JP2006043200A - Intracerebral hemorrhage/subarachnoid hemorrhage diagnostic support system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an intracerebral hemorrhage/subarachnoid hemorrhage diagnostic support system for calculating a degree of intracerebral hemorrhage/subarachnoid hemorrhage based on the brain image of a patient. <P>SOLUTION: This intracerebral hemorrhage/subarachnoid hemorrhage diagnostic support system is equipped with an image acquiring device 1 for acquiring the brain image of the patient, a cerebral region detection part 21 for determining a cerebral region from the brain image, a wrinkle range detection part 22 for detecting a region of the brain wrinkle and an uneven region of the brain by determining the difference between a convex closure in the cerebral region and the brain region, a blood vessel removing part 23 for removing a blood vessel region from the wrinkle region and the uneven region of the brain, and a risk calculating part 24 for calculating the degree of bleeding based on the luminosity of the wrinkle region and the uneven region of the brain from which the blood vessel region is removed. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system, and more particularly to an intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system that calculates the degree of intracerebral hemorrhage / subarachnoid hemorrhage based on an image of a brain of a patient.

  Intracerebral hemorrhage and subarachnoid hemorrhage are highly mortal diseases, and sequelae such as involuntary half-body tend to remain. For this reason, an emergency treatment by a doctor is necessary. Prior to this procedure, a diagnosis of intracerebral hemorrhage / subarachnoid hemorrhage is made. This diagnosis is generally performed by performing a CT (Computed Tomography) / MR (Magnetic Resonance) examination of the head and visually checking the result.

As a technique for detecting intracerebral hemorrhage, for example, a technique for detecting intracerebral hemorrhage by displaying information obtained by biological light measurement on a morphological image obtained by an image diagnostic apparatus has been proposed (Patent Document 1). reference).
JP 2001-198112 A

  When the amount of hemorrhage due to intracerebral hemorrhage / subarachnoid hemorrhage is small (in the case of subarachnoid hemorrhage, even if it is very small), it is a problem that it is too late due to a misdiagnosis by a doctor. Such a misdiagnosis is particularly likely to occur in an emergency hospital (such as a medical institution) in which no specialized neurosurgeon is employed. That is, aneurysms are generally less than about 90% concentrated near the branch of the main artery (eg, internal carotid artery), but are otherwise difficult to identify. In the case of a small amount of bleeding, it is difficult to specify the position from the MRI image. On the other hand, once ruptured aneurysms are prone to rebleeding, which further increases the risk of life and remains a severe aftereffect. From the above, there is a demand for diagnostic support in which the risk of intracerebral hemorrhage / subarachnoid hemorrhage is estimated by a computer for each two-dimensional image based on, for example, a two-dimensional MRI image of the patient.

  An object of the present invention is to provide an intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system that calculates the degree of intracerebral hemorrhage / subarachnoid hemorrhage based on an image of the brain of a patient.

  The intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system according to the present invention is based on an image capturing unit that acquires an image of a patient's brain, and the brightness of a brain wrinkle region and a brain non-uniform region in the brain image. A risk degree calculation unit for calculating the degree of bleeding.

  Preferably, in the intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system of the present invention, the bleeding is intracerebral hemorrhage or subarachnoid hemorrhage.

  Preferably, the intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system further includes a cerebral region detection unit for determining a cerebral region from the brain image.

  Preferably, in the intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system of the present invention, the cerebral region detection unit obtains a cerebral region from the brain image by a region expansion method.

  Preferably, the intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system further obtains a difference between the convex hull and the brain region in the cerebral region obtained from the brain image, thereby obtaining the wrinkle region of the brain and the brain A wrinkle region detection unit for detecting a non-uniform region is provided.

  Preferably, the intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system further determines, from the brain image, a cerebral region detection unit that obtains a cerebral region, and a difference between the convex hull and the brain region in the cerebral region, A wrinkle region detection unit for detecting the wrinkle region of the brain and the non-uniform region of the brain.

  Preferably, in the intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system according to the present invention, the brain image of the patient is composed of a plurality of two-dimensional images, and the cerebral region is obtained for each of the two-dimensional images, based on this A wrinkle region of the brain and a non-uniform region of the brain are detected, and the degree of bleeding is calculated based on the brightness of the image in the wrinkle region and non-uniform region of the brain.

  Preferably, the intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system further includes a blood vessel removing unit that removes a blood vessel region from the wrinkle region and a non-uniform region of the brain, and the risk level calculating unit The degree of bleeding is calculated based on the brightness of the removed wrinkle region and the non-uniform brain region.

  Preferably, in the intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system of the present invention, the risk level calculation unit calculates the risk level of bleeding based on the degree of bleeding.

  Preferably, the intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system further includes a display device that displays an image of the brain of the patient, and the display device has a predetermined risk of bleeding for each of the two-dimensional images. A two-dimensional image higher than the value is displayed so as to be specified.

  Experienced neurosurgeons have the experience of the present inventor to focus on the cerebral sulcus (brain wrinkles) for each slice (two-dimensional image) of tomography when diagnosing sub-membrane hemorrhage even if it is microscopic This has been revealed through interviews with neurosurgeons. Therefore, the present inventor considered that the diagnostic accuracy of the presence or absence of intracerebral hemorrhage or subarachnoid hemorrhage can be improved by focusing on the sulcus region rather than evaluating the entire brain region of each image.

  According to the intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system of the present invention, the intracerebral hemorrhage or the subarachnoid hemorrhage is generated, for example, for each two-dimensional image in the brain, based on the luminance of the wrinkle region of the brain and the nonuniform region of the brain. Calculate the degree of bleeding. At this time, for example, an image of a patient's brain is acquired, a cerebral area is obtained, a wrinkle area and a non-uniform area of the brain are detected by obtaining a difference between the convex hull and the brain area in the cerebral area, and if necessary The blood vessel area is removed from the wrinkle area and the uneven brain area, and the degree of bleeding is calculated based on the brightness of the image thus obtained, or the risk of bleeding is calculated, and the risk of bleeding is calculated. A two-dimensional image having a degree higher than a predetermined value is displayed so as to be specified.

  As a result, even in the case of aneurysms other than the vicinity of the branch of the main artery (for example, internal carotid artery) that is difficult to identify based on the image of the patient's brain (for example, two-dimensional MRI image), The risk of intracerebral hemorrhage / subarachnoid hemorrhage can be estimated by a computer for each two-dimensional image in the brain. As a result, diagnosis of intracerebral hemorrhage / subarachnoid hemorrhage can be supported, for example, in a medical institution where a specialized neurosurgeon is not working, and misdiagnosis in the hospital or the like can be prevented. Therefore, even in the case of minute intracerebral hemorrhage / subarachnoid hemorrhage, it is possible to prevent a doctor from making a misdiagnosis and avoid the treatment being too late.

  FIG. 1 is a configuration diagram of a diagnosis support system for intracerebral hemorrhage / subarachnoid hemorrhage and shows a configuration of a diagnosis support system for intracerebral hemorrhage / subarachnoid hemorrhage according to the present invention.

  The intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system of the present invention includes an image acquisition device 1, a diagnosis support device 2, and a display device 3. As described above, the intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system or diagnosis support apparatus 2 according to the present invention provides the diagnosis support for the minute intracerebral hemorrhage / subarachnoid hemorrhage in the same manner as an experienced neurosurgeon. Attention is focused on the cerebral sulcus (brain wrinkles) in each tomographic image (cross-sectional image), that is, a slice (two-dimensional image) in the (tomographic) image.

  The image acquisition device 1 is a device that acquires an image of a patient's brain, and in this example, is a known MR device. As a result, a brain image (image data) in a known data format (for example, DICOM) is obtained from the image acquisition device 1. The image acquisition apparatus 1 may be an apparatus that can obtain a two-dimensional image (planar image) of the human brain by well-known tomography or the like, such as an X-ray CT apparatus or a positron CT apparatus. DICOM is a data format mainly used in MR apparatus and CT apparatus. By following DICOM, not only image data but also related information such as patient information and photographing date / time can be obtained.

  The diagnosis support apparatus 2 receives a brain image (image data) according to a predetermined data format (for example, DICOM) from the image acquisition apparatus 1 and holds it in an image memory (not shown). The diagnosis support apparatus 2 and the image acquisition apparatus 1 may be directly connected by a cable or the like, or may be connected via a network such as a wireless or wired LAN (Local Area Network) or the Internet. For example, the image acquisition device 1 may be installed in a plurality of hospitals without specialized neurosurgeons, the diagnosis support device 2 may be installed in a hospital with specialized neurosurgeons, and these may be connected via a network. In this case, the display device 3 may be provided in both a hospital without a neurosurgeon and a hospital with a neurosurgeon.

  The diagnosis support apparatus 2 includes a cerebral region detection unit 21, a wrinkle region detection unit 22, a blood vessel removal unit 23, and a risk level calculation unit 24. Each of these processing units 21 to 24 executes a processing program for executing the processing resident on a main memory (not shown) of the diagnosis support apparatus 2 which is a computer on a CPU (not shown) of the diagnosis support apparatus 2. It is realized by executing with.

  The cerebral region detection unit 21 obtains a “cerebral region” from the brain image (image data) held in the image memory by a known means. In this example, the cerebral region detection unit 21 obtains a cerebral region from a brain image by, for example, a known region growing method. The region expansion method is a method of dividing a region by sequentially executing the process of integrating into one region when the small region of interest and the small region (or pixel) adjacent to it have the same characteristics. It is. In this example, “same feature” means that the pixels in the region are similar in the image data. Specifically, when the image data is a monochrome image of, for example, 256 gradations, It is necessary that the values are similar and that their positions are close to each other, and the range can be determined empirically.

  The wrinkle region detection unit 22 obtains a difference between the “convex hull” and the “brain region” in the cerebral region obtained from the brain image, thereby obtaining a “brain wrinkle region” and a “brain non-uniform region”. Is detected. In other words, the remaining regions after excluding the region that is the brain region (overlapping the brain region) from the region that is the convex hull are the wrinkle region of the brain and the non-uniform region of the brain. Here, the brain region refers to a region (uniform region) having similar pixel values and a group, and does not include a region such as “hippocampus”, for example. Accordingly, the non-uniform region of the brain refers to a region such as the hippocampus. In this way, by detecting not only the wrinkled region of the brain but also the non-uniform region of the brain, it is possible to detect a portion that may bleed without omission, so that the degree of bleeding and the degree of risk can be determined. An erroneous result can be prevented.

  To find the convex hull, we used 2D Active Net, a two-dimensional deformation model. 2D Active Net (exactly, “network model using energy minimization principle”) gives a circular network as the initial shape, and the network approaches the part to be extracted by repeating the process based on the energy minimum principle. That's it. As for 2D Active Net, for example, “Sakagami, Yamamoto” Dynamic Network Model Active Net and its Application to Region Extraction, Journal of Television Society, Vol. 45, No. 10, pp. 1153-1163, (1991) "And" Shun Ohara, Toshiro Fujiwara, Koichi Matsuda, Akio Doi "Active Net model considering pressure energy and its application" IPSJ Graphics and CAD Summer Research Group, pp1-4, 2001/9 " . Specifically, by applying 2D Active Net to the brain region (uniform region, in this example, the cerebral region obtained earlier) as an energy image, the region where the brain region is bundled with rubber bands, that is, The (approximate) convex hull is determined. Then, the “brain wrinkle region” and “brain non-uniform region” can be detected by obtaining the difference between the inside of the region and the brain region.

  An example of the application of 2D Active Net is shown in FIG. In FIG. 2, FIG. 2 (A) is an initial image and shows a brain region (uniform region). FIG. 2B shows an image when the process based on the principle of minimum energy is repeated 400 times. FIG. 2C shows an image when the above process is repeated 800 times. FIG. 2D shows an image when the above process is repeated 2000 times. It can be seen from FIG. 2D that “brain wrinkle region” and “brain non-uniform region” have been detected.

  Note that the convex hull obtained by using 2D Active Net is not a convex hull in a strict sense. That is, the convex hull obtained by using 2D Active Net partially bites into the actual convex hull, and therefore the proximity value of the convex hull (region) is obtained.

  The blood vessel removing unit 23 removes the “blood vessel region” from the wrinkled region of the brain and the non-uniform region of the brain. The reason for removing the blood vessel region is that a blood vessel (a round tube in a tomographic slice) becomes very white (high brightness) in the case of an MRI image, so this portion should be removed first. This is because the accuracy of detection of bleeding can be improved. In the case of an image obtained by a CT apparatus, this does not occur, and therefore it is not necessary to remove the blood vessel region.

  As described above, the blood vessel region has a round shape with high brightness in the slice of tomography. In the case of an MRI image, since such a high-luminance region is only a blood vessel region, a region having a value larger than the threshold value can be removed as a blood vessel region using a predetermined threshold value. That is, the pixel value of the image data of the blood vessel region is changed to a value that is the same as or similar to the pixel value of the surrounding region (or an average value thereof). As a result, a region obtained by removing the “blood vessel region” from the wrinkled region of the brain and the non-uniform region of the brain, that is, the “region of the wrinkle of the brain” and the “remaining region where the luminance value is not uniform” is obtained. The “remaining region with non-uniform luminance values” is a portion obtained by subtracting a pixel region that is substantially uniform and connected (connected) from a brain region (or cerebral region).

  The risk level calculation unit 24 calculates the degree of bleeding based on the luminance of the image (original image, in this case, MRI image) in the wrinkled region of the brain and the non-uniform region of the brain. calculate. Therefore, the bleeding for which the degree of bleeding is calculated is intracerebral hemorrhage or subarachnoid hemorrhage. Specifically, in this example, the risk degree calculation unit 24 creates the degree of bleeding as a histogram as shown in FIG. For example, when the image data is a monochrome image of, for example, 256 gradations, a histogram is created by counting the number of pixels having each pixel value in the region.

  In this example, actually, the risk level calculation unit 24 calculates the degree of bleeding based on the luminance of the wrinkled region of the brain from which the blood vessel region has been removed and the uneven region of the brain (creates a histogram). ). In these areas, in the case of an MRI image, a non-bleeding area is a black pixel, and a bleeding area is a white pixel. According to the present invention, it is possible to consider the case where blood does not flow into the wrinkled region of the brain even if there is bleeding, by calculating the degree of bleeding for the non-uniform region (internal region) of the brain. . As a result, it is possible to prevent erroneous determination of bleeding as described above.

  In this example, the risk level calculation unit 24 further calculates the risk level of bleeding based on the calculated bleeding level (histogram). Accordingly, the risk of intracerebral hemorrhage or subarachnoid hemorrhage is calculated. Thereby, when displaying the result of analyzing the image of the patient's brain for the presence or absence of bleeding, the analysis result can be displayed more easily. At this time, as will be described later, a threshold for calculating (determining) the degree of risk is input from the outside of the diagnosis support apparatus 2. In this example, the threshold value is determined with reference to the aforementioned histogram.

  Note that the threshold is set by inputting from the outside of the diagnosis support apparatus 2 each time based on the result of creating the histogram, but this threshold is automatically set based on the result of creating the histogram. You may make it do. For example, the risk calculating unit 24 adopts the threshold when the histogram has a boundary as shown in FIG. 5 as the threshold, and the distribution when the histogram has no boundary as shown in FIG. The rightmost value may be adopted.

  In practice, an image of a patient's brain is composed of a plurality of continuous two-dimensional images obtained by tomography. Accordingly, for each two-dimensional image, a cerebral region is obtained, and a wrinkle region and a non-uniform region of the brain are detected based on the cerebral region, and the wrinkle region and non-uniform region of the brain are detected as necessary. The blood vessel region is removed, the degree of bleeding is calculated based on the luminance of the wrinkled region of the brain and the non-uniform region of the brain, and the risk of bleeding is calculated.

  The display device 3 displays, for example, for each two-dimensional image, a two-dimensional image in which the risk of intracerebral hemorrhage / subarachnoid hemorrhage is higher than a predetermined value. Thereby, determination can be performed for each two-dimensional image, and an indication of a part at risk of bleeding can be displayed. In this example, as will be described later, since the presence or absence of bleeding can be determined only by the shape of the histogram, the display device 3 displays the calculated degree of cerebral hemorrhage / subarachnoid hemorrhage (histogram) together with the degree of risk. Any one of the displays may be omitted.

  Further, as will be described later, since it is possible to specify a bleeding site in units of pixels using a histogram, the brain in which the image acquisition device 1 has acquired the pixel that the risk level calculation unit 24 has specified as the bleeding site. An image superimposed on this image (two-dimensional image) may be created and displayed on the display device 3. At this time, the pixel identified as the bleeding site may be displayed in red, for example.

  FIG. 3 is an intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support processing flow, and shows an example of intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support processing in the intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system of the present invention. In FIG. 2, the image acquisition device 1 acquires an MRI image (two-dimensional image) of the patient's brain (step S <b> 1) and transmits it to the diagnosis support device 2. In the diagnosis support apparatus 2 that has received this, the cerebral region detection unit 21 obtains the cerebral region of the patient's brain using the region expansion method (step S2), and the wrinkle region detection unit 22 uses the convex hull in the obtained cerebral region. The brain wrinkle region and the brain non-uniform region are detected by calculating the difference between the brain wrinkle region and the brain non-uniform region (step S3), and the blood vessel removing unit 23 detects the brain wrinkle region and the brain non-uniform region from the blood vessel region. (Step S4), the risk calculation unit 24 calculates the degree (histogram) of intracerebral hemorrhage / subarachnoid hemorrhage based on the luminance in the wrinkle region of the brain and the non-uniform region of the brain, The risk of intracerebral hemorrhage / subarachnoid hemorrhage is calculated (step S5), and the calculation result is transmitted to the display device 3. Receiving this, the display device 3 displays the calculation result (histogram and risk) for each two-dimensional image (step S6).

  4 to 7 show the actual diagnosis support for intracerebral hemorrhage / subarachnoid hemorrhage in the intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system of the present invention.

  FIG. 4A is an original image of the brain of the patient read by the MR apparatus that is the image acquisition apparatus 1. The image size is 460 dots × 460 dots. This original image (MRI image) is image data captured on the next day for a patient who actually developed subarachnoid hemorrhage.

  FIG. 4B is an image after segmenting the cerebral region using the region expansion method. As the parameters at this time, the seed point was selected interactively, the global parameter was set to 20, and the local parameter was set to 10.

  By applying 2D Active Net to the image of FIG. 4B, an image of a converged net area can be obtained as shown in FIG. From the image of FIG. 4C, a convex hull of the brain region is obtained as shown in FIG. The parameters at this time were 30 × 100 for the mesh shape and 2000 iterations, and α, β, and γ were, for example, 1.0, 1.0, and 0.5, respectively. The 2D Active Net program is written in OpenGL, and the image of the convex hull shown in FIG. 4D was obtained by painting the inside of the image net as the application result.

  An image obtained by taking the difference between the convex hull image (image in FIG. 4D) and the cerebral region image (image in FIG. 4B) is the image in FIG. By this processing, the “region including the wrinkle part of the brain” (that is, the wrinkle region of the brain and the non-uniform region of the brain) can be extracted.

  Next, the presence / absence of bleeding (and its degree) is estimated from the pixel value of the MRI image corresponding to the region including the wrinkle portion. FIG. 5 is a histogram of the region (that is, all pixels of the difference mask) obtained in FIG. In FIG. 5, the horizontal axis represents pixel values (1 to 256 gradations, but one part is omitted), and the vertical axis represents the number of pixels having the pixel values (same in FIG. 7). As shown in FIG. 5, it can be seen that a portion with a high luminance value (ie, a bleeding portion) and a portion with a low luminance (ie, a normal portion) are separated. Further, since the portion with a high luminance value is a bleeding portion, the right peak of the two peaks is the bleeding portion, and the pixel value is approximately “at 256 gradations” from FIG. If it is about 120 ", it turns out that it is bleeding. Therefore, by using the pixel value as a threshold value, it is possible to easily specify the bleeding site in pixel units.

  Next, the total number of pixels of the difference mask is used as the denominator, and the number of pixels higher than the threshold value as the numerator is calculated as the “risk level”. In this example, the boundary of the histogram is the threshold value (90). As a result, the degree of risk in this image (image in FIG. 4A) is “0.489476”. The total number of pixels was “20904”, and the number of pixels higher than the threshold was “10232”.

  Here, for comparison, when the present invention is applied to an MRI image of a healthy person, the results are as shown in FIGS. That is, an image of the cerebral region shown in FIG. 6B is obtained from the original brain image (image size is 256 dots × 256 dots) read by the MR apparatus shown in FIG. 6D is obtained by applying 2D Active Net, and an image of a convex hull (not shown) is obtained based on this to obtain an image of the cerebral area (image of FIG. 6B). The image obtained by taking the difference from () is the image of FIG. As described above, the histogram of the region obtained in FIG. 6D is obtained as shown in FIG. In the histogram of FIG. 7, since there is no bleeding portion, the luminance distribution is concentrated in one place, and the luminance is low, which is clearly different from FIG. Thus, at the time when the above-mentioned histogram is taken, it can be seen that the presence or absence of bleeding can be determined to some extent from the graph.

  Since there is no boundary in the histogram of FIG. 7, the right end portion of the distribution is used as a threshold value (100) as a guide. As a result, the danger level was “0.029474”. That is, it can be seen that the risk is about 1/16. Thus, the presence / absence of intracerebral hemorrhage / subarachnoid hemorrhage can be determined sufficiently from the risk level.

  As described above, according to the present invention, in the intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system, focusing on the cerebral sulcus (brain wrinkles) as in the experienced neurosurgeon, Based on the brightness of the image in the region, the degree of hemorrhage such as intracerebral hemorrhage / subarachnoid hemorrhage, or the risk of bleeding is calculated. The risk of subarachnoid hemorrhage can be estimated by a computer. For example, in a hospital where a specialized neurosurgeon is not working, diagnosis of intracerebral hemorrhage / subarachnoid hemorrhage can be supported. Even in the case of subarachnoid hemorrhage, misdiagnosis by a doctor can be prevented, and it can be avoided that treatment is too late.

1 is a configuration diagram of an intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system. FIG. FIG. 6 is an explanatory diagram of intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support. It is an intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support processing flow. FIG. 6 is an explanatory diagram of intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support. FIG. 6 is an explanatory diagram of intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support. FIG. 6 is an explanatory diagram of intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support. FIG. 6 is an explanatory diagram of intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Image acquisition apparatus 2 Diagnosis assistance apparatus 3 Display apparatus 21 Cerebral area | region detection part 22 Wrinkle area | region detection part 23 Blood vessel removal part 24 Risk degree calculation part

Claims (10)

An image capturing unit for acquiring an image of a patient's brain;
A diagnosis support system for intracerebral hemorrhage / subarachnoid hemorrhage, comprising: a risk level calculation unit that calculates the degree of bleeding based on the luminance of a wrinkle region and a non-uniform region of the brain in the brain image . The intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system according to claim 1, wherein the hemorrhage is intracerebral hemorrhage or subarachnoid hemorrhage. The intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system further includes:
The intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system according to claim 1, further comprising a cerebral region detection unit that obtains a cerebral region from the brain image. The intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system according to claim 3, wherein the cerebral region detection unit obtains a cerebral region from the brain image by a region expansion method. The intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system further includes:
A wrinkle region detection unit that detects the wrinkle region of the brain and the non-uniform region of the brain by obtaining a difference between the convex hull and the brain region in the cerebral region obtained from the brain image, The intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system according to claim 1. The intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system further includes:
From the image of the brain, a cerebral region detection unit for obtaining a cerebral region;
The wrinkle region detection unit that detects the wrinkle region of the brain and the non-uniform region of the brain by obtaining a difference between the convex hull and the brain region in the cerebral region. Diagnosis support system for intracerebral hemorrhage / subarachnoid hemorrhage. The patient's brain image consists of a plurality of two-dimensional images,
For each of the two-dimensional images, the cerebrum region is obtained, and a wrinkle region of the brain and a non-uniform region of the brain are detected based on the cerebral region. The intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system according to claim 6, wherein the degree of bleeding is calculated based on luminance. The intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system further includes:
A blood vessel removing unit that removes a blood vessel region from the wrinkled region and the non-uniform region of the brain,
The degree of bleeding is calculated based on the brightness of the wrinkle region from which the blood vessel region has been removed and the non-uniform region of the brain. The diagnosis support system for intracerebral hemorrhage / subarachnoid hemorrhage described in 1. The intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system according to claim 1, wherein the risk calculation unit calculates a risk of bleeding based on the degree of bleeding. The intracerebral hemorrhage / subarachnoid hemorrhage diagnosis support system further includes:
A display device for displaying an image of the patient's brain;
The intracerebral hemorrhage / subarachnoid hemorrhage according to claim 9, wherein the display device displays, for each of the two-dimensional images, a two-dimensional image in which the risk of bleeding is higher than a predetermined value. Diagnosis support system.