature area to the rear volume of the feature area. May be. Further, the volume calculation unit 41 may calculate the ratio of the volume of the entire gray matter (cortex) area to the volume of the entire brain area.

  Then, we attempted quantitative analysis of brain compartment in 20 patients with multiple sclerosis (9 males, 11 females, mean age 42.2 ± 9.2 years old) with cortical (gray matter) / white matter lesions. It was. The cortical lesions of multiple sclerosis were evaluated with DIR (double inversion recovery) images. According to the classification of Nelson et al., The disease type was classified into a cortical lesion limited type, a mixed type (cortical lesion + subcortical white matter lesions are similar), and a paracortical type (mainly subcortical white matter lesions).

  The lesions were found to be cortical lesion-limited type 7, mixed type 6, and paracortical type, 29 in total. The ratio (%) of the volume of the entire cortical area to the volume of the entire brain area is 43.67 ± 7.06% in the cortical lesion group (cortical lesion limited type + mixed type), and no cortical lesion group (paracortical type). In 44.02 ± 4.87%, there was no difference between the two groups in the degree of cortical atrophy.

  On the other hand, the ratio (%) of the volume of the entire white matter region to the volume of the entire brain region was 29.74 ± 2.99% in the group with cortical lesions and 32.82 ± 3.09% in the group without cortical lesions. This was the result of progressing white matter atrophy in a certain group. The EDSS (Expanded Disability Status Scale) score, which is a measure of neurological function, was significantly higher in the group of patients analyzed this time than in the group with and without cortical lesions. It is speculated that The white matter volume of the group with cortical lesions was small, suggesting that the progression of white matter chronic inflammation affects cortical lesion formation and nerve function. Evaluation of patients by EDSS is objective by evaluating as described above because the evaluator evaluates the patient's behavior and evaluates the subject because the evaluator's subjectivity is included and depends greatly on the mental state such as the patient's motivation. This suggests the possibility of proper evaluation.

  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 above-described embodiment, T2 image data is used in the processing in the alignment unit 15, the separation processing unit 19, and the conversion parameter calculation unit 25. However, processing may be performed using T1 image data.

  In the above embodiment, the T2 image 50 after alignment is aligned with the standard brain template 60 to calculate a conversion parameter, and based on the calculated conversion parameter, the mask image 70 of the standard brain is converted to that of the individual brain. The mask image 80 was converted. However, a conversion parameter is calculated by aligning the standard brain template 60 with the aligned T2 image 50, and the standard brain mask image 70 is converted into a personal brain mask image 80 based on the conversion parameter. Also good.

  In the above-described embodiment, the feature region extraction unit 37 and the feature region re-extraction unit 20 arbitrarily set the reference value and the allowable value of the pixel value used when calculating the volume of the feature region, and set the target pixel. Although extracted, a histogram composed of the pixel value and the frequency may be created, and the upper 5% on the side with the larger pixel value may be extracted as the target pixel, or a value of 80% or more with respect to the maximum value of the pixel value May be extracted as a target pixel. That is, it may be automatically extracted based on preset extraction conditions.

  Although the white matter extraction unit 33 and the feature region re-extraction unit 20 extract the white matter from the FLAIR image 90, the white matter may be extracted from the T1 image or the T2 image 50.

  Although the SPECT image is used as an image related to cerebral blood flow, a PET image related to cerebral blood flow obtained by a PET (Positive Emission Tomography) examination may be used.

  In addition, the re-extraction region extraction unit 16 extracts third region data including white matter from the FLAIR image data using the first mask image data and the second mask image data of the personal brain, and re-extracts the feature region. The unit 20 extracts a white matter lesion as a feature region from the extracted third region 94. However, the re-extraction region extraction unit 16 extracts the overlapping region 93 from the FLAIR image data using only the first mask image data of the personal brain, and the feature region re-extraction unit 20 uses the overlapping region 93 as a feature region. A white matter lesion may be extracted. Even with such a configuration, all feature regions (white matter lesion regions) can be reliably extracted.

  Further, the volume calculation unit 41 calculates the total white matter of the subject based on the first region data, the second region data, the third region data, the fourth region data, the fifth region data, and the feature region data. A region, a white matter lesion region, a total gray matter region of the subject based on the fourth region, a gray matter lesion region, a white matter region located on the anterior horn side, and a posterior corner side based on a predetermined boundary White matter area, white matter lesion area located on the anterior horn side, white matter lesion area located on the posterior corner side, gray matter lesion area located on the anterior horn side, gray matter lesion area located on the posterior corner side, located on the anterior horn side Calculating the number of pixels (volume) of at least two of the gray matter region, the gray matter region located on the back corner side, and the entire brain region of the subject, and calculating the number of pixels (volume) from the calculated region Select two areas and calculate the number of pixels (volume) Hazuki two selected areas ratio of may be calculated.

DESCRIPTION OF SYMBOLS 1,102 Information processing apparatus, 11 MRI data storage part, 16 Reextraction area extraction part, 19 Separation processing part, 20 Feature area reextraction part, 23 Standard brain data storage part, 25 Conversion parameter calculation part, 29 Mask image conversion part , 33 White matter extraction unit, white matter lesion extraction unit, 41 volume calculation unit, 45 superposition unit, 50 T2 image, 51 first region, 60 standard brain template, 61 anatomical standardized T2 image, 71 ventricular region 80, 80 brain images, 90 FLAIR images, 91, 92 second region, 93 overlapping region, 94 third region, 170 first mask image set on standard brain, 172 set on standard brain Second mask image, 180 first mask image of individual brain, 182 second mask image of individual brain

Claims (16)

A computer program for causing an information processing apparatus having a storage unit to store an MRI image of a head of a subject and a mask image for excluding a ventricle in a standard brain,
Separating a first region including a ventricle and white matter from a first image that is an MRI image of the subject;
Obtaining a conversion function for converting from one of the first image and the standard brain to the other;
Using the conversion function to convert a mask image for excluding the ventricle in the standard brain into a mask image for excluding the ventricle in the subject's individual brain;
Extracting a second region containing white matter excluding the ventricle from a second image that is an MRI image of the subject based on the first region and the mask image of the subject;
A computer program for causing the information processing apparatus to execute a step of extracting a feature region from the second region. An MRI image of the head of the subject, a first mask image showing the whole brain region in the standard brain, and a second mask image corresponding to a region that is substantially ring-shaped and includes an edge region of the whole brain region in the standard brain A computer program for causing an information processing apparatus having a storage unit to store
Separating a first region containing white matter from a first image that is an MRI image of the subject;
Obtaining a conversion function for converting from one of the first image and the standard brain to the other;
Using the conversion function, a first mask image indicating the whole brain region in the standard brain and a second mask image corresponding to a region including an edge region of the whole brain region in the standard brain, A first mask image showing an entire brain region in the subject's individual brain and a second mask image corresponding to a region including an edge region of the entire brain region in the standard brain in the subject's individual brain;
Extracting a second region containing white matter from a second image, which is an MRI image of the subject, based on the first region and the first mask image of the subject;
Extracting a feature region from the second region;
An area that overlaps the entire brain area of the first mask image of the subject is extracted from the second image, and an area inside the overlapping area of the second image with respect to the second mask image of the subject is extracted. Extracting as a third region containing white matter;
A computer program for causing the information processing apparatus to execute a step of extracting a feature region from the third region.   The computer program product according to claim 2, wherein the feature region is a white matter lesion region.   The computer program according to claim 2, further comprising a step of displaying the feature region and the SPECT image of the subject in a superimposed manner. Separating a fourth region containing gray matter from a first image that is an MRI image of the subject;
Extracting a fifth region containing gray matter from a second image that is an MRI image of the subject based on the fourth region and the first mask image of the subject;
The computer program according to claim 2, further comprising: extracting a feature region from the fifth region. The feature region extracted from the second region is a white matter lesion region;
The feature region extracted from the third region is a white matter lesion region;
The feature region extracted from the fifth region is a gray matter lesion region;
The subject's white matter region based on the first region, the white matter lesion region based on the second region and the third region, the subject's whole gray matter region based on the fourth region, the fifth region The gray matter lesion area based on the area of the white matter, the white matter area located on the anterior horn side, the white matter area located on the posterior corner side, the white matter lesion area located on the anterior horn side, and the posterior corner side based on a predetermined boundary White matter lesion area, gray matter lesion area located on the anterior horn side, gray matter lesion area located on the posterior horn side, gray matter area located on the anterior horn side, gray matter area located on the posterior corner side, and subject The number of pixels of at least two of the whole brain regions is calculated, two regions are selected from the regions for which the number of pixels is calculated, and the ratio of the two selected regions is calculated based on the calculated number of pixels The compilation of claim 5 further comprising a step. Over data program. Wherein in the step of calculating calculates the number of pixels white matter lesion area located in white matter lesion area and the front corner side is located on the rear angle side, based on the number of pixels calculated before with respect to white matter lesion area located to the rear corner side The computer program according to claim 6, wherein a ratio of a white matter lesion area located on a corner side is calculated. The step of calculating further comprises calculating the number of pixels of the subject's white matter region and the subject's white matter lesion region, and calculating the ratio of the subject's white matter lesion region to the subject's whole white matter region. 6. The computer program according to 6.   The computer program according to any one of claims 2 to 8, wherein the second image is a T1 image, a T2 image, or a FLAIR image. An MRI image of the head of the subject, a first mask image showing the whole brain region in the standard brain, and a second mask image corresponding to a region that is substantially ring-shaped and includes an edge region of the whole brain region in the standard brain A storage unit for storing
Separating means for separating a first region containing white matter from a first image that is an MRI image of the subject;
Calculating means for obtaining a conversion function for converting from one of the first image and the standard brain to the other;
Using the conversion function, a first mask image indicating the whole brain region in the standard brain and a second mask image corresponding to a region including an edge region of the whole brain region in the standard brain A first mask image showing an entire brain area in the brain, and a conversion means for converting into a second mask image corresponding to an area including an edge area of the entire brain area in the individual brain of the subject;
First extraction means for extracting a second region containing white matter from a second image that is an MRI image of the subject based on the first region and the first mask image of the subject;
First extraction means for extracting a feature region from the second region;
A portion that overlaps the entire brain region of the first mask image of the subject is extracted from the second image, and a region inside the overlapping portion of the second image with respect to the second mask image of the subject A third extraction means for extracting a third region containing white matter;
An information processing apparatus comprising: fourth extraction means for extracting a feature region from the third region.   The information processing apparatus according to claim 10, wherein the feature region is a white matter lesion region. An MRI image of the head of the subject, a first mask image showing the whole brain region in the standard brain, and a second mask image corresponding to a region that is substantially ring-shaped and includes an edge region of the whole brain region in the standard brain An information processing apparatus having a storage unit for storing
The information processing apparatus is
Separating a first region containing white matter from a first image that is an MRI image of the subject;
Obtaining a conversion function for converting from one of the first image and the standard brain to the other;
Using the conversion function, a first mask image indicating the whole brain region in the standard brain and a second mask image corresponding to a region including an edge region of the whole brain region in the standard brain Converting to a first mask image indicating the entire brain region in the brain and a second mask image corresponding to a region including an edge region of the entire brain region in the individual brain of the subject;
Extracting a second region containing white matter from a second image, which is an MRI image of the subject, based on the first region and the first mask image of the subject;
Extracting a feature region from the second region;
An area that overlaps the entire brain area of the first mask image of the subject is extracted from the second image, and an area inside the overlapping area of the second image with respect to the second mask image of the subject is extracted. Extracting as a third region containing white matter;
Extracting a feature region from the third region. An MRI image of the subject's head, a first mask image indicating the entire brain region in the standard brain, and a second mask image that is substantially ring-shaped and corresponds to an edge region of the entire brain region in the standard brain are stored. A computer program for causing an information processing apparatus having a storage unit to be executed,
Separating a first region containing white matter from a first image that is an MRI image of the subject;
Obtaining a conversion function for converting from one of the first image and the standard brain to the other;
Using the transformation function, a first mask image indicating the whole brain region in the standard brain and a second mask image corresponding to the edge region of the whole brain region in the standard brain Converting to a first mask image representing the entire brain region in the brain and a second mask image corresponding to an edge region of the entire brain region in the standard brain in the subject's individual brain;
Extracting a second region containing white matter from a second image, which is an MRI image of the subject, based on the first region and the first mask image of the subject;
Extracting a feature region from the second region;
An area that overlaps the entire brain area of the first mask image of the subject is extracted from the second image, and an area inside the overlapping area of the second image with respect to the second mask image of the subject is extracted. Extracting as a third region containing white matter;
A computer program for causing the information processing apparatus to execute a step of extracting a feature region from the third region. An MRI image of the subject's head, a first mask image indicating the entire brain region in the standard brain, and a second mask image that is substantially ring-shaped and corresponds to an edge region of the entire brain region in the standard brain are stored. A storage unit to
Separating means for separating a first region containing white matter from a first image that is an MRI image of the subject;
Calculating means for obtaining a conversion function for converting from one of the first image and the standard brain to the other;
Using the conversion function, a first mask image indicating the whole brain region in the standard brain and a second mask image corresponding to an edge region of the whole brain region in the standard brain are converted into a brain in the individual brain of the subject. A first mask image showing an entire area, and a conversion means for converting into a second mask image corresponding to an edge area of the entire brain area in the subject's individual brain;
First extraction means for extracting a second region containing white matter from a second image that is an MRI image of the subject based on the first region and the first mask image of the subject;
First extraction means for extracting a feature region from the second region;
A portion that overlaps the entire brain region of the first mask image of the subject is extracted from the second image, and a region inside the overlapping portion of the second image with respect to the second mask image of the subject A third extraction means for extracting a third region containing white matter;
An information processing apparatus comprising: fourth extraction means for extracting a feature region from the third region.   A computer-readable recording medium storing the computer program according to any one of claims 2 to 9 and claim 13. An MRI image of the subject's head, a first mask image indicating the entire brain region in the standard brain, and a second mask image that is substantially ring-shaped and corresponds to an edge region of the entire brain region in the standard brain are stored. A method executed by an information processing apparatus having a storage unit
The information processing apparatus is
Separating a first region containing white matter from a first image that is an MRI image of the subject;
Obtaining a conversion function for converting from one of the first image and the standard brain to the other;
Using the conversion function, a first mask image indicating the whole brain region in the standard brain and a second mask image corresponding to an edge region of the whole brain region in the standard brain are converted into a brain in the individual brain of the subject. Converting to a first mask image indicating an entire area and a second mask image corresponding to an edge area of the entire brain area in the individual brain of the subject;
Extracting a second region containing white matter from a second image, which is an MRI image of the subject, based on the first region and the first mask image of the subject;
Extracting a feature region from the second region;
An area that overlaps the entire brain area of the first mask image of the subject is extracted from the second image, and an area inside the overlapping area of the second image with respect to the second mask image of the subject is extracted. Extracting as a third region containing white matter;
Extracting a feature region from the third region.

Images