JP5878125B2 - Dementia diagnosis support apparatus and dementia diagnosis support method - Google Patents

Description

  The present invention relates to a dementia diagnosis support apparatus and a dementia diagnosis support method that support dementia diagnosis using medical images such as CT images, MR images, and US images.

Recent research has revealed that there are various types of dementia (Non-Patent Document 1, Non-Patent Document 2).
Broadly classified, there are cerebrovascular dementia (VD), which is a general term for dementia that appears in connection with cerebrovascular disorders, and Alzheimer's dementia, which occurs when a substance called beta amyloid develops in brain cells. .

  VD is further divided into multiple infarct dementia (large vessel occlusion), small vessel dementia (small vessel occlusion), hypoperfusion cerebrovascular dementia, cerebral hemorrhagic cerebrovascular dementia, etc. It is done. Furthermore, multiple infarct dementia is further divided into multiple lacunar infarctions and Binswanger disease (progressive subcortical vascular encephalopathy).

  As described above, dementia can be divided into various types, but it is important to know which type of dementia belongs to first in order to perform appropriate treatment. For example, in early Alzheimer-type dementia, it is thought that the parahippocampal gyrus is the earliest atrophy (Non-patent Document 3), but the parahippocampal gyrus has a small volume, and visual evaluation should be performed on CT images and MRI images. It was considered difficult.

  In recent years, however, methods for performing brain image diagnosis have been proposed (Patent Document 1, Non-Patent Document 3). For example, Patent Document 1 describes an apparatus for determining brain atrophy. In the technique shown in Patent Document 1, the ratio of gray matter volume and white matter volume with respect to the entire brain is calculated using the brain MRI image, and the ratio between the intracranial volume and the white matter volume is also calculated. Determine atrophy. Non-Patent Document 3 describes an early stage Alzheimer-type dementia diagnosis support system called VSRAD (registered trademark). In the diagnosis support system of Non-Patent Document 3, MRI images are used to perform numerical evaluation by comparing the degree of atrophy of the parahippocampal volume with that of a normal brain. Non-Patent Document 4 describes a method of diagnosing dementia by measuring the lateral ventricular lower angle using CT images, quantitatively evaluating the hippocampus and its surrounding atrophy.

International Publication No.WO2007 / 114238

"Study on clinical and imaging of multiple infarct dementia and Alzheimer type dementia" Endo Riyuko, Tojo Medical University, Vol. 59, No. 6, p693-704 “Image diagnosis of Alzheimer's disease and cerebrovascular dementia from the viewpoint of neurology” Shinya Kawabata, Journal of Japanese Society of Radiological Technology, Vol. 58, No. 10, p.1335-1343 “Early Alzheimer-type dementia diagnosis support system VSRAD (registered trademark)” Hiroshi Matsuda, Journal of Japanese Society of Radiological Technology, Vol. 62, No. 8, p.1066-1072 "Development of an automatic measurement method for temporal medial part using head CT images" Masumi Hattori and Shuji Koyama, Journal of Japanese Society of Radiological Technology, Vol. 60, No. 7, p.993-998

  However, as shown in Patent Document 1 described above, it is difficult to separate and extract both gray matter and white matter with a CT image. Therefore, Patent Document 1 is a method that can be applied only to MRI images. Further, in order to use the VSRAD shown in Non-Patent Document 3 described above, it was premised on having an image (MRI image) in which the hippocampus was reflected first. Non-patent document 4 was also premised on having an image in which the lower angle of the lateral ventricle was shown.

  The present invention has been made in view of the above problems, and utilizes brain images in which evaluation target sites such as parahippocampal gyrus and lower ventricular angle are not clearly shown, and CT images widely used in clinical practice. It is an object of the present invention to provide a dementia diagnosis support apparatus and a dementia diagnosis support method that can perform image analysis of the brain and can effectively support the diagnosis of the type of dementia.

In order to achieve the above-described object, the present invention relates to a focused density area extracting unit that extracts a pixel corresponding to a specific density range from an input brain image as a focused density area, and a distribution of the focused density area as a figure. , A graph or a numerical value, and a display means for displaying a figure, graph or numerical value representing the distribution of the concentration area of interest obtained by the calculating means. This is a dementia diagnosis support device.
Preferably, a brain region extracting means for extracting a brain region from the inputted brain image is provided.

  The present invention also provides a focused density region extraction step for extracting pixels corresponding to a specific density range from the input brain image as a focused density region, and the distribution of the focused density region as a figure, a graph, or a numerical value. A dementia diagnosis support method comprising: an operation step for performing an operation for representing; and a display step for displaying a figure, graph, or numerical value representing the distribution of the concentration region of interest obtained by the operation means. It is.

  According to the present invention, it is possible to analyze a brain image using a brain image in which an evaluation target site such as a parahippocampal gyrus or a lateral ventricular lower angle is not clearly shown or a CT image widely used in clinical practice, and further, dementia It is possible to provide a dementia diagnosis support apparatus and a dementia diagnosis support method capable of effectively supporting the types of diagnosis.

The figure which shows the whole structure of the dementia diagnosis support apparatus 100 The flowchart explaining the procedure of the concentration distribution graphicalization process which the dementia diagnosis assistance apparatus 100 which concerns on this invention performs Examples of brain images (input images) by type of dementia Figure clearly showing the low density area of the input image (a) Image obtained by extracting the low density area, (b) Image obtained by removing the ventricle area from the low density area An example of brain region division Example showing the size of the low-concentration area from which the ventricular area has been removed as a figure for each divided area Display example in left-right overlay mode (multiple infarct type) Display example in left-right overlay mode (Alzheimer type) Display example in image difference mode Setting example of division center position Setting example of dividing line Other examples of brain region segmentation and other graphic display formats Examples of brain images (input images) by type of dementia Flow chart explaining the procedure of density distribution graph and digitization processing Examples of images to be processed (CT images and binarized images) Graph G1 showing CT value distribution against standardized radius by type Graph G2 showing binary distribution for normalized radius by type A diagram that explains the characteristics of the distribution of low-concentration regions by type in a diagram in which the ventricles are removed from the original CT image and the low-concentration regions are binarized An example of a display screen that displays the analysis results Image showing average distance R _CT (θ) (average distance from center to low density pixels) The figure explaining calculation of the distance between curves used for type judgment

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
First, the configuration of an image processing system 1 to which the dementia diagnosis support apparatus 100 of the present invention is applied will be described with reference to FIG.

  As shown in FIG. 1, the image processing system 1 includes a display device 107, a dementia diagnosis support device 100 having an input device 109, an image database 111 connected to the dementia diagnosis support device 100 via a network 110, And a medical image photographing device 112.

  The dementia diagnosis support apparatus 100 is a computer that performs processing such as image generation and image analysis. For example, a medical image processing apparatus installed in a hospital or the like is included.

  As shown in FIG. 1, the dementia diagnosis support apparatus 100 includes a CPU (Central Processing Unit) 101, a main memory 102, a storage device 103, a communication interface (communication I / F) 104, a display memory 105, a mouse 108, etc. An interface (I / F) 106 with a device is provided, and each unit is connected via a bus 113.

  The CPU 101 calls a program stored in the main memory 102 or the storage device 103 to the work memory area on the RAM of the main memory 102 and executes it, and drives and controls each unit connected via the bus 113 to diagnose dementia. Various processes performed by the support device 100 are realized.

In addition, the CPU 101 executes a density distribution graphing process (see FIG. 2), a density distribution graph / digitization process (see FIG. 15), which will be described later, and the like on the captured brain region image (hereinafter referred to as a brain image). The distribution of the concentration region of interest in the brain image is displayed as a figure, graph, or numerical value.
The density distribution graphicizing process and the density distribution graph / digitizing process will be described in each embodiment described later.

  The main memory 102 includes a ROM (Read Only Memory), a RAM (Random Access Memory), and the like. The ROM permanently holds a computer boot program, a BIOS program, data, and the like. The RAM temporarily holds programs, data, and the like loaded from the ROM, the storage device 103, and the like, and includes a work area that the CPU 101 uses for performing various processes.

  The storage device 103 is a storage device that reads / writes data to / from an HDD (hard disk drive) or other recording medium, and stores programs executed by the CPU 101, data necessary for program execution, an OS (operating system), and the like. . As for the program, a control program corresponding to the OS and an application program are stored. Each of these program codes is read by the CPU 101 as necessary, transferred to the RAM of the main memory 102, and executed as various means.

  The communication I / F 104 includes a communication control device, a communication port, and the like, and mediates communication between the dementia diagnosis support apparatus 100 and the network 110. The communication I / F 104 performs communication control with the image database 111, another computer, or a medical image photographing apparatus 112 such as an X-ray CT apparatus or an MRI apparatus via the network 110.

  The I / F 106 is a port for connecting a peripheral device, and transmits / receives data to / from the peripheral device. For example, a pointing device such as a mouse 108 or a stylus pen may be connected via the I / F 106.

  The display memory 105 is a buffer that temporarily accumulates display data input from the CPU 101. The accumulated display data is output to the display device 107 at a predetermined timing.

  The display device 107 includes a display device such as a liquid crystal panel and a CRT monitor, and a logic circuit for executing display processing in cooperation with the display device, and is connected to the CPU 101 via the display memory 105. The display device 107 displays the display data stored in the display memory 105 under the control of the CPU 101.

  The input device 109 is an input device such as a keyboard, for example, and outputs various instructions and information input by the operator to the CPU 101. The operator interactively operates the dementia diagnosis support apparatus 100 using external devices such as the display device 107, the input device 109, and the mouse 108.

  The network 110 includes various communication networks such as a LAN (Local Area Network), a WAN (Wide Area Network), an intranet, and the Internet, and communication between the image database 111, a server, other information devices, and the like and the dementia diagnosis support apparatus 100. Mediate the connection.

The image database 111 stores and stores image data captured by the medical image capturing device 112. In the image processing system 1 shown in FIG. 1, the image database 111 is connected to the dementia diagnosis support apparatus 100 via the network 110, but the image database 111 is stored in, for example, the storage device 103 in the dementia diagnosis support apparatus 100. May be provided.
[First embodiment]
Next, a first embodiment of the operation of the dementia diagnosis support apparatus 100 according to the present invention will be described with reference to FIGS.

  In the first embodiment, the CPU 101 of the dementia diagnosis support apparatus 100 reads a program and data related to the concentration distribution graphic processing shown in FIG. 2 from the main memory 102, and executes processing based on the program and data.

  At the start of execution of the following processing, the brain image data to be calculated is acquired from the image database 111 or the like via the network 110 and the communication I / F 104 and stored in the storage device 103 of the dementia diagnosis support apparatus 100. It shall be.

  The brain image is a tomographic image (two-dimensional) or three-dimensional volume image of the head imaged by an X-ray CT apparatus, an MRI apparatus, an ultrasonic apparatus or the like. In the following example, a two-dimensional axial CT image is used as a brain image, but a coronal image can also be used.

  In the concentration distribution graphing process of FIG. 2, first, the CPU 101 of the dementia diagnosis support apparatus 100 captures a brain image to be processed.

  At this stage, for example, an image as shown in FIG. 3 is captured. Fig. 3 (a) is a head CT image 20a of a patient with multiple infarct type dementia, (b) is a head CT image 20b of a patient with Alzheimer type dementia, (c) is a head CT image 20c of a patient with mixed dementia. Indicates.

  First, the CPU 101 performs alignment using the skull region or the cerebrospinal fluid region as preprocessing of the captured image (input image) (step S1). This is to match the shape of the brain image based on the brain image of each individual case.

The skull region and cerebrospinal fluid region can be extracted from the CT value of the input image. Further, the image alignment processing is performed on the basis of a reference image given in advance or a past image of the same patient.
By performing this preprocessing, it becomes possible to statistically and accurately analyze brain images of various cases.

  Next, the CPU 101 extracts a focused density region from the preprocessed image. In the present invention, a low concentration region from which the ventricle region is removed is extracted as the concentration region of interest (step S2).

In step S2, the first CPU101 is a brain region of the preprocessed image thresholded to extract a lower concentration region than the predetermined threshold value t ₂ (low concentration region). Then, regions indicated by thick lines in the images 21a, 21b, and 21c in FIG. 4, that is, the ventricle region 23 and other low-concentration regions 22 as shown in the image 24b ′ in FIG. 5 (a) are extracted. The CPU 101 further removes the ventricular region 23 from the extracted region and removes the ventricular region as shown in the image 24b of FIG.5 (b) (hereinafter, the ventricular removal low concentration region 22). Extract). The ventricular region has a particularly low CT value and can be easily recognized from anatomical information that it is located near the center of the brain.

Incidentally, not only the above procedure, previously at the stage of the threshold processing, by setting the low CT value threshold so as not to extract the ventricular area in the region (less than t ₁ or t _2), previously recognizing the ventricular region There is no need to perform processing.

  Next, in order to display the information more clearly, the CPU 101 executes the processing of steps S3 to S5 in FIG. 2 to display the complicated distribution of the ventricular removal low-concentration region 22 in a simple figure.

First, the CPU 101 divides the brain region or the inside of the skull into a plurality of regions (step S3).
In step S3, the CPU 101 divides the brain region or the inside of the skull symmetrically by a center line 3a that divides at least the left brain and the right brain. The left-right symmetry here is based on the arrangement of the left and right brains and does not necessarily match the left-right symmetry on the image. Therefore, the center line 3a is, for example, a line connecting brain depressions by cerebral columns. Further, when increasing the number of divided areas, the CPU 101 moves the center line 3a within the image plane by a predetermined angle around the center of the brain or the center point of the center line 3a (the line connecting the brain depressions by the cerebral column). Rotate and divide into desired number of divisions. FIG. 6 shows an example in which the brain region is divided into 16 by the center line 3a and the dividing lines 3b to 3h. In FIG. 6, the areas divided by the center line 3a and the dividing lines 3b to 3h are referred to as a divided area A, a divided area B,.

  Regarding the brain region segmentation method, `` Development of an automatic measurement method for the medial part of the temporal lobe using head CT images '' by Masumi Hattori and Shuji Koyama, Journal of Japanese Society of Radiological Technology, Vol. 60, No. 7, p. 994 etc. may be referred to and other methods may be used.

  Next, the CPU 101 obtains the total area and average coordinates of the ventricular removal low concentration region 22 for each divided region (step S4). If the image to be processed is two-dimensional, the total area is obtained. If the image to be processed is a three-dimensional image, the total volume is obtained instead of the total area.

  The CPU 101 generates graphics 5a to 5p having a size corresponding to the size of the total area (or total volume) calculated in step S4, and displays the graphics 5a to 5p in the divided regions A to B as shown in FIG. The display position is displayed at a position centered on the average coordinate of the ventricular removal low concentration region 22 in each of the divided regions A to B (step S5).

  In step S5, the sizes of the figures 5a to 5p are the same as the total area calculated in step S4 or an area proportional to the total area (in the case of two dimensions). In the case of three dimensions, the volume is the same as the total volume or a volume proportional to the total volume.

A display example of step S5 is shown in FIG.
As shown in FIG. 7, figure display images 25a, 25b, and 25c are generated in which figures 5a to 5p indicating the size of the ventricular removal low-concentration region 22 are displayed in each divided region of the brain image. The size of each figure 5a to 5p represents the size of the ventricular removal low concentration region 22 in the corresponding divided region, and the position of each figure 5a to 5p is the position of the ventricular removal low concentration region 22 in the corresponding divided region. Represents average coordinates.

  By generating and displaying such a graphic display image, it is possible to display in an easy-to-understand manner how and where the ventricular removal low-concentration region 22 is distributed in the brain region depending on the positions and sizes of the graphics 5a to 5p. In addition, the distribution of the ventricular removal low concentration region 22 is displayed as a graphic for each of the plurality of divided regions divided symmetrically, so the left-right symmetry of the ventricular removal low concentration region 22 is clearly displayed. it can.

  In addition, although the example which displays the figures 5a-5p with a circle was shown, it is good also as a rectangle, a mark, and other figures, for example. In addition, when applied to a three-dimensional image, a sphere may be used instead of a circle, or another three-dimensional figure may be used.

  The distribution of the ventricular removal low-concentration region 22 is shown for each type of dementia as shown in FIGS. 7 (a), (b), and (c). FIG. 7 (a) is a graphic display image 25a in the multiple infarct type, FIG. 7 (b) is a graphic display image 25b in the Alzheimer type, and FIG. 7 (c) is a graphic display image 25c in the mixed type.

  When comparing FIGS. 7 (a), (b), and (c), the Alzheimer type of (b) is more symmetrical than the multiple infarct type of (a), and the focused concentration region (ventricular removal low concentration region 22) Can be seen at a glance. Such a difference in the distribution of the low-concentration region by type is useful information for determining the type of dementia.

  Further, the dementia diagnosis support apparatus 100 includes a left-right superimposed display mode (see FIGS. 8 and 9), a left-right difference display mode, and an inter-image difference display mode (FIG. 10) as display modes different from the display example of FIG. Etc.). Further, as a user interface for switching between these display modes, the dementia diagnosis support apparatus 100 includes a “measurement” button 42, a “horizontal overlap” button 43, a “left / right difference” button 44, and an “image difference” button 45. It is desirable to provide software buttons including the like. Further, a “next patient number” button 46 for switching images and an “end” button 47 for ending the processing may be provided.

  When the “Measurement” button 42 shown in FIG. 8 is pressed with the mouse pointer 41 or the like, the mode is switched to the measurement mode, and the graphic display image 25a shown in FIG. Is displayed.

  When the “horizontal overlay” button 43 is pressed with the mouse pointer 41 or the like, the mode is switched to the horizontal overlay mode shown in FIGS. In the left-right superposition mode, the CPU 101 calculates the right region of the line (center line 3a in FIG. 8) connecting the brain depressions in the cerebral column among the graphics 5a to 5p of the divided regions A to P calculated in the measurement mode. Alternatively, the graphic in the left area is duplicated and displayed at a position symmetrical to the center line 3a. In order to prevent confusion, it is desirable to display the superimposed graphic (hereinafter referred to as the superimposed graphic) and the original graphic in different display formats (color, pattern, graphic, etc.). In the display device 107 shown in FIG. 8, a multiple infarct type graphic display image 25a and a left-right superimposed image 26a of the graphic display image 25a are displayed side by side. In the left-right superimposed image 26a, the figure 5p in the right area P is displayed as a superimposed figure 6p in the symmetry area A with the center line 3a as the symmetry axis, and the figure 5o in the right area O is in the symmetry area B. The figure 5n in the right area N is displayed as the superimposed figure 6n in the symmetrical area C, and the figure 5m in the right area M is displayed as the superimposed figure 6m in the symmetrical area D. Then, the graphic 5l in the right region L is displayed as a superimposed graphic 6l in the symmetric region E, and the graphic 5i in the right region I is displayed as a superimposed graphic 6i in the symmetric region H. The superimposed figures 6i to 6p displayed at the left and right symmetrical positions are displayed in a display format different from the original figures 5a to 5p.

  When the left-right superimposed mode button 43 is pressed with the mouse pointer 41 or the like for the Alzheimer-type graphic display image 25b, a left-right superimposed image 26b as shown in FIG. 9 is displayed.

  Comparing the left and right superimposed image 26a shown in FIG. 8 with the left and right superimposed image 26b shown in FIG. 9, the Alzheimer type dementia in FIG. That is, it can be seen that the distribution of the ventricular removal low concentration region 22 is highly symmetric, and this symmetric information is useful for the diagnosis of dementia type.

  When the left / right difference mode button 44 is pressed with the mouse pointer 41 or the like, the mode is switched to the left / right difference mode. In the left-right difference mode, the CPU 101 sets the left-right symmetric area with respect to the line (center line 3a in FIG. 8) connecting the brain depressions in the cerebral column among the graphics 5a to 5p of each divided area calculated in the measurement mode. A difference in size of each figure (hereinafter, left-right difference) is taken, and a figure indicating the left-right difference in size (hereinafter, referred to as a left-right difference graphic) is displayed (not shown). The shape of the left-right difference graphic is arbitrary, but the size is the same as the size of the left-right difference or a size proportional to the size of the left-right difference. In addition, the display position of the left / right difference graphic is the average position of the display positions of the respective figures at the symmetrical positions (average coordinates in each divided region of the ventricular removal low concentration region 22). Or you may display a left-right difference figure in the position where the big figure is displayed among each figure in a left-right symmetric position.

  When the inter-image difference button 45 is pressed with the mouse pointer 41 or the like, the mode is switched to the inter-image difference mode. In the inter-image difference mode, the CPU 101 determines the distribution of the ventricular removal low concentration region 22 of the input image and the distribution of the ventricular removal low concentration region 22 of the image to be compared, similar to the above-described steps S1 to S5. Each of these images is measured, and the difference between the ventricular removal low concentration regions 22 (hereinafter referred to as inter-image differences) is taken for each corresponding divided region, and a figure corresponding to the size of the inter-image difference (hereinafter referred to as the inter-image difference). This is called an inter-image difference graphic (corresponding to 7b to 7o in FIG. 10) and is displayed for each divided region. The shapes of the inter-image difference graphics 7b to 7o are arbitrary, but the size is the same as the size of the inter-image difference or a size proportional to the size of the inter-image difference. In order to prevent confusion, it is desirable to display the inter-image difference graphics 7b-7o and the original graphics 5a-5p, 5A-5P in different display formats (colors, patterns, graphics, etc.). Also, the display positions of the inter-image difference graphics 7b to 7o are the average positions of the respective figures in the corresponding areas of the past image 25p and the current image 25P (for example, the average position of the display position of the graphic 5a and the display position of the graphic 5A). To do.

  FIG. 10 is an example of a display screen in the inter-image difference mode. In FIG. 10, a coronal image is a measurement target. As shown in FIG. 10, on the display screen of the display device 107, a graphic display image 25p for the past image of the same patient and a graphic display image 25P for the current image are displayed side by side. An inter-image difference image 28 obtained by taking the inter-image difference of the distribution of the ventricular removal low-concentration region 22 is displayed for each corresponding divided region of these graphic display images 25p and 25P. In FIG. 10, since the areas of the graphic 5a and the graphic 5A are the same, the corresponding inter-image difference graphic is not displayed. The same applies to the figure 5p and the figure 5P.

  By observing the inter-image difference graphics 7b to 7o displayed on the inter-image difference image 28, it is possible to easily confirm how much difference in the distribution of the target concentration area is in which divided area. It becomes easier to check the degree of progress.

  Note that the image for which the difference is taken is not limited to the past image and the current image, and may be a reference image that is input in advance and a comparison image that is input. The reference image is preferably stored in the storage device 103 of the dementia diagnosis support apparatus 100 for each type of dementia.

  In each of the above modes, as shown in FIGS. 8 to 10, a division number input field 48 may be provided as a user interface for designating the division number of the divided area. The CPU 101 divides the brain region into the division number input in the division number input field 48 in the above-described step S3 (FIG. 2).

  As described above, in the first embodiment, the CPU 101 of the dementia diagnosis support apparatus 100 extracts the brain region from the input brain image, and the attention belonging to a specific concentration range from the extracted brain region A concentration region (here, a low concentration region from which the ventricle region has been removed; a ventricular removal low concentration region 22) is extracted. In addition, the CPU 101 displays the distribution of the ventricular removal low-concentration region 22 as simple graphics 5a to 5p in the original image. The graphics 5a to 5p are displayed in the divided areas A to P obtained by dividing the brain area based on the symmetry of the left and right brains, respectively. The sizes of the figures 5a to 5p indicate the size of the ventricular removal low concentration area 22 in each divided area A to P, and the display positions of the figures 5a to 5p are the ventricular removal in each divided area A to P. The average position of the low concentration region 22 is used.

  Thereby, it is possible to display in an easy-to-understand manner how and where the concentration region of interest (the ventricular removal low concentration region 22) is distributed in the brain region by the position and size of the figures 5a to 5p. In addition, since the distribution of the ventricular removal low concentration region 22 is displayed in a graphic form for each of the divided regions A to P divided symmetrically, the left and right symmetry of the ventricular removal low concentration region 22 is increased. Can be clearly displayed. Therefore, if a doctor or the like uses the dementia diagnosis support apparatus 100 of the present invention for diagnosis of dementia, the distribution of the ventricular removal low-concentration region 22 can be confirmed in an easy-to-understand manner, and the discrimination of the type of dementia is particularly effective. Can help. In addition, since the ventricular removal low concentration region 22 can be easily extracted from the CT image, it becomes possible to perform dementia diagnosis using CT images widely used in clinical practice.

  In the left-right superimposed display mode, the dementia diagnosis support apparatus 100 displays a graphic displayed on either the left or right brain symmetrically with respect to the center line 3a dividing the left and right brain.

  This makes it easier to confirm the left-right symmetry of the ventricular removal low concentration region 22, and can more effectively support the type of dementia. In the Alzheimer type, it is known that the ventricular removal low concentration region 22 is distributed symmetrically compared to the case of multiple infarct dementia, so pay attention to the left-right symmetry of the ventricular removal low concentration region 22 It is also effective for the diagnosis of dementia type.

  Also, in the left / right difference display mode, the difference in size (left / right difference) of each figure in the left / right symmetrical area with respect to the center line 3a dividing the left / right brain is taken. A right-and-left differential graphic having a size proportional to the size is generated and displayed.

  This makes it easier to confirm the left-right symmetry of the ventricular removal low-concentration region 22 as in the case of overlapping display. More effective support for determining the type of dementia.

  Further, in the inter-image difference display mode, a graphic display image is displayed for each of the reference brain image (for example, the past image of the patient) and the comparative brain image (for example, the current image of the patient), and the reference brain image and the comparative brain image are also displayed. For each corresponding divided area of each graphic display image, calculate the difference in size of the ventricular removal low-density region 22 (difference between images), and the difference between the images in proportion to the size of the difference between images A figure is generated and the inter-image difference image 28 is displayed.

This makes it possible to easily compare the distribution of the ventricular removal low-concentration region 22 between the two images, and can support diagnosis of the type and course of dementia.
In the above example, the number of divisions of the brain region is 16, but the number is not limited to 16.

  The center of the division may be not only the center of gravity of the brain region but also any point on the line connecting the brain depressions by the cerebral column. For example, it may be divided into portions effective for diagnosis from an anatomical viewpoint. In this case as well, it is desirable to divide the brain region so that at least the symmetry of the left and right brains can be seen.

  For example, as shown in FIG. 11, as a user interface 291 for setting a center point, a center line, and a dividing line, a handle H0 for setting the center point of the division, a center line for setting the division It is desirable to provide a handle H1 and handles H2a and H2b for setting a dividing line. The handles H0, H1, H2a, and H2b can be moved by an operation using a pointing device such as the mouse 108.

  When the operator operates the handle H1, the center line can be manually set based on the judgment of a doctor or the like. If the initial position of the handle H1 is a straight line connecting the brain depressions by the cerebral tandem lines detected by the CPU 101, the setting of the center line becomes easy and flexible.

  The handle H0 moves on the center line. The position of the handle H0 moved by the operation of the operator is set as the position of the center point.

  The handles H2a and H2b are handles that are moved relative to one another, and a straight line connecting the positions of the handles H2a and H2b and the center point (the position of the handle H0) is set as a dividing line. The positions of the handles H2a and H2b are interlocked so as to be always symmetrical with respect to the center line. When the number of divisions is large, H2a and H2b for setting a division line are provided in a number corresponding to the number of divisions.

Further, the dividing lines 3b to 3h of the region may be not only a straight line but also a curved line.
In this case, for example, as shown in FIG. 12, a handle h1, h2, h3, h4 may be provided on each dividing line as a user interface 292 for setting the shape of the dividing line.
The handles h1, h2, h3, and h4 can be moved by an operation using a pointing device such as the mouse 108.

  For example, by moving the position of the handle h1, the shape of a line (partition line) connecting the handle H2a and the center line can be transformed into an arbitrary curve. The same applies to the handles h2, h3, h4. Furthermore, a plurality of handles similar to h1, h2, h3, and h4 may be provided on each dividing line. In addition, it is desirable that the corresponding handles are interlocked so that the left and right dividing lines are symmetrical with respect to the center line.

  By providing such user interfaces 291 and 292, it is possible to freely set the size and shape of the divided areas, which is useful for diagnosis of individual patients. For example, the dividing line can be set to an appropriate position and shape according to the cross section (axial, coronal) of the image, or the dividing line can be set to an appropriate position and shape based on anatomical information.

Further, the graphic representing the distribution of the ventricular removal low concentration region 22 is not limited to a circle. As in the graphic display image 293 shown in FIG. 13, the graphics 5A to 5F may be rectangular or other shapes. Also, the pattern and color of the figure are arbitrary.
[Second Embodiment]
Next, a second embodiment of the operation of the dementia diagnosis support apparatus 100 according to the present invention will be described with reference to FIGS.

  In FIG. 14, (a) is a Binswanger type, (b) is a multiple infarct type, (c) is an Alzheimer type, and (d) is a typical brain image of mixed type dementia. A region surrounded by a central line in each image in FIG. 14 is a ventricle region.

  As shown in FIG. 14, in (a) Binswanger type, low-concentration regions other than ventricular region 23 are densely distributed throughout (b) In multiple infarct types, low-concentration regions other than ventricular region 23 are distributed. (C) In the Alzheimer type, low-concentration regions other than the ventricular region 23 are distributed around the edge of the brain region. (D) In the mixed type, low-concentration regions other than the ventricular region 23 are distributed. It is distributed near the ventricle and the edge of the brain area. That is, the type of dementia depends not only on the left-right symmetry of the ventricular removal low concentration region, but also on how the ventricular removal low concentration region spreads.

  In the dementia diagnosis support apparatus 100 according to the second embodiment, the characteristic concentration distribution of such an image is graphed or digitized so that a doctor or the like can easily determine which type it belongs to. Process.

  In the second embodiment, the CPU 101 of the dementia diagnosis support apparatus 100 reads out the program and data related to the concentration distribution graph / digitization processing shown in FIG. 15 from the main memory 102, and based on this program and data, A target density area is extracted, a calculation for graphing and digitizing the distribution of the extracted target density area is performed, and a calculation result is displayed.

  At the start of execution of the following processing, the brain image data to be calculated is acquired from the image database 111 or the like via the network 110 and the communication I / F 104 and stored in the storage device 103 of the dementia diagnosis support apparatus 100. It shall be.

  The brain image is a tomographic image (two-dimensional) or three-dimensional volume image of the head imaged by an X-ray CT apparatus, an MRI apparatus, an ultrasonic apparatus or the like. In the following example, a two-dimensional axial CT image is used as a brain image, but the present invention can also be applied to a coronal image.

  In the concentration distribution graph / digitization process of FIG. 15, first, the CPU 101 of the dementia diagnosis support apparatus 100 captures a brain image to be processed. Further, the CPU 101 generates an addition memory in the RAM work memory, and initializes (zero clear) (step S21). The addition memory is a one-dimensional array memory.

Next, the CPU 101 extracts a brain region based on the CT value of the captured brain image (input image) (step S22). Further, the CPU 101 obtains the center of gravity of the extracted brain region (step S23). Next, the CPU 101 recognizes the ventricle region of the input image (step S24). The ventricular region has a particularly low CT value and can be easily recognized from anatomical information that it is located near the center of the brain.
The ventricular region is removed from the brain region by the processing as described above.

In the subsequent procedure, the concentration distribution is analyzed with the region excluding the ventricular region from the brain region as the processing target.
For the image to be processed, for example, a CT image of a region obtained by removing the ventricle region from the brain region as shown in FIG. 16 (a) or only a low concentration region of interest as shown in FIG. 16 (b) is extracted. Any of binarized binarized images may be used.

  In the case of the CT image in FIG. 16 (a), the CT value of each pixel is a calculation target of a profile described later, and in the binarized image in FIG. Profile calculation target. This binarized image represents a low density pixel value “1” and the other pixels “0” in a region obtained by removing the ventricle from the brain region.

  As shown in FIG. 16, the CPU 101 sets a moving radius with the center of gravity (or the above-mentioned center point) of the brain region obtained in step S23 as the rotation center (step S25).

  Next, the CPU 101 obtains a density distribution profile along the moving radius at a certain rotation angle θ (step S26). When the CT value is a calculation target as in the CT image of FIG. 16 (a), the CPU 101 acquires the pixel value (CT value) at each point on the radius vector as a one-dimensional array profile. When a binary value is to be calculated as in the binarized image of FIG. 16B, the pixel value “1” or “0” at each point on the radius vector is acquired as a one-dimensional array profile.

  Depending on the rotational position of the radius and the patient, the length from the origin of the radius to the end of the brain region (until the extracted brain region or just before the skull) is different, so the CPU 101 determines the radius length at each rotation angle. Normalization is performed, the profile obtained in step S26 is interpolated, and the normalized profile is calculated (step S27). That is, when the radial length of each rotational angle is normalized with reference to the radial radius at a certain rotational angle, the profile obtained at a short angle of the total length of the radial radius (distance from the origin to the brain region end) Since the number of data to be acquired (number of pixels) is small compared to the profile obtained at the long angle of the total length, this is interpolated. For interpolation, the pixel value of the interpolated pixel may be estimated from the pixel value (CT value or binary) of the adjacent pixel.

  The CPU 101 adds the standardized profile to the addition memory (step S28). At this stage, the pixel value of each point on the radius vector at a certain rotation angle θ is normalized and stored in the addition memory.

  When the CT value is a calculation target (in the case of FIG. 16 (a)), the average CT value of the entire brain region may be subtracted from the normalized profile before addition to the addition memory. In this way, even if a CT value having a large value is set as a calculation target, the added value becomes small and overflow during calculation can be prevented. In addition, the profile change is easy to understand.

  Thereafter, the CPU 101 rotates the moving radius by a predetermined angle dθ (step S29), and executes the processing of step S26 to step S29 over all angle directions (360 degrees). When the profile for 360 degrees is obtained (step S30; YES), the CPU 101 divides the data (normalized profile) acquired in the addition memory by a predetermined value to obtain smoothed data. For example, if the data in the addition memory is divided by the number of additions, the data is smoothed over 360 degrees. The CPU 101 displays the obtained smoothed data as a graph (step S31). The graph represents the cumulative density value (CT value or binary value) with respect to the normalized radius.

  FIG. 17 is a graph G1 of the smoothed data (profile) with each CT image of FIG. 14 as the calculation target, and FIG. 18 is the smoothed data with the binary image as shown in FIG. (Profile) graph G2.

In the graph G1 of FIG. 17, the horizontal axis represents the normalized radius Rn, and the vertical axis represents the normalized average CT value (profile representing the relationship between the distance from the origin and the CT value at each normalized radius).
The thin line represents the Binswanger type, the broken line represents the multiple infarct type, the alternate long and short dash line represents the Alzheimer type, and the thick line represents the mixed data.

  Referring to graph G1, the Binswanger type has a low CT value as a whole, the multiple infarct type has a low CT value concentrated in a region close to the center of the brain (region with a small radius), and the Alzheimer type has a brain edge. It can be seen that low CT values are distributed in a concentrated manner (regions with a large radius), and the CT value is almost constant in any region in the mixed type.

  Thus, by graphing the relationship between the distance from the center of gravity of the brain region and the CT value, the dementia type features appearing in the image can be displayed in an easy-to-understand manner.

  In graph G2 of FIG. 18, the horizontal axis is the normalized radius Rn, the vertical axis is the count number (smoothed data; the value obtained by adding and smoothing the number of pixels with a normalized pixel value of 1 at each radial radius over 360 degrees. ). The thin line represents the Binswanger type, the broken line represents the multiple infarct type, the alternate long and short dash line represents the Alzheimer type, and the thick line represents the mixed data.

  Referring to the graph G2, the Binswanger type has a large count number regardless of radius (i.e., high density and low concentration as a whole), and the multiple infarct type has a count number near the center of the brain (part with a small radius) ( In the Alzheimer type, the number of counts (low density pixels) in the brain edge (large radius) is large, and in the mixed type, the count number is almost constant (low density overall). The ratio is small).

  Similarly, for smoothed data obtained from a binarized image, the relationship between the distance from the center of gravity of the brain region and the binary (whether or not it is low concentration) is graphed, and the perception that appears in the image The characteristics of the type of symptoms can be displayed in an easy-to-understand manner.

As described above, the CPU 101 of the dementia diagnosis support apparatus 100 adds the CT value over 360 degrees for each point on the radial center with the center of the brain (e.g., the center of gravity) as the origin, or 360 only low-density pixels. The graphs G1 and G2 representing the relationship between the distance from the center of the brain and the density value (CT value or low-density pixel) are generated by accumulating (counting) over time. Since the density distribution of the image of the brain region differs depending on the type of dementia, it can be displayed in a form that can be easily compared with graphs G1 and G2, and the type of dementia can be easily determined.
In addition to graphs, image features may be represented by numerical values.

FIG. 19 shows each type of binarized image.
Referring to FIG. 19, in the (a) Binswanger type, the number of low-density pixels is large and distributed at high density. The average distance over which the low density pixels are distributed is medium, and the distribution spread (dispersion) is large. (b) In the multiple infarct type, the number of low-density pixels is small and distributed at a low density. The average distance over which low density pixels are distributed is small, and the distribution spread (dispersion) is also small. (c) In the Alzheimer type, the number of low-density pixels is small and distributed at a low density. The average distance over which the low density pixels are distributed is large, and the spread (distribution) of the distribution is small. (d) In the mixed type, the number of low-density pixels is small and distributed at a low density. The average distance over which the low density pixels are distributed is medium, and the distribution spread (distribution) is large.

  Thus, since the distribution of pixels (except for the ventricle) indicating a low concentration varies depending on the type of dementia, the CPU 101 calculates the feature of the concentration distribution as a feature amount (step S32 in FIG. 15).

Features include the average distance (R _av , R _CT ) of the concentration area of interest from the center of gravity of the brain area, the deviation of concentration values (M), the ratio of the concentration area of interest in the brain area (disappearance rate), and the brain area Examples include the ratio of ventricles (ventricular rate).

  For example, as shown in FIG. 20, the CPU 101 displays the graph G1 in FIG. 17 and the graph G2 in FIG. 18 on the display device 107, and further calculates a reference feature amount to display as a numerical index 8, The correlation between the graph of the profile obtained from the input image and the reference graph calculated in advance from the reference image of each dementia type may be calculated and displayed as the type determination result 9.

For example, as the numerical index 8, average distances R _av , R _CT , feature quantity M, disappearance rate, ventricular rate, etc. are calculated and displayed on the display screen.

The average distance R _av is the average (distance after standardization) from the center of the above-mentioned radius to each low density pixel (low density pixel (pixel value = 1) in the binarized image of FIG. 16 (b)) (Average) of 0.0 to Rn. Rn is a normalized value of the distance from the center of the radius to the end of the brain region (or inside the skull).

The average distance R _CT is a value representing the characteristic of how the density value spreads from the center of the above-mentioned moving radius for the CT image of FIG. 16 (a), and is calculated by the following equation (1).

式(1)において、CT_maxは、CT値の最大値、CT_iは画素iにおけるCT値である。また、脳室を除去した低濃度領域に該当する画素についてのみ、演算される(t1＜CT_i＜t2)。

In Equation (1), CT _max is the maximum CT value, and CT _i is the CT value at pixel i. Also, calculation is performed only for pixels corresponding to the low density region from which the ventricle is removed (t1 <CT _i <t2).

  Equation (1) is calculated for each radius angle (θ). That is, the target pixel i is a pixel on the radius vector forming the pixel θ.

For a certain brain image, a graph of the calculation result of Expression (1) is shown in G3 of FIG. In the graph G3, since the average position of the positions where the low density pixels (pixels having CT _i ) exist in the image is displayed on the image, the characteristics of the density distribution can be expressed. As a criterion for determination, a graph (reference graph G10) of the average distance R _CT (θ) in images of various cases may be displayed together with the graph G3.

  The density value deviation M is obtained by the following equation (2).

式(2)において、CT_maxは、CT値の最大値、CT_iは画素iにおけるCT値である。また、脳室を除去した低濃度領域に該当する画素についてのみ、演算される(t1＜CT_i＜t2)。

In Equation (2), CT _max is the maximum CT value, and CT _i is the CT value at pixel i. Also, calculation is performed only for pixels corresponding to the low density region from which the ventricle is removed (t1 <CT _i <t2).

  The disappearance rate is a numerical value indicating the ratio of the low concentration region (the concentration region of interest) in the brain region, and is calculated by the following equation (3).

脳室率は、脳領域における脳室の割合を示す数値であり、以下の式(4)により算出される。

The ventricular rate is a numerical value indicating the proportion of the ventricle in the brain region, and is calculated by the following equation (4).

これらの特徴量R_av、R_CT、M、消失率、脳室率の他、認知症の診断に役立つ他の特徴量(例えば脳の萎縮率等)を算出し、表示するようにしてもよい。

In addition to these feature values R _av , R _CT , M, disappearance rate, ventricular rate, other feature values useful for diagnosis of dementia (for example, brain atrophy rate, etc.) may be calculated and displayed. .

  Further, as in the type determination result 9 shown in FIG. 20, the CPU 101 uses the graph G1 or G2 indicating the relationship between the normalized radius calculated for the input image and the concentration value, and the reference curve pattern for each type of dementia ( A graph indicating the relationship between the normalized radius and the density value) is compared with each other to determine which pattern the input image is close to (step S33; FIG. 15).

  The distance between curves can be obtained from the following equation (5).

すなわち、図22に示すように、現在画像を解析している患者の曲線(グラフG1、G2)と、基準曲線パターン(例えば、タイプ別の確定患者の平均的なグラフ)と、を用いて、それぞれ対応する点Pi，Qiでの値を比較し、数値化する。

That is, as shown in FIG. 22, using the curve of the patient who is currently analyzing the image (graphs G1, G2) and the reference curve pattern (for example, an average graph of confirmed patients by type), The values at the corresponding points Pi and Qi are compared and digitized.

  Instead of the inter-curve distance, the correlation between the graph and the reference pattern may be digitized from the Pearson product moment correlation coefficient r in Equation (6).

このように基準曲線との相関をタイプ別に表示することで、入力画像がどのタイプに近いかを数値により簡単に確認でき、認知症の診断に有効となる。

By displaying the correlation with the reference curve for each type in this way, it is possible to easily confirm the type of the input image that is close to the numerical value, which is effective for the diagnosis of dementia.

  As described above, in the second embodiment, the CPU 101 of the dementia diagnosis support apparatus extracts the brain region from the input brain image, removes the ventricle region from the extracted brain region, and the concentration of interest The pixel distribution (CT value distribution or low density region distribution) is analyzed and displayed in a graph or numerical form. CPU101 sets a radial radius centered on the center of gravity of the brain region, acquires a profile of the distance from the center and the concentration value (CT value or binary value) of each point on the radial radius over 360 degrees, A profile is obtained by sequentially adding density values (CT value or binary value) at each angle at the same distance from the center. At this time, the difference in the radial length of the profile at each angle is normalized. In addition, the points that lack data when normalized are interpolated by CT values or binary data of adjacent points on the radius vector. Then, the CPU 101 calculates smoothed data obtained by dividing the acquired normalized profile by a predetermined value, generates graphs G1 and G2 representing the cumulative density values with respect to the normalized radius from the smoothed data, and displays them on the display screen. .

As a numerical index 8 that can be calculated from the density distribution of the brain region image, the CPU 101 calculates the average distance R _av and R _CT that can be calculated from the density value of the image, the deviation M of the density value, the disappearance rate, the ventricular rate, etc. The feature amount is calculated and displayed. This makes it possible to recognize the type and progression of dementia quantitatively, and is effective for diagnosis.

  Further, as the type determination result 9, the CPU 101 compares the graph G1 or G2 indicating the relationship between the normalized radius calculated for the input image and the concentration value with the reference graph for each type of dementia, and the input image is Find which pattern is close. As a result, it is possible to easily confirm by numerical values which type the density distribution of the input image is close to and can effectively support the diagnosis of dementia.

  As mentioned above, although preferred embodiment of the dementia diagnosis assistance apparatus which concerns on this invention was described referring an accompanying drawing, this invention is not limited to this example.

  In the second embodiment described above, the low density pixels are extracted as “1” and the others as “0” as the binarized image. On the contrary, the high density pixels are extracted and the binarized image is extracted. You may get In this case, the ventricular region having a low CT value can be excluded without performing the ventricular region recognition process in step S24.

When high density pixels are extracted, the above average distance R _CT (θ) (average length R _HCT (θ) from the center to the high CT value pixel) is calculated from the following equation (7).

また第2の実施の形態では、CT画像を例として説明したが、MRI画像や超音波画像にも適用できる。MRI画像の場合、画像の濃淡はCT画像と反転するため、本実施の形態における「低濃度領域」は「高濃度領域」に置き換えられる。

In the second embodiment, a CT image is described as an example, but the present invention can also be applied to an MRI image or an ultrasound image. In the case of an MRI image, since the density of the image is reversed from that of the CT image, the “low density area” in the present embodiment is replaced with a “high density area”.

  Further, the graphic display image of the first embodiment described above and the graph, numerical index, and type determination result of the second embodiment may be displayed in combination.

  In addition, it is obvious that those skilled in the art can arrive at various changes or modifications within the scope of the technical idea disclosed in the present application, and these naturally belong to the technical scope of the present invention. It is understood.

  1 Image processing system, 100 Dementia diagnosis support device, 101 CPU, 102 Main memory, 103 Storage device, 104 Communication I / F, 105 Display memory, 106 I / F, 107 Display device, 108 Mouse, 109 Input device, 20a , 20b, 20c Input image, 22 Ventricular removal low concentration region, 23 Ventricular region, 24b Ventricular removal image, A, B, ..., P division region, 3a center line, 3b-3h division line, 25a, 25b, 25c figure display image, 5a-5p figure, 26a, 26b, 26c left and right superimposed image, 6a-6p left and right superimposed figure, 25p reference image, 25P comparison image, 28 image difference image, 7b-7o image difference Figure, H0 center point setting handle, H1 center line setting handle, H2a, H2b, H3a, H3b Dividing line setting handle, h1 to h4 dividing line deformation handle, G1, G2 Relationship between distance from brain center and density value , 8 numeric indicators, 9 type judgment results

Claims (11)

A focused dark area extracting means for extracting, as a focused density area, pixels corresponding to a specific density range from the input brain image;
A calculation means for performing a calculation for expressing the distribution of the concentration region of interest by a figure, a graph, or a numerical value;
Display means for displaying a figure, graph, or numerical value representing the distribution of the concentration region of interest obtained by the computing means;
A brain region extracting means for extracting a brain region from the inputted brain image,
When the distribution of the concentration region of interest is represented by the graphic,
The computing means is
A dividing unit that divides the brain region of the brain image into a plurality of divided regions based on left-right brain symmetry;
The size of the target density area extracted by the target density area extracting unit is measured for each divided area divided by the dividing unit, and a graphic representing the size of the measured target density area is generated for each divided area. Graphic generation means,
The display means includes
A dementia diagnosis support apparatus comprising graphic display means for generating and displaying a graphic display image in which the graphic generated by the graphic generating means is arranged in a corresponding divided area . The display means includes
The graphic generated by the graphic generating means is displayed in the corresponding divided area, and the graphic displayed on either the left or right brain is displayed symmetrically with respect to the center line dividing the left and right brain. 2. The dementia diagnosis support apparatus according to claim 1 , further comprising a left-right superimposed display means. The display means includes
The left and right differences are generated by displaying the graphics generated by the graphic generation means in the corresponding divided areas, and generating and displaying the left and right differential figures that represent the difference in the size of each figure displayed in the symmetrical divided areas. 2. The dementia diagnosis support apparatus according to claim 1 , further comprising display means. The display means includes
Displaying the graphic display image for the reference brain image and the graphic display image for the comparative brain image, respectively;
For each corresponding divided region of each graphic display image for the reference brain image and the comparative brain image, calculate the difference in the size of the concentration region of interest, and generate an inter-image difference graphic representing the size, 2. The dementia diagnosis support apparatus according to claim 1 , further comprising an inter-image difference display means for displaying in the divided area. A focused density region extracting means for extracting a pixel corresponding to a specific density range from the input brain image as a focused density region;
A calculation means for performing a calculation for expressing the distribution of the concentration region of interest by a figure, a graph, or a numerical value;
Display means for displaying a figure, graph, or numerical value representing the distribution of the concentration region of interest obtained by the computing means;
A brain region extracting means for extracting a brain region from the inputted brain image,
When representing the distribution of the concentration region of interest in the graph,
The computing means is
A radial setting means for setting a radial with the center of gravity of the brain region as the origin;
Profile acquisition means for acquiring a profile representing the relationship between the distance from the origin and the concentration value over all angles for each point on the radius vector;
Normalizing the profile acquired by the profile acquisition means based on a reference radial length, and generating a graph for the normalized profile, and a graph generation means,
The display means includes
Graph display means for displaying the graph generated by the graph generation means;
Dementia diagnosis support apparatus shall be the characterized in that it comprises. The dementia diagnosis support apparatus according to claim 5 , wherein the profile acquisition unit acquires the profile for a pixel value of a region obtained by removing a ventricle region from the brain region. The dementia diagnosis support apparatus according to claim 5 , wherein the profile acquisition unit acquires the profile for a binary value indicating whether or not the target concentration region corresponds. A reference graph holding means for holding a reference graph obtained by graphing a standardized profile acquired for each type of dementia;
Correlation value calculating means for calculating a correlation value for determining the type of dementia based on the graph generated by the graph generating means for the input brain image and the reference graph;
Type display means for displaying the correlation value calculated by the correlation value calculation means for each type of dementia;
6. The dementia diagnosis support apparatus according to claim 5 , further comprising: A focused density region extracting means for extracting a pixel corresponding to a specific density range from the input brain image as a focused density region;
A calculation means for performing a calculation for expressing the distribution of the concentration region of interest by a figure, a graph, or a numerical value;
Display means for displaying a figure, graph, or numerical value representing the distribution of the concentration region of interest obtained by the computing means;
A brain region extracting means for extracting a brain region from the inputted brain image,
When expressing the distribution of the concentration region of interest by the numerical value,
The calculation means includes a feature quantity calculation means for calculating a feature quantity indicating a feature of the concentration distribution of the brain region,
The display means includes numerical index display means for displaying the feature amount calculated by the feature amount calculation means,
The feature amount is
It includes any one of an average distance of the concentration region of interest from the center of gravity of the brain region, a deviation in concentration value, a proportion of the concentration region of interest in the brain region, and a proportion of the ventricle in the brain region. that dementia diagnosis support apparatus. A focused density region extraction step of extracting pixels corresponding to a specific density range from the input brain image as a focused density region;
A calculation step of performing a calculation for representing the distribution of the concentration region of interest by a figure, a graph, or a numerical value;
A display step for displaying a figure, a graph, or a numerical value representing the distribution of the concentration region of interest obtained by the calculation step;
A brain region extraction step of extracting a brain region from the input brain image;
When the distribution of the concentration region of interest is represented by the graphic,
The calculation step includes:
A division step of dividing the brain region of the brain image into a plurality of divided regions based on left-right brain symmetry;
The size of the target density region extracted by the target density region extraction step is measured for each divided region divided by the division step, and a figure having a size representing the size of the measured target density region is measured for each divided region. And a graphic generation step for generating
The display step includes
A dementia diagnosis support method , comprising: a graphic display step of generating and displaying graphic display images in which the graphic generated by the graphic generating step is arranged in a corresponding divided region . A focused density region extraction step of extracting pixels corresponding to a specific density range from the input brain image as a focused density region;
A calculation step of performing a calculation for representing the distribution of the concentration region of interest by a figure, a graph, or a numerical value;
A display step for displaying a figure, a graph, or a numerical value representing the distribution of the concentration region of interest obtained by the calculation step;
A brain region extraction step of extracting a brain region from the input brain image;
When representing the distribution of the concentration region of interest in the graph,
The calculation step includes:
A radial setting step for setting a radial with the center of gravity of the brain region as the origin;
For each point on the radius vector, a profile acquisition step of acquiring a profile representing the relationship between the distance from the origin and the density value over all angles;
Normalizing the profile acquired by the profile acquisition step based on a reference radial length, and generating a graph for the normalized profile; and
With
The display step includes
A graph display step for displaying the graph generated by the graph generation step;
Dementia diagnosis support how to, characterized in that it comprises a.

Images