Intracranial hemorrhage anatomy auxiliary positioning method and system based on CT image

Abstract

The invention discloses an intracranial hemorrhage anatomy auxiliary positioning method and system based on a CT image, and relates to the field of medical imagines.The main scheme is that a hemorrhage part probability model is trained based on lateral ventricle data, third ventricle data and fourth ventricle data corresponding to different CT images; obtaining bleeding probabilities of different ventricles of the patient to be diagnosed according to the bleeding part probability model, judging the bleeding risk level of each ventricle, and judging whether each region is a bleeding region or not according to the bleeding region judgment index HEe and the bleeding region judgment threshold SU when it is judged that any ventricle has a high bleeding risk; the problems that individual differences exist in reactions of different patients to intracranial hemorrhage, and when multiple intracranial hemorrhage parts exist, symptoms and signs caused by hemorrhage of different parts can be overlapped or mixed, so that the judgment time is prolonged, and treatment of the patients is affected are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical imaging, and specifically to an anatomical assisted positioning method and system for intracranial hemorrhage based on CT images. Background Art

[0002] When judging the location of intracerebral hemorrhage, by analyzing CT images and using advanced image processing techniques and anatomical knowledge, the specific area of intracranial hemorrhage can be accurately identified, the hemorrhage focus can be distinguished from the surrounding normal brain tissue, the positioning deviation caused by image interpretation errors can be reduced, the bleeding point can be quickly located, and the subjective error of doctors can be reduced.

[0003] The traditional method for positioning intracerebral hemorrhage is based on the fact that intracerebral hemorrhage in different parts will cause different clinical symptoms and signs. By examining the clinical symptoms and signs of patients, including the state of consciousness, pupil changes, limb movement and sensory disorders, pathological reflexes, etc., the location of intracranial hemorrhage is inferred.

[0004] The existing technology still has the following deficiencies: there are individual differences in the responses of different patients to intracranial hemorrhage. Some patients may not show typical symptoms and signs, or the symptoms and signs are relatively mild and vague. At the same time, when there are multiple intracranial hemorrhages, the symptoms and signs caused by hemorrhages in different parts may overlap or be confused with each other, and the observation area and the difficulty of observation are increased, making it difficult to quickly lock the suspicious area where the bleeding point is located according to the CT image. Summary of the Invention

[0005] (1) Technical Problems to be Solved Aiming at the deficiencies of the existing technology, the present invention provides an anatomical assisted positioning method and system for intracranial hemorrhage based on CT images. The probability model of the hemorrhage location is trained based on the lateral ventricle data, third ventricle data, and fourth ventricle data corresponding to the CT images of different patients. The probability of lateral ventricle hemorrhage PA, the probability of third ventricle hemorrhage PB, and the probability of fourth ventricle hemorrhage PC to be diagnosed are obtained according to the output of the hemorrhage location probability model, and the hemorrhage risk levels of each ventricle are judged in combination with the hemorrhage threshold set. When it is determined that there is a high hemorrhage risk in any ventricle, according to the hemorrhage area determination index HE e and the hemorrhage area determination threshold SU to judge whether each area is a hemorrhage area, solving the problems that there are individual differences in the responses of different patients to intracranial hemorrhage. Some patients may not show typical symptoms and signs, or the symptoms and signs are relatively mild and vague. At the same time, when there are multiple intracranial hemorrhages, the symptoms and signs caused by hemorrhages in different parts may overlap or be confused with each other, the observation area range and the difficulty of observation are large, and it is difficult to quickly lock the suspicious area where the bleeding point is located according to the CT image.

[0006] (2) Technical Solutions To achieve the above objectives, the present invention is implemented through the following technical solutions: an anatomical auxiliary positioning system for intracranial hemorrhage based on CT images, comprising: The data acquisition module can obtain the corresponding lateral ventricle data, third ventricle data and fourth ventricle data based on the CT images of different patients' medical records; The index calculation module can calculate the left lateral ventricle change rate XA, the right lateral ventricle change rate XB, the lateral ventricle anterior horn bleeding index WA, the body bleeding index WB, the posterior horn bleeding index WC and the lower horn bleeding index WD based on the lateral ventricle data; calculate the third ventricle change rate RB based on the third ventricle data; calculate the fourth ventricle change rate RC and the hematoma position index WF based on the fourth ventricle data; The model training module builds an initial model based on a convolutional neural network; the CT images of different patients' medical records and the corresponding lateral ventricle data, third ventricle data, fourth ventricle data and all data values ​​obtained by the indicator calculation module are used as input data items, and the diagnosis results of whether each ventricle in the corresponding patient's medical record is bleeding are used as output items to train and optimize the initial model. After the training is completed, a probability model of the bleeding site is formed; The model judgment module can input the CT image of the patient to be diagnosed, the corresponding lateral ventricle data, the third ventricle data, the fourth ventricle data and the data values ​​calculated by the index calculation module into the bleeding site probability model, and the bleeding site probability model outputs the lateral ventricle bleeding probability PA, the third ventricle bleeding probability PB and the fourth ventricle bleeding probability PC; preset a bleeding threshold set; judge the bleeding risk level of each ventricle according to the lateral ventricle bleeding probability PA, the third ventricle bleeding probability PB, the fourth ventricle bleeding probability PC and the bleeding threshold set; The positioning module calculates the hemorrhage area determination index HE in different areas of the ventricle based on the corresponding ventricle data when any ventricle is judged to have a high risk of hemorrhage e ; Preset bleeding area determination threshold SU; According to the bleeding area determination index HE e and the bleeding area determination threshold SU to determine whether each area is a bleeding area.

[0007] In the preferred embodiment of the CT image-based intracranial hemorrhage anatomical auxiliary positioning system, the method for calculating the left lateral ventricle change rate XA and the right lateral ventricle change rate XB is: The lateral ventricle data included the left lateral ventricle volume AA, the right lateral ventricle volume AB, the normal lateral ventricle volume AC, the left lateral ventricle average CT value TA, the right lateral ventricle average CT value TB and the normal cerebrospinal fluid CT value TC; The left lateral ventricle change rate XA was calculated based on the left lateral ventricle volume AA, the normal lateral ventricle volume AC, the left lateral ventricle average CT value TA and the normal cerebrospinal fluid CT value TC. The formula is: ; The right lateral ventricle change rate XB was calculated based on the right lateral ventricle volume AB, the normal lateral ventricle volume AC, the right lateral ventricle average CT value TB and the normal cerebrospinal fluid CT value TC. The formula is: .

[0008] In the preferred embodiment of the intracranial hemorrhage anatomical auxiliary positioning system based on CT images, the method for calculating the lateral ventricle anterior horn hemorrhage index WA, body hemorrhage index WB, posterior horn hemorrhage index WC and lower horn hemorrhage index WD is as follows: The lateral ventricle data also include the anterior horn angle value DA, anterior horn cross-sectional area DB, normal anterior horn angle value DC, normal anterior horn cross-sectional area DD, corpus callosum thickness DE, thalamus coordinate value (DF, DG), normal corpus callosum thickness DH, normal thalamus coordinate value (DI, DJ), posterior horn cross-sectional area DK, posterior horn length DL, normal posterior horn cross-sectional area DM, normal posterior horn length DN, inferior horn volume DQ and normal inferior horn volume DP of the hemorrhage side lateral ventricle; The anterior horn hemorrhage index WA is calculated based on the anterior horn angle value DA, the anterior horn cross-sectional area DB, the normal anterior horn angle value DC and the normal anterior horn cross-sectional area DD. The formula is: ; The body hemorrhage index WB is calculated based on the corpus callosum thickness DE, thalamus coordinate values ​​(DF, DG), normal corpus callosum thickness DH, and normal thalamus coordinate values ​​(DI, DJ). The formula is: ; The posterior angle bleeding index WC is calculated based on the posterior angle cross-sectional area DK, the posterior angle length DL, the normal posterior angle cross-sectional area DM and the normal posterior angle length DN. The formula is: ; The inferior angle bleeding index WD is calculated based on the inferior angle volume DQ and the normal inferior angle volume DP, and the formula is: .

[0009] In the preferred embodiment of the CT image-based intracranial hemorrhage anatomical auxiliary positioning system, the method for calculating the third ventricle change rate RB is: The third ventricle data include the third ventricle blood volume BA, the third ventricle area BB, the normal third ventricle volume BC and the average normal third ventricle area BE; The third ventricle change rate RB is calculated based on the third ventricle data according to the following formula: ; Among them, 1 for The weight coefficient is 0.3~0.7; 2 for The weight coefficient is 0.3~0.7; and 1 +ɑ 2 =1.

[0010] In the preferred embodiment of the CT image-based intracranial hemorrhage anatomical auxiliary positioning system, the method for calculating the fourth ventricle change rate RC is: The fourth ventricle data include the fourth ventricle long diameter CA, the fourth ventricle short diameter CB, the normal fourth ventricle long diameter CC and the normal fourth ventricle short diameter CD; The fourth ventricle change rate RC is calculated based on the fourth ventricle data, and the formula is: .

[0011] In the preferred embodiment of the intracranial hemorrhage anatomical auxiliary positioning system based on CT images, the method for calculating the hematoma position index WF is: The fourth ventricle data also include the maximum projection area FA of the hematoma, the minimum projection area FB of the hematoma, the volume FC of the hematoma located in the top area of ​​the fourth ventricle, and the total volume FD of the hematoma; The hematoma location index WF is calculated based on the maximum projection area FA of the hematoma, the minimum projection area FB of the hematoma, the volume FC of the hematoma located at the top of the fourth ventricle, and the total volume FD of the hematoma. The formula is: ; Among them, γ 1 for The weight coefficient is 0.4 to 0.6; 2 for The weight coefficient is 0.4 to 0.6; and γ 1 +γ 2 =1.

[0012] In the preferred embodiment of the intracranial hemorrhage anatomical auxiliary positioning system based on CT images, the method for determining the hemorrhage risk level of each ventricle is: The bleeding threshold set includes the lateral ventricle bleeding threshold SA, the third ventricle bleeding threshold SB and the fourth ventricle bleeding threshold SC; The formula for judging whether there is bleeding risk in each part is based on the probability of lateral ventricle hemorrhage PA, the probability of third ventricle hemorrhage PB, the probability of fourth ventricle hemorrhage PC and the bleeding threshold set: ; ; .

[0013] In the preferred embodiment of the above-mentioned intracranial hemorrhage anatomical auxiliary positioning system based on CT images: the hemorrhage area determination index HE is calculated e The method is: Ventricular data including actual ventricular volume HC e 、Normal ventricular volume HD e , actual CT value CTA e 、 Normal CT value CTB e , average displacement distance of peripheral blood vessels CTC e and vascular disruption ratio (CTD) e ; Calculate the bleeding area determination index HE based on the bleeding data e , the formula based on is: ; Among them, HE e is the hemorrhage area determination index of the e-th region in the brain area; e is the serial number corresponding to different regions, and the value is a positive integer; HC e is the actual ventricular volume of the eth region in the brain region; HD e is the normal ventricular volume of the eth region in the brain; CTA e is the actual CT value of the e-th area in the brain; CTB e is the normal CT value of the eth area in the brain; CTC e is the average displacement distance of the peripheral blood vessels in the e-th region of the brain; CTD e is the proportion of vascular interruption in the e-th region of the brain; β 1 for The weight coefficient is 0.2 to 0.5; β 2 for The weight coefficient is 0.3~0.5; β 3 for The weight coefficient is 0.2 to 0.5; and β 1 +β 2 +β 3 =1.

[0014] In the preferred embodiment of the CT image-based intracranial hemorrhage anatomical auxiliary positioning system, the method for determining whether each area is a hemorrhage area is as follows: HE index is determined based on the bleeding area e The bleeding area determination threshold SU is used to determine whether bleeding exists in each area, and the formula is as follows: .

[0015] The present invention also discloses a method for assisting anatomical positioning of intracranial hemorrhage based on CT images, comprising the following steps: Based on the CT images of different patients' medical records, the corresponding lateral ventricle data, third ventricle data, and fourth ventricle data were obtained; Based on the lateral ventricle data, the left lateral ventricle change rate XA, the right lateral ventricle change rate XB, the lateral ventricle anterior horn bleeding index WA, the body bleeding index WB, the posterior horn bleeding index WC and the inferior horn bleeding index WD were calculated; based on the third ventricle data, the third ventricle change rate RB was calculated; based on the fourth ventricle data, the fourth ventricle change rate RC and the hematoma location index WF were calculated; An initial model is constructed based on a convolutional neural network; CT images of different patients' medical records and the corresponding lateral ventricle data, third ventricle data, fourth ventricle data, and all data values ​​obtained by the index calculation module are used as input data items, and the diagnosis results of whether each ventricle in the corresponding patient's medical record is bleeding are used as output items, and the initial model is trained and optimized. After the training is completed, a bleeding site probability model is formed; Input the CT image of the patient to be diagnosed, the corresponding lateral ventricle data, the third ventricle data, the fourth ventricle data and the data values ​​calculated by the index calculation module into the bleeding site probability model, and the bleeding site probability model outputs the lateral ventricle bleeding probability PA, the third ventricle bleeding probability PB and the fourth ventricle bleeding probability PC; preset a bleeding threshold set; determine the bleeding risk level of each ventricle according to the lateral ventricle bleeding probability PA, the third ventricle bleeding probability PB, the fourth ventricle bleeding probability PC and the bleeding threshold set; When any ventricle is judged to have a high risk of bleeding, the hemorrhage area determination index HE of different areas of the ventricle is calculated based on the corresponding ventricle data. e ; Preset bleeding area determination threshold SU; According to the bleeding area determination index HE e and the bleeding area determination threshold SU to determine whether each area is a bleeding area.

[0016] (III) Beneficial effects The present invention provides an anatomical auxiliary positioning method for intracranial hemorrhage based on CT images, which has the following beneficial effects: (1) By obtaining the data of the lateral ventricle, the third ventricle, and the fourth ventricle through the patient's CT images, the relationship between the bleeding location and the ventricle can be clarified, providing a basis for determining the treatment plan.

[0017] (2) The left lateral ventricle change rate XA, the right lateral ventricle change rate XB, the lateral ventricle anterior horn hemorrhage index WA, the body hemorrhage index WB, the posterior horn hemorrhage index WC, and the inferior horn hemorrhage index WD are calculated based on the lateral ventricle data. The third ventricle change rate RB is calculated based on the third ventricle data. This allows for timely detection of abnormalities in the early stages of third ventricle hemorrhage, when hemorrhage may only cause minor changes in the morphology of the ventricle. The fourth ventricle change rate RC and the hematoma location index WF are calculated based on the fourth ventricle data, which can more objectively and accurately reflect the degree of morphological change of the fourth ventricle and provide a more accurate basis for disease assessment.

[0018] (3) By training the probability model of bleeding sites, the brain condition can be reflected from multiple dimensions, and the risk status of each ventricle can be quickly determined, which is conducive to improving the efficiency of diagnosis.

[0019] (4) Calculate the hemorrhage area determination index HE in different ventricular areas based on the corresponding ventricular data e , can more accurately reflect the actual bleeding possibility of each area, and determine the index HE according to the bleeding area e The bleeding area determination threshold SU determines whether each area is a bleeding area, which can timely detect potential bleeding risks, provide positioning assistance for doctors, reduce the observation area, and provide a basis for early intervention and treatment of patients. At the same time, it provides a clear and unified standard for bleeding diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 The figure is a schematic diagram of the system composition of the CT image-based intracranial hemorrhage anatomical auxiliary positioning system of the present invention. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] See also Figure 1 The present invention provides an intracranial hemorrhage anatomical auxiliary positioning system based on CT images, comprising: The data acquisition module can obtain the corresponding lateral ventricle data, third ventricle data and fourth ventricle data based on the CT images of different patients' medical records.

[0023] In the above scheme, the data of the lateral ventricle, the third ventricle and the fourth ventricle are obtained through the patient's CT images, which can clarify the relationship between the bleeding location and the ventricle, and provide a basis for determining the treatment plan.

[0024] The indicator calculation module can calculate the left lateral ventricle change rate XA, the right lateral ventricle change rate XB, the lateral ventricle anterior corner bleeding index WA, the body bleeding index WB, the posterior corner bleeding index WC and the inferior corner bleeding index WD based on the lateral ventricle data; calculate the third ventricle change rate RB based on the third ventricle data; calculate the fourth ventricle change rate RC and the hematoma position index WF based on the fourth ventricle data.

[0025] Specifically, the method for calculating the left lateral ventricle change rate XA and the right lateral ventricle change rate XB is: The lateral ventricle data include the left lateral ventricle volume AA, the right lateral ventricle volume AB, the normal lateral ventricle volume AC, the left lateral ventricle average CT value TA, the right lateral ventricle average CT value TB and the normal cerebrospinal fluid CT value TC.

[0026] It should be noted that the left lateral ventricle volume AA and the right lateral ventricle volume AB represent the size of the internal space of the left lateral ventricle and the right lateral ventricle of the brain respectively, and the unit is milliliter or cubic centimeter. This indicator is of great significance for evaluating whether the ventricular system is normal. It is obtained by analyzing the CT images using medical image analysis software such as TotalSegmentator and LiviaNET that can use automatic segmentation algorithms. The normal lateral ventricle volume AC refers to the standard volume range that the lateral ventricle should have under normal physiological conditions, and the unit is milliliter or cubic centimeter. The acquisition method is: collect a large number of brain CT image data from healthy people with no brain diseases and complete nervous system development, use TotalSegmentator, LiviaNET and other software to obtain the lateral ventricle volume data of this population, add these data up and divide them by the number of data, as the normal lateral ventricle volume AC. The average CT value TA of the left lateral ventricle indicates the average density of cerebrospinal fluid in the left lateral ventricle on the CT image. The CT value is an indicator used to measure tissue density in CT images. The unit is Hounsfield unit. The acquisition method is: use medical image processing software such as BRAINSABC and 3D-Slicer to select ROI and measure CT values ​​for all layers containing the left lateral ventricle in the CT image. The measured CT values ​​are accumulated and divided by the number of data as the average CT value TA of the left lateral ventricle. Among them, ROI selection refers to drawing areas on each layer of the CT image. For the left lateral ventricle, the ROI needs to accurately cover the cerebrospinal fluid part of the lateral ventricle, while avoiding the ventricular wall, surrounding brain tissue and pathological tissue, such as hemorrhage, tumor Tumor, etc. The average CT value TB of the right lateral ventricle represents the average density of cerebrospinal fluid in the right lateral ventricle on CT images. Medical image processing software such as BRAINSABC and 3D-Slicer are used to select ROIs and measure CT values ​​for all layers containing the right lateral ventricle in the CT images. The measured CT values ​​are accumulated and divided by the number of data as the average CT value TB of the right lateral ventricle. The normal cerebrospinal fluid CT value TC refers to the density value range of cerebrospinal fluid on CT images under normal circumstances. This value is relatively stable. It is an important reference standard for judging whether there are abnormal density substances in or around the lateral ventricle. It is obtained by referring to medical literature such as "CT Scan to Evaluate Intracranial Cerebrospinal Fluid Content" and "Normal Brain CT Manifestations".

[0027] The left lateral ventricle change rate XA was calculated based on the left lateral ventricle volume AA, the normal lateral ventricle volume AC, the left lateral ventricle average CT value TA and the normal cerebrospinal fluid CT value TC. The formula is: ; It should be noted that the operating principle of this formula is: is the relative change in the volume of the left lateral ventricle. is the relative change in the density of cerebrospinal fluid in the left lateral ventricle. It is the product of the first two parts, taking into account the interactive effects of volume change and cerebrospinal fluid density change. This formula takes into account the relative change in volume, the relative change in cerebrospinal fluid density and their interactive effects, and divides it by 3 for normalization to obtain the left lateral ventricle change rate XA. The higher the left lateral ventricle change rate XA, the higher the risk of left lateral ventricle hemorrhage.

[0028] The right lateral ventricle change rate XB was calculated based on the right lateral ventricle volume AB, the normal lateral ventricle volume AC, the right lateral ventricle average CT value TB and the normal cerebrospinal fluid CT value TC. The formula is: ; It should be noted that the operating principle of this formula is: is the relative change in the volume of the right lateral ventricle. is the relative change in the density of cerebrospinal fluid in the right lateral ventricle, It is the product of the first two parts, taking into account the interactive effects of volume changes and cerebrospinal fluid density changes. This formula takes into account the relative change in volume, the relative change in cerebrospinal fluid density and their interactive effects, and divides it by 3 for normalization to obtain the right lateral ventricle change rate XB. The higher the right lateral ventricle change rate XB, the higher the risk of right lateral ventricle hemorrhage.

[0029] Specifically, the method for calculating the anterior horn hemorrhage index WA, body hemorrhage index WB, posterior horn hemorrhage index WC and inferior horn hemorrhage index WD is as follows: The lateral ventricle data also include the anterior horn angle value DA, anterior horn cross-sectional area DB, normal anterior horn angle value DC, normal anterior horn cross-sectional area DD, corpus callosum thickness DE, thalamus coordinate value (DF, DG), normal corpus callosum thickness DH, normal thalamus coordinate value (DI, DJ), posterior horn cross-sectional area DK, posterior horn length DL, normal posterior horn cross-sectional area DM, normal posterior horn length DN, inferior horn volume DQ and normal inferior horn volume DP of the hemorrhage side lateral ventricle.

[0030] It should be noted that the anterior angle value DA refers to the angle formed by the anterior angle of the lateral ventricle in the coronal plane or axial plane, which can reflect the morphological characteristics of the anterior angle of the lateral ventricle, and the unit is degree. The cross-sectional area of ​​the anterior angle DB refers to the area occupied by the anterior angle of the lateral ventricle on a certain cross-sectional image, which is used to evaluate the size change of the anterior angle of the lateral ventricle, and the unit is square millimeter or square centimeter. The thickness of the corpus callosum DE refers to the thickness measurement of the corpus callosum in a specific part, such as the knee, body or splenium of the corpus callosum, which can reflect the development or pathological condition of the corpus callosum, and the unit is millimeter. The thalamus coordinate value (D F, DG) is the coordinate position of the center point of the thalamus in the imaging coordinate system. The cross-sectional area of ​​the posterior horn DK refers to the area of ​​the posterior horn of the lateral ventricle on a specific cross-sectional image. Its size changes may indicate pathological changes in the ventricular system. The unit is square millimeters or square centimeters. The length of the posterior horn DL refers to the distance from the starting point to the end of the posterior horn of the lateral ventricle, and the unit is millimeters. The volume of the inferior horn DQ refers to the volume of the space contained in the inferior horn of the lateral ventricle, and the unit is cubic millimeters or cubic centimeters. The inferior horn is a part of the lateral ventricle, and its volume changes may be related to diseases such as brain atrophy and hydrocephalus.

[0031] The anterior angle value DA, anterior angle cross-sectional area DB, corpus callosum thickness DE, thalamus coordinate values ​​(DF, DG), posterior angle cross-sectional area DK, posterior angle length DL and inferior angle volume DQ are obtained as follows: use professional image processing software such as 3D-Slicer and MIMICS to reconstruct the original scanning data in the CT image, generate images in different planes such as the coronal plane, axial plane, and sagittal plane, select the two boundary lines of the anterior angle of the lateral ventricle on the coronal or axial plane image through the angle measurement tool of the software, and the software automatically calculates the anterior angle value DA, and on the cross-sectional image, use the area measurement tool of the software to outline the contour of the anterior angle of the lateral ventricle, and the software automatically calculates the area of ​​the enclosed area, that is, the anterior angle cross-sectional area. The software automatically calculates the area of ​​the enclosed area, i.e., the cross-sectional area DK of the posterior horn. The software uses the angle measurement tool on the sagittal image to obtain the length DL of the posterior horn. The software uses the 3D reconstruction function to perform 3D modeling of the inferior horn of the lateral ventricle, and then automatically calculates the volume DQ of the inferior horn. The normal anterior angle value DC refers to the angle formed by the anterior angle of the lateral ventricle in the coronal or axial plane under normal physiological conditions, which reflects the morphological characteristics of the anterior angle of the lateral ventricle. The unit is degree. The acquisition method is: collect a large number of CT images of healthy people with no brain diseases and normal nervous system development, use 3D-Slicer, MIMICS and other software to obtain the anterior angle value data of this population, add up these data and divide them by the number of data to obtain the normal anterior angle value DC.

[0032] The normal anterior horn cross-sectional area DD represents the area occupied by the anterior horn of the lateral ventricle on the visible cross-section of the genu of the corpus callosum under normal circumstances. It is used to measure the spatial size of the anterior horn of the lateral ventricle. The unit is square millimeters or square centimeters. The acquisition method is: collect CT images of a large number of healthy people with no brain diseases and normal nervous system development, use 3D-Slicer, MIMICS and other software to obtain the anterior horn cross-sectional area data of this population, add up these data and divide them by the number of data to obtain the normal anterior horn cross-sectional area DD.

[0033] Normal corpus callosum thickness DH refers to the thickness measurement value of the corpus callosum in a specific part of the normal human body, such as the knee, body or pressure part, in millimeters. It is obtained by collecting CT images of a large number of healthy people with no brain diseases and normal nervous system development, and using 3D-Slicer, MIMICS and other software to obtain the corpus callosum thickness data of the population. These data are accumulated and divided by the number of data to obtain the normal corpus callosum thickness DH.

[0034] The normal thalamic coordinate value (DI, DJ) refers to the coordinate position of the center point of the normal thalamus in the imaging coordinate system. The acquisition method is: collect a large number of CT images of healthy people without brain diseases and with normal nervous system development, use 3D-Slicer, MIMICS and other software to obtain the thalamic coordinate data of the population, and calculate the average value of the horizontal coordinate and the average value of the vertical coordinate in the coordinate data as the normal thalamic coordinate value (DI, DJ).

[0035] The normal posterior horn cross-sectional area DM refers to the area of ​​the normal posterior horn of the lateral ventricle on the cross-sectional image showing the complete state of the posterior horn of the lateral ventricle. It is used to evaluate whether there are abnormal changes in the morphology or size of the posterior horn of the lateral ventricle. The unit is square millimeters or square centimeters. The acquisition method is: collect a large number of CT images of healthy people with no brain diseases and normal nervous system development, use 3D-Slicer, MIMICS and other software to obtain the corpus callosum thickness data of the population, add up these data and divide them by the number of data to obtain the normal posterior horn cross-sectional area DM.

[0036] The normal posterior horn length DN refers to the distance from the starting point to the end of the posterior horn of the lateral ventricle under normal physiological conditions, measured in millimeters. It is obtained by collecting CT images of a large number of healthy people with no brain diseases and normal nervous system development, and using 3D-Slicer, MIMICS and other software to obtain the posterior horn length data of the population. These data are accumulated and divided by the number of data to obtain the normal posterior horn length DN.

[0037] The normal inferior horn volume DP refers to the volume of the space contained in the inferior horn of the normal lateral ventricle, and the unit is cubic millimeter or cubic centimeter; the acquisition method is: collect a large number of CT images of healthy people with no brain diseases and normal nervous system development, use 3D-Slicer, MIMICS and other software to obtain the inferior horn volume data of this population, add up these data and divide them by the number of data to obtain the normal inferior horn volume DP.

[0038] The anterior horn hemorrhage index WA is calculated based on the anterior horn angle value DA, the anterior horn cross-sectional area DB, the normal anterior horn angle value DC and the normal anterior horn cross-sectional area DD. The formula is: ; It should be noted that the operating principle of this formula is: It is the relative change of the anterior angle of the lateral ventricle, reflecting the degree of deviation of the anterior angle of the lateral ventricle from the normal state. When anterior angle hemorrhage occurs, the anterior angle value will change. The higher the deviation of the anterior angle of the lateral ventricle from the normal state, the higher the probability of anterior angle hemorrhage. It is the relative change of the cross-sectional area of ​​the anterior horn of the lateral ventricle. The change of the cross-sectional area of ​​the anterior horn of the lateral ventricle is an important basis for judging whether there is bleeding in the anterior horn. The larger the value is, the higher the probability of anterior horn bleeding occurs. This formula comprehensively considers the effects of the two important factors of anterior horn angle and cross-sectional area on anterior horn bleeding. The square of the relative change of the anterior horn angle value of the lateral ventricle is added to the square of the relative change of the cross-sectional area of ​​the anterior horn and the square root is taken to obtain the anterior horn bleeding index WA of the lateral ventricle.

[0039] The body hemorrhage index WB is calculated based on the corpus callosum thickness DE, thalamus coordinate values ​​(DF, DG), normal corpus callosum thickness DH, and normal thalamus coordinate values ​​(DI, DJ). The formula is: ; It should be noted that the operating principle of this formula is: Indicates the relative change in the thickness of the corpus callosum. Changes in the thickness of the corpus callosum are related to brain lesions such as body hemorrhage. It is the spatial distance deviation of the actual position of the thalamus relative to the normal position. The change of the thalamic position may be caused by brain tissue displacement caused by lesions such as brain hemorrhage. This formula quantifies the abnormality of the thalamus by comprehensively considering the degree of deviation of the corpus callosum thickness from the normal and the spatial distance deviation, and obtains the body hemorrhage index WB.

[0040] The posterior angle bleeding index WC is calculated based on the posterior angle cross-sectional area DK, the posterior angle length DL, the normal posterior angle cross-sectional area DM and the normal posterior angle length DN. The formula is: ; It should be noted that the operating principle of this formula is: Represents the relative change in the cross-sectional area of ​​the back angle, The actual change in the length of the back angle is expressed by the natural logarithm operation, which comprehensively considers the change in the cross-sectional area and length of the back angle to obtain the back angle bleeding index WC, where: is the natural logarithm.

[0041] The inferior angle bleeding index WD is calculated based on the inferior angle volume DQ and the normal inferior angle volume DP, and the formula is: ; It should be noted that this formula obtains the lower angle bleeding index WD by calculating the relative change ratio of the lower angle volume.

[0042] Specifically, the method for calculating the third ventricle change rate RB is: The third ventricle data include the third ventricle blood volume BA, the third ventricle area BB, the normal third ventricle volume BC and the normal third ventricle area average BE.

[0043] It should be noted that the third ventricle blood volume BA refers to the volume of blood contained in the third ventricle, which is of great significance for evaluating the severity of ventricular hemorrhage and the changes in the condition. The unit is cubic millimeter or cubic centimeter, and it is obtained through medical imaging processing software such as 3DSlicer and MIMICS. The third ventricle area BB refers to the area of ​​the third ventricle on a specific imaging section, such as the coronal or axial plane, which reflects the shape and size of the third ventricle. The unit is square millimeter or square centimeter, and the method of obtaining is: through 3DSlicer, MIMICS and other medical imaging software The area of ​​the third ventricle on the coronal plane in the CT image is obtained by medical image processing software, which is used as the third ventricle area BB. The normal third ventricle volume BC refers to the volume of cerebrospinal fluid and other contents contained in the third ventricle under normal physiological conditions, and the unit is cubic millimeter or cubic centimeter. The acquisition method is: collect a large number of CT images of healthy people without brain diseases and normal nervous system development, obtain the third ventricle volume data of this population through medical image processing software such as 3DSlicer and MIMICS, and calculate the average value of the data as the normal third ventricle volume BC. The average BE of the normal third ventricle area is the arithmetic mean of the third ventricle area measurements in the normal population. It is an important statistic of the normal third ventricle area range, which can more intuitively reflect the concentration trend of the third ventricle area in the normal population. The unit is square millimeters or square centimeters. The acquisition method is: select a large number of CT images of healthy people without brain diseases and normal nervous system development as samples, and use medical image processing software such as 3DSlicer and MIMICS to obtain the area of ​​the third ventricle on the coronal plane in different medical record samples. These area data are added up and divided by the number of data to obtain the average BE of the normal third ventricle area.

[0044] The third ventricle change rate RB is calculated based on the third ventricle data according to the following formula: ; Among them, 1 for The weight coefficient is 0.3 to 0.7. Determine the degree of influence on the change rate of the third ventricle RB; 2 for The weight coefficient is 0.3 to 0.7. The degree of influence on the change rate of the third ventricle RB is determined; and 1 +ɑ 2 =1.

[0045] It should be noted that the working principle of this formula is: the third ventricular hemorrhage often appears as a cast, that is, the blood fills the entire third ventricle. By calculating the ratio of the volume of blood in the third ventricle to the normal volume of the third ventricle , can quantify the degree of casting and judge the third ventricle hemorrhage. The normal third ventricle has a specific shape. When bleeding causes casting or partial blood accumulation, its shape will change. By calculating the relative change in the area of ​​the patient's third ventricle relative to the normal third ventricle area, the degree of morphological deformation can be quantified to assist in judging the third ventricle hemorrhage. The formula uses a weighted method to comprehensively consider the influence of both aspects and obtain the third ventricle change rate RB.

[0046] Specifically, the method for calculating the change rate of the fourth ventricle RC is: The fourth ventricle data include the fourth ventricle long diameter CA, the fourth ventricle short diameter CB, the normal fourth ventricle long diameter CC and the normal fourth ventricle short diameter CD.

[0047] It should be noted that the long axis CA of the fourth ventricle refers to the diameter length of the fourth ventricle in the direction of its longest axis, reflecting the size and shape of the fourth ventricle in this direction. The short axis CB of the fourth ventricle refers to the shortest diameter length of the fourth ventricle measured in the direction perpendicular to the long axis. The long axis CA and the short axis CB of the fourth ventricle are obtained by analyzing CT images using medical image processing software such as 3DSlicer and MIMICS. The normal fourth ventricle long diameter CC refers to the reference range value of the long diameter of the fourth ventricle under normal physiological conditions, which is used to assist in judging whether the long diameter of the fourth ventricle is normal, whether there are abnormal conditions such as enlargement or reduction, and the normal fourth ventricle short diameter CD refers to the reference range value of the short diameter of the fourth ventricle under normal physiological conditions, which is used to assist in judging whether the shape and size of the fourth ventricle are normal. The method for obtaining the normal fourth ventricle long diameter CC and the normal fourth ventricle short diameter CD is as follows: a large number of CT images of healthy people without brain diseases and normal nervous system development are selected as samples, and the fourth ventricle long diameter data and the fourth ventricle short diameter data of different samples are obtained through medical image processing software such as 3DSlicer and MIMICS, and the average values ​​of the fourth ventricle long diameter data and the fourth ventricle short diameter data are calculated respectively as the normal fourth ventricle long diameter CC and the normal fourth ventricle short diameter CD.

[0048] The fourth ventricle change rate RC is calculated based on the fourth ventricle data, and the formula is: ; It should be noted that the operating principle of this formula is: It indicates the difference between the actual value and the normal value of the long diameter of the fourth ventricle. It represents the difference between the actual value and the normal value of the short axis of the fourth ventricle. The numerator adds the changes in the long axis and the short axis, comprehensively considering the morphological changes of the fourth ventricle in both dimensions, and more comprehensively reflects the overall changes of the fourth ventricle. The denominator plays a normalization role, dividing the comprehensive changes in the numerator by a quantity related to the size of the normal ventricle, so that the calculated change rate of the fourth ventricle is within a relatively reasonable range. This formula calculates the sum of the differences between the actual values ​​of the long axis and the short axis of the fourth ventricle and the normal values, and divides it by the sum of the normal long axis and short axis to obtain the change rate of the fourth ventricle RC, which reflects the degree and trend of change of the fourth ventricle relative to the normal state. Specifically, when bleeding occurs in the fourth ventricle, blood accumulates in the fourth ventricle, which will increase the contents in the ventricle, thereby increasing the pressure in the fourth ventricle. Under the action of pressure, the fourth ventricle may expand, and the long axis and short axis will increase accordingly. The higher the change rate RC of the fourth ventricle, the higher the probability of bleeding in the fourth ventricle.

[0049] Specifically, the method for calculating the hematoma location index WF is: The fourth ventricle data also include the maximum projection area FA of the hematoma, the minimum projection area FB of the hematoma, the volume FC of the hematoma located in the top area of ​​the fourth ventricle, and the total volume FD of the hematoma.

[0050] It should be noted that the maximum projection area FA of the hematoma refers to the maximum value of the two-dimensional spatial range occupied by the hematoma on different planes. This area can reflect the maximum coverage of the hematoma and is usually used to preliminarily evaluate the size of the hematoma and the degree of impact on the ventricle. The minimum projection area FB of the hematoma refers to the minimum value of the two-dimensional spatial range occupied by the hematoma on different planes, which is of certain significance for evaluating the morphology of the hematoma and the local impact on the ventricle. The method for obtaining the maximum projection area FA and the minimum projection area FB of the hematoma is: use 3DSlicer, MIMICS and other medical image processing software to analyze the CT images, obtain the projection area data of the hematoma in different planes, traverse these data, and take the maximum value as the maximum projection area FA of the hematoma, and the minimum value as the minimum projection area FB of the hematoma. The volume FC of the hematoma located in the top area of ​​the fourth ventricle refers to the volume of the hematoma in the top area of ​​the fourth ventricle. The acquisition method is: use the threshold setting function in medical image processing software such as 3DSlicer and MIMICS to set the threshold according to the gray value range of the hematoma on the CT image, so that the software can automatically identify and segment the hematoma in the top area of ​​the fourth ventricle. At the same time, the software will calculate the volume of this part of the hematoma area based on the segmented hematoma area using its built-in algorithm. This volume is the volume FC of the hematoma located in the top area of ​​the fourth ventricle. The total hematoma volume FD refers to the total volume of the fourth ventricle hematoma. It is an important indicator for evaluating the size of the hematoma and its impact on the ventricular system. The method for obtaining it is as follows: browse each layer of the CT image, obtain the number of layers occupied by the hematoma on the CT image, and then multiply it by the layer thickness of the CT scan to obtain the thickness of the hematoma. Use the measurement tools in medical image processing software such as 3DSlicer and MIMICS to measure the area of ​​the hematoma in each layer, find the layer corresponding to the maximum hematoma area, and use the measurement tools to accurately measure the longest diameter and widest diameter of the hematoma on the largest layer. Among them, the longest diameter refers to the longest distance of the hematoma on this layer, and the widest diameter refers to the widest distance in the direction perpendicular to the longest diameter. Finally, the total hematoma volume FD is calculated by the formula (longest diameter × widest diameter × thickness of hematoma × π) / 6.

[0051] The hematoma location index WF is calculated based on the maximum projection area FA of the hematoma, the minimum projection area FB of the hematoma, the volume FC of the hematoma located at the top of the fourth ventricle, and the total volume FD of the hematoma. The formula is: ; Among them, γ 1 is the weight coefficient of the hematoma morphology index, ranging from 0.4 to 0.6, and is determined according to the influence of the hematoma morphology index on the hematoma location index WF; γ 2 is the weight coefficient of the hematoma distribution position index, which is between 0.4 and 0.6 and is determined according to the influence of the hematoma distribution position index on the hematoma position index WF; and γ 1+γ 2 =1; It should be noted that the operating principle of this formula is: It is a hematoma morphology index used to measure the irregularity of the hematoma shape. Under normal circumstances, the difference in the projection area of ​​a regular-shaped object in different directions is relatively small, and the irregularity of the hematoma shape is related to factors such as the bleeding location and bleeding speed. By calculating the ratio of the maximum and minimum projection areas, the irregularity of the hematoma shape can be quantified, thereby assisting in determining the location of the bleeding point. If the bleeding point is close to the ventricular wall, the hematoma may be restricted in stretching in a certain direction due to the limitation of the ventricular wall, resulting in a large difference in the projection area. It is the hematoma distribution position index, which is used to determine the bleeding point location according to the distribution of hematoma in different areas of the fourth ventricle. If the bleeding point is close to the top of the fourth ventricle, the volume of the hematoma in the top area will account for a relatively large proportion. By calculating the ratio of the hematoma volume in the top area to the total volume, the distribution of the hematoma can be quantified. This formula uses a weighted method to comprehensively consider the influence of hematoma morphology indicators and hematoma distribution position indicators to obtain the hematoma position index WF.

[0052] In the above scheme, the left lateral ventricle change rate XA, the right lateral ventricle change rate XB, the lateral ventricle anterior horn bleeding index WA, the body bleeding index WB, the posterior horn bleeding index WC and the inferior horn bleeding index WD are calculated based on the lateral ventricle data, so that the specific site of lateral ventricle hemorrhage can be accurately located. The third ventricle change rate RB is calculated based on the third ventricle data, so that abnormalities can be discovered in time in the early stage of third ventricle hemorrhage, that is, when the bleeding may only cause slight changes in the ventricular morphology. The fourth ventricle change rate RC and the hematoma position index WF are calculated based on the fourth ventricle data, so that the degree of morphological change of the fourth ventricle can be more objectively and accurately reflected, providing a more accurate basis for disease assessment.

[0053] The model training module builds an initial model based on a convolutional neural network. It uses the CT images of different patients' medical records and the corresponding lateral ventricle data, third ventricle data, fourth ventricle data, and all data values ​​obtained by the indicator calculation module as input data items, and uses the diagnosis results of whether each ventricle in the corresponding patient's medical record is bleeding as output items. The initial model is trained and optimized, and a bleeding site probability model is formed after the training is completed.

[0054] It should be noted that in the stage of building a convolutional neural network, after defining the basic network architecture, such as convolutional layer, pooling layer, fully connected layer, etc., the output layer is set to multiple neurons. On the basis of a single neuron used to output whether bleeding occurs (assuming that the sigmoid activation function is used for binary classification), the number of neurons is increased, which is the same as the number of ventricular locations to be predicted. For example, 3 neurons are added, corresponding to the lateral ventricle, the third ventricle, the fourth ventricle, etc., and the Softmax activation function is used for these 3 neurons. In this way, the model output will be a vector containing 3 probability values, each value representing the probability of bleeding in the corresponding ventricle.

[0055] In the above scheme, by training the probability model of bleeding sites, the brain condition can be reflected from multiple dimensions, which is conducive to improving the accuracy of diagnosis.

[0056] The model judgment module can input the CT image of the patient to be diagnosed, the corresponding lateral ventricle data, the third ventricle data, the fourth ventricle data and the data values ​​calculated by the indicator calculation module into the bleeding site probability model, and the bleeding site probability model outputs the lateral ventricle bleeding probability PA, the third ventricle bleeding probability PB and the fourth ventricle bleeding probability PC; presets a set of bleeding thresholds; and judges the bleeding risk level of each ventricle according to the lateral ventricle bleeding probability PA, the third ventricle bleeding probability PB, the fourth ventricle bleeding probability PC and the bleeding threshold set.

[0057] Specifically, the method for determining the bleeding risk level of each ventricle is: The bleeding threshold set includes a lateral ventricle bleeding threshold SA, a third ventricle bleeding threshold SB, and a fourth ventricle bleeding threshold SC.

[0058] It should be noted that the method for determining the lateral ventricle hemorrhage threshold SA, the third ventricle hemorrhage threshold SB and the fourth ventricle hemorrhage threshold SC is as follows: brain CT image data of a large number of cerebral hemorrhage patients and healthy people without brain diseases are collected as sample data, the lateral ventricle data, the third ventricle data and the fourth ventricle data of different sample data are calculated according to the method in the indicator calculation module and input into the bleeding site probability model, the lateral ventricle hemorrhage probability, the third ventricle hemorrhage probability and the fourth ventricle hemorrhage probability output by the bleeding site probability model are obtained, the lower 50% lateral ventricle hemorrhage probability value is screened out and its average value is calculated as the reference value of the lateral ventricle hemorrhage threshold SA, the lower 50% third ventricle hemorrhage probability value is screened out and its average value is calculated as the reference value of the third ventricle hemorrhage threshold SB, the lower 50% fourth ventricle hemorrhage probability is screened out and its average value is calculated as the reference value of the fourth ventricle hemorrhage threshold SC.

[0059] The formula for judging whether there is bleeding risk in each part is based on the probability of lateral ventricle hemorrhage PA, the probability of third ventricle hemorrhage PB, the probability of fourth ventricle hemorrhage PC and the bleeding threshold set: ; ; .

[0060] The positioning module calculates the hemorrhage area determination index HE in different areas of the ventricle based on the corresponding ventricle data when any ventricle is judged to have a high risk of hemorrhage e ; Preset bleeding area determination threshold SU; According to the bleeding area determination index HE e and the bleeding area determination threshold SU to determine whether each area is a bleeding area.

[0061] Specifically, the bleeding area determination index HE is calculated e The method is: Corresponding ventricle data include actual ventricle volume HC e 、Normal ventricular volume HD e , actual CT value CTA e 、 Normal CT value CTB e , average displacement distance of peripheral blood vessels CTC e and vascular disruption ratio (CTD) e .

[0062] It should be noted that when any ventricle is determined to have a high risk of bleeding, the corresponding ventricular data of this ventricle is obtained. Specifically, the CT images are analyzed using medical image processing software such as 3DSlicer and MIMICS, and a three-dimensional coordinate system for the ventricle with high bleeding risk is established. The ventricle with high bleeding risk is divided into small cubic areas at a fixed interval of 0.5 mm × 0.5 mm × 1 mm, and each small area is numbered from left to right and from front to back.

[0063] Actual ventricular volume HC e It refers to the size of the space occupied by each ventricular area in the actual CT image of the patient, in cubic millimeters. The acquisition method is: use 3DSlicer, MIMICS and other medical image processing software to analyze the CT image data, obtain the number of voxels contained in each small area, and calculate the number of voxels × 0.25 mm according to the formula 3 Calculate the actual ventricular volume HC e , where CT images are composed of many tiny three-dimensional units, these units are called voxels, which are the smallest units representing volume in CT images.

[0064] Normal ventricular volume HD eIt refers to the size of the space occupied by each ventricular area under normal physiological conditions. The acquisition method is: collect a large number of CT images of healthy people with no brain diseases and normal nervous system development as samples, use 3D-Slicer, MIMICS and other medical image processing software to obtain the number of voxels contained in each small area of ​​different samples, for each small area, the number of voxels contained in the small area of ​​different samples is accumulated and divided by the number of samples to obtain the average number of voxels. According to the formula, the average number of voxels × 0.25㎜ 3 Calculate the normal ventricular volume HD e .

[0065] Actual CT value CTA e It refers to the average density of cerebrospinal fluid in each small area on the CT image, measured in Hounsfield units. It is obtained by analyzing CT image data using medical image processing software such as 3DSlicer and MIMICS.

[0066] Normal CT value CTB e It refers to the density value of cerebrospinal fluid in each small area on the CT image under normal physiological conditions. The unit is Hounsfield unit. The method of obtaining it is: collect a large number of CT images of healthy people with no brain diseases and normal nervous system development as samples, use 3D-Slicer, MIMICS and other medical image processing software to obtain the CT value of each small area in different samples, and for each small area, add up the CT values ​​of different samples in the small area and divide it by the number of samples to obtain the normal CT value CTB. e .

[0067] Average displacement distance of peripheral blood vessels CTC e It refers to the average movement distance of blood vessels within a certain range of each small area relative to their normal position. It can reflect the impact of the lesion on the surrounding blood vessels. The unit is millimeter and is obtained using image registration software such as ANTS and Elastix.

[0068] CTD e It refers to the ratio of the number or length of interrupted blood vessels around each small area in the vascular network displayed by CT images to the total number or length of blood vessels. It is obtained using medical image processing software such as OsiriX and 3D-Slicer.

[0069] Calculate the bleeding area determination index HE based on the bleeding data e , the formula based on is: ; Among them, HE e is the curvature change index of the ventricular wall in the e-th region of the brain; e is the serial number corresponding to different regions, and its value is a positive integer; HCe is the actual ventricular volume of the eth region in the brain region; HD e is the normal ventricular volume of the eth region in the brain; CTA e is the actual CT value of the e-th area in the brain; CTB e is the normal CT value of the eth area in the brain; CTC e is the average displacement distance of the peripheral blood vessels in the e-th region of the brain; CTD e is the proportion of vascular interruption in the e-th region of the brain; β 1 for The weight coefficient is 0.2 to 0.5. HE is an indicator for determining the bleeding area. e The degree of influence is determined; 2 for The weight coefficient is 0.3 to 0.5. HE is an indicator for determining the bleeding area. e The degree of influence is determined; 3 for The weight coefficient is 0.2 to 0.5. HE is an indicator for determining the bleeding area. e The degree of influence of β is determined; and 1 +β 2 +β 3 =1.

[0070] It should be noted that the operating principle of this formula is: It reflects the relative difference between the actual ventricular volume of the eth region in the brain and the normal ventricular volume. It reflects the relative difference between the actual CT value of the e-th area in the brain and the CT value in the normal state. When there is bleeding, the hematoma may compress the surrounding blood vessels to cause displacement, or it may damage the blood vessels and cause interruption. Taking into account the changes in the position and structural integrity of the blood vessels, the formula uses a weighted method to comprehensively consider the influence of the above three parts and obtain the bleeding area determination index HE e .

[0071] Specifically, the method for determining whether each area is a bleeding area is as follows: HE index is determined based on the bleeding area e The bleeding area determination threshold SU is used to determine whether bleeding exists in each area, and the formula is as follows: ; It should be noted that the method for determining the bleeding area determination threshold SU is: collect a large number of CT images of healthy people with no brain diseases and normal nervous system development as samples, calculate the bleeding area determination indicators of different samples according to the above method, screen out the higher 50% of the data and calculate its average value as the reference value of the bleeding area determination threshold SU.

[0072] In the above scheme, the hemorrhage area determination index HE of different regions of the ventricle is calculated based on the corresponding ventricle data. e , can more accurately reflect the actual bleeding possibility of each area, and determine the index HE according to the bleeding area e The bleeding area determination threshold SU determines whether each area is a bleeding area, which can timely detect potential bleeding risks and provide a basis for early intervention and treatment of patients, while providing a clear and unified standard for bleeding diagnosis.

[0073] On the other hand, the present invention also discloses a method for anatomically assisting the positioning of intracranial hemorrhage based on CT images, which is used to implement the above-mentioned anatomically assisting the positioning of intracranial hemorrhage based on CT images, and comprises the following steps: Based on the CT images of different patients' medical records, the corresponding lateral ventricle data, third ventricle data, and fourth ventricle data were obtained; Based on the lateral ventricle data, the left lateral ventricle change rate XA, the right lateral ventricle change rate XB, the lateral ventricle anterior horn bleeding index WA, the body bleeding index WB, the posterior horn bleeding index WC and the inferior horn bleeding index WD were calculated; based on the third ventricle data, the third ventricle change rate RB was calculated; based on the fourth ventricle data, the fourth ventricle change rate RC and the hematoma location index WF were calculated; An initial model is constructed based on a convolutional neural network; CT images of different patients' medical records and the corresponding lateral ventricle data, third ventricle data, fourth ventricle data, and all data values ​​obtained by the index calculation module are used as input data items, and the diagnosis results of whether each ventricle in the corresponding patient's medical record is bleeding are used as output items, and the initial model is trained and optimized. After the training is completed, a bleeding site probability model is formed; Input the CT image of the patient to be diagnosed, the corresponding lateral ventricle data, the third ventricle data, the fourth ventricle data and the data values ​​calculated by the index calculation module into the bleeding site probability model, and the bleeding site probability model outputs the lateral ventricle bleeding probability PA, the third ventricle bleeding probability PB and the fourth ventricle bleeding probability PC; preset a bleeding threshold set; determine the bleeding risk level of each ventricle according to the lateral ventricle bleeding probability PA, the third ventricle bleeding probability PB, the fourth ventricle bleeding probability PC and the bleeding threshold set; When any ventricle is judged to have a high risk of bleeding, the hemorrhage area determination index HE of different areas of the ventricle is calculated based on the corresponding ventricle data. e ; Preset bleeding area determination threshold SU; According to the bleeding area determination index HEe and the bleeding area determination threshold SU to determine whether each area is a bleeding area.

[0074] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product. A person of ordinary skill in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented in electronic hardware or in a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution.

[0075] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0076] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. The anatomical auxiliary positioning system for intracranial hemorrhage based on CT images is characterized by: include: The data acquisition module can obtain the corresponding lateral ventricle data, third ventricle data and fourth ventricle data based on the CT images of different patients' medical records; The index calculation module can calculate the left lateral ventricle change rate XA, the right lateral ventricle change rate XB, the lateral ventricle anterior horn bleeding index WA, the body bleeding index WB, the posterior horn bleeding index WC and the lower horn bleeding index WD based on the lateral ventricle data; and calculate the third ventricle change rate RB based on the third ventricle data; The fourth ventricle change rate RC and hematoma location index WF were calculated based on the fourth ventricle data; The model training module builds an initial model based on a convolutional neural network; the CT images of different patients' medical records and the corresponding lateral ventricle data, third ventricle data, fourth ventricle data and all data values ​​obtained by the indicator calculation module are used as input data items, and the diagnosis results of whether each ventricle in the corresponding patient's medical record is bleeding are used as output items to train and optimize the initial model. After the training is completed, a probability model of the bleeding site is formed; The model judgment module can input the CT image of the patient to be diagnosed, the corresponding lateral ventricle data, the third ventricle data, the fourth ventricle data and the data values ​​calculated by the index calculation module into the bleeding site probability model, and the bleeding site probability model outputs the lateral ventricle bleeding probability PA, the third ventricle bleeding probability PB and the fourth ventricle bleeding probability PC; preset a bleeding threshold set; judge the bleeding risk level of each ventricle according to the lateral ventricle bleeding probability PA, the third ventricle bleeding probability PB, the fourth ventricle bleeding probability PC and the bleeding threshold set; The positioning module calculates the hemorrhage area determination index HE in different areas of the ventricle based on the corresponding ventricle data when any ventricle is judged to have a high risk of hemorrhage e ; Preset bleeding area determination threshold SU; according to bleeding area determination index HE e and the bleeding area determination threshold SU to determine whether each area is a bleeding area.

2. The CT image-based intracranial hemorrhage anatomical auxiliary positioning system according to claim 1, characterized in that: The method for calculating the left lateral ventricle change rate XA and the right lateral ventricle change rate XB is: The lateral ventricle data included the left lateral ventricle volume AA, the right lateral ventricle volume AB, the normal lateral ventricle volume AC, the left lateral ventricle average CT value TA, the right lateral ventricle average CT value TB and the normal cerebrospinal fluid CT value TC; The left lateral ventricle change rate XA was calculated based on the left lateral ventricle volume AA, the normal lateral ventricle volume AC, the left lateral ventricle average CT value TA and the normal cerebrospinal fluid CT value TC. The formula is: ； The right lateral ventricle change rate XB was calculated based on the right lateral ventricle volume AB, the normal lateral ventricle volume AC, the right lateral ventricle average CT value TB and the normal cerebrospinal fluid CT value TC. The formula is: 。 3. The CT image-based intracranial hemorrhage anatomical auxiliary positioning system according to claim 2, characterized in that: The method for calculating the lateral ventricle anterior horn hemorrhage index WA, body hemorrhage index WB, posterior horn hemorrhage index WC and inferior horn hemorrhage index WD is as follows: The lateral ventricle data also include the anterior horn angle value DA, anterior horn cross-sectional area DB, normal anterior horn angle value DC, normal anterior horn cross-sectional area DD, corpus callosum thickness DE, thalamus coordinate value (DF, DG), normal corpus callosum thickness DH, normal thalamus coordinate value (DI, DJ), posterior horn cross-sectional area DK, posterior horn length DL, normal posterior horn cross-sectional area DM, normal posterior horn length DN, inferior horn volume DQ and normal inferior horn volume DP of the hemorrhage side lateral ventricle; The anterior horn hemorrhage index WA is calculated based on the anterior horn angle value DA, the anterior horn cross-sectional area DB, the normal anterior horn angle value DC and the normal anterior horn cross-sectional area DD. The formula is: ； The body hemorrhage index WB is calculated based on the corpus callosum thickness DE, thalamus coordinate values ​​(DF, DG), normal corpus callosum thickness DH, and normal thalamus coordinate values ​​(DI, DJ). The formula is: ； The posterior angle bleeding index WC is calculated based on the posterior angle cross-sectional area DK, the posterior angle length DL, the normal posterior angle cross-sectional area DM and the normal posterior angle length DN. The formula is: ； The inferior angle bleeding index WD is calculated based on the inferior angle volume DQ and the normal inferior angle volume DP, and the formula is: 。 4. The CT image-based intracranial hemorrhage anatomical auxiliary positioning system according to claim 3, characterized in that: The method for calculating the third ventricle change rate RB is: The third ventricle data include the third ventricle blood volume BA, the third ventricle area BB, the normal third ventricle volume BC and the average normal third ventricle area BE; The third ventricle change rate RB is calculated based on the third ventricle data according to the following formula: ； Among them, ɑ1 is The weight coefficient is 0.3~0.7; ɑ2 is The weight coefficient is between 0.3 and 0.7, and ɑ1+ɑ2=1.

5. The CT image-based intracranial hemorrhage anatomical auxiliary positioning system according to claim 4, characterized in that: The method for calculating the change rate of the fourth ventricle RC is: The fourth ventricle data include the fourth ventricle long diameter CA, the fourth ventricle short diameter CB, the normal fourth ventricle long diameter CC and the normal fourth ventricle short diameter CD; The fourth ventricle change rate RC is calculated based on the fourth ventricle data, and the formula is: 。 6. The intracranial hemorrhage anatomical auxiliary positioning system based on CT images according to claim 5 is characterized in that: The method for calculating the hematoma location index WF is: The fourth ventricle data also include the maximum projection area FA of the hematoma, the minimum projection area FB of the hematoma, the volume FC of the hematoma located in the top area of ​​the fourth ventricle, and the total volume FD of the hematoma; The hematoma location index WF is calculated based on the maximum projection area FA of the hematoma, the minimum projection area FB of the hematoma, the volume FC of the hematoma located at the top of the fourth ventricle, and the total volume FD of the hematoma. The formula is: ； Among them, γ1 is The weight coefficient is 0.4~0.6; γ2 is The weight coefficient is between 0.4 and 0.6, and γ1+γ2=1.

7. The intracranial hemorrhage anatomical auxiliary positioning system based on CT images according to claim 6 is characterized in that: The method for determining the bleeding risk level of each ventricle is: The bleeding threshold set includes the lateral ventricle bleeding threshold SA, the third ventricle bleeding threshold SB and the fourth ventricle bleeding threshold SC; The formula for judging whether there is bleeding risk in each part is based on the probability of lateral ventricle hemorrhage PA, the probability of third ventricle hemorrhage PB, the probability of fourth ventricle hemorrhage PC and the bleeding threshold set: ； ； 。 8. The CT image-based intracranial hemorrhage anatomical auxiliary positioning system according to claim 7, characterized in that: Calculate the bleeding area determination index HE e The method is: Ventricular data including actual ventricular volume HC e 、Normal ventricular volume HD e , actual CT value CTA e 、Normal CT value CTB e , average displacement distance of peripheral blood vessels CTC e and vascular disruption ratio (CTD) e ; Calculate the bleeding area determination index HE based on the bleeding data e , the formula based on is: ； Among them, HE e is the hemorrhage area determination index of the e-th region in the brain area; e is the serial number corresponding to different regions, and the value is a positive integer; HC e is the actual ventricular volume of the eth region in the brain region; HD e is the normal ventricular volume of the eth region in the brain; CTA e is the actual CT value of the e-th area in the brain; CTB e is the normal CT value of the eth area in the brain; CTC e is the average displacement distance of the peripheral blood vessels in the e-th region of the brain; CTD e is the proportion of vascular interruption in the e-th region of the brain; β1 is The weight coefficient is 0.2~0.5; β2 is The weight coefficient is 0.3~0.5; β3 is The weight coefficient is between 0.2 and 0.5, and β1+β2+β3=1.

9. The intracranial hemorrhage anatomical auxiliary positioning system based on CT images according to claim 8, characterized in that: The method for determining whether each area is a bleeding area is: HE index is determined based on the bleeding area e The bleeding area determination threshold SU is used to determine whether bleeding exists in each area, and the formula is as follows: 。 10. A method for anatomically assisting the localization of intracranial hemorrhage based on CT images, characterized in that: The following steps are involved: Based on the CT images of different patients' medical records, the corresponding lateral ventricle data, third ventricle data, and fourth ventricle data were obtained; Based on the lateral ventricle data, the left lateral ventricle change rate XA, the right lateral ventricle change rate XB, the lateral ventricle anterior horn bleeding index WA, the body bleeding index WB, the posterior horn bleeding index WC and the inferior horn bleeding index WD were calculated; based on the third ventricle data, the third ventricle change rate RB was calculated; The fourth ventricle change rate RC and hematoma location index WF were calculated based on the fourth ventricle data; An initial model is constructed based on a convolutional neural network; CT images of different patients' medical records and the corresponding lateral ventricle data, third ventricle data, fourth ventricle data, and all data values ​​obtained by the index calculation module are used as input data items, and the diagnosis results of whether each ventricle in the corresponding patient's medical record is bleeding are used as output items, and the initial model is trained and optimized. After the training is completed, a bleeding site probability model is formed; Input the CT image of the patient to be diagnosed, the corresponding lateral ventricle data, the third ventricle data, the fourth ventricle data and the data values ​​calculated by the index calculation module into the bleeding site probability model, and the bleeding site probability model outputs the lateral ventricle bleeding probability PA, the third ventricle bleeding probability PB and the fourth ventricle bleeding probability PC; preset a bleeding threshold set; determine the bleeding risk level of each ventricle according to the lateral ventricle bleeding probability PA, the third ventricle bleeding probability PB, the fourth ventricle bleeding probability PC and the bleeding threshold set; When any ventricle is judged to have a high risk of bleeding, the hemorrhage area determination index HE of different areas of the ventricle is calculated based on the corresponding ventricle data. e ; Preset bleeding area determination threshold SU; according to bleeding area determination index HE e and the bleeding area determination threshold SU to determine whether each area is a bleeding area.

Images