A multi-planar brain CT reconstruction method for patients with cerebral venous thrombosis

Abstract

The present invention relates to the field of image processing technology, and in particular to a multi-plane cranial CT reconstruction method for patients with cerebral venous thrombosis, comprising: obtaining a closed area of ​​each plane of a CT image of a patient's CT scan; obtaining an area to be analyzed on each plane according to the shape distribution and grayscale distribution of each closed area of ​​each CT image; dividing the area to be analyzed on each plane according to the angle distribution of the area extension lines between the area to be analyzed on each plane and the area to be analyzed on the adjacent plane, and obtaining suspected lesion areas on each plane; performing edge matching on the suspected lesion areas on each plane and the adjacent plane according to the distance distribution and relative position distribution between the edge pixels of the suspected lesion areas on each plane and the edge pixels of the suspected lesion areas on the adjacent plane; and performing three-dimensional reconstruction of the patient's cranial CT according to the matching results. The present invention improves the effect of three-dimensional reconstruction of cranial CT.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a multi-plane brain CT reconstruction method for patients with cerebral venous thrombosis. Background Art

[0002] Currently, computed tomography (CT) has become the preferred imaging modality for brain diseases due to its rapidity, ubiquity, and high resolution. Particularly in emergency settings, plain CT scans can quickly rule out cerebral hemorrhage. Enhanced scanning and post-processing techniques (such as multiplanar reconstruction (MPR) and maximum intensity projection (MIP)) can reveal direct signs such as venous sinus filling defects and the empty triangle sign, as well as indirect signs such as cerebral edema and venous infarction.

[0003] The reconstruction of brain CT images for patients with venous thrombosis relies on the combination of multi-layer CT images. Therefore, during the three-dimensional reconstruction of the CT images, it is necessary to ensure the clarity of the suspected thrombosis area in each layer of the CT image. However, due to the presence of other tissue structures in the brain with a density close to that of the thrombosis lesion area, there is interference with the collected CT images. Therefore, when the collected CT images are segmented based on the existing threshold segmentation technology to distinguish the suspected thrombosis lesion area, misjudgment is prone to occur, which ultimately affects the three-dimensional reconstruction results of the brain CT images and makes the CT reconstruction effect poor. Summary of the Invention

[0004] In order to solve the technical problem that existing image segmentation methods are prone to misjudgment, resulting in poor CT reconstruction results, the present invention aims to provide a multi-plane cranial CT reconstruction method for patients with cerebral venous thrombosis. The technical solution adopted is as follows:

[0005] Obtaining CT images of each plane of the patient's CT scan and extracting closed areas of each CT image;

[0006] According to the shape distribution and grayscale distribution of each closed area in each CT image, the closed area is screened to obtain the area to be analyzed in each plane layer;

[0007] According to the angle distribution of the extension lines between the area to be analyzed on each plane and the area to be analyzed on the adjacent planes, as well as the area distance distribution, the area to be analyzed on each plane is divided to obtain the suspected lesion area on each plane;

[0008] Based on the distance distribution and relative position distribution between the edge pixel points of the suspected lesion area on each plane and the edge pixel points of the suspected lesion area on the adjacent plane, edge matching is performed on the suspected lesion area on each plane and the adjacent plane; based on the matching results, the patient's brain CT is three-dimensionally reconstructed.

[0009] Preferably, the method of screening the closed areas according to the shape distribution and grayscale distribution of each closed area in each CT image to obtain the area to be analyzed in each plane layer specifically includes:

[0010] The characteristic expression degree of each closed area in each CT image is obtained according to the shape distribution of the minimum circumscribed rectangle of each closed area in each CT image;

[0011] Determining a lesion possibility index for each closed area in each CT image based on the mean grayscale value of all pixels in each closed area in each CT image and the degree of feature expression;

[0012] All closed areas in each CT image are clustered, and the evaluation index of each clustering result is obtained according to the fluctuation of the lesion possibility index of the closed areas in the same cluster in each clustering result;

[0013] In the clustering results corresponding to the maximum value of the evaluation index, all closed areas corresponding to the maximum value of the balance of the lesion possibility index of all closed areas in the cluster are determined as the areas to be analyzed on the corresponding layer plane of the CT image.

[0014] Preferably, obtaining the characteristic expression degree of each closed area in each CT image according to the shape distribution of the minimum circumscribed rectangle of each closed area in each CT image specifically includes:

[0015] Based on the area ratio of each closed area in each CT image to the corresponding minimum circumscribed rectangle, a first ratio coefficient is determined. Based on the aspect ratio of the minimum circumscribed rectangle of each closed area in each CT image, a second ratio coefficient is determined. The first ratio coefficient and the second ratio coefficient are combined to obtain the characteristic expression degree of each closed area in each CT image.

[0016] Preferably, the evaluation index of each clustering result is obtained according to the fluctuation of the lesion possibility index of the closed area in the same cluster in each clustering result, specifically including:

[0017] For any clustering result, the characteristic dispersion coefficient of each cluster is determined based on the dispersion degree of the lesion possibility index of all closed areas in the same cluster;

[0018] The global dispersion coefficient of the clustering result is determined based on the dispersion degree of the mean value of the lesion possibility index of all closed areas in each cluster;

[0019] The evaluation index of the clustering result is determined based on the ratio of the cumulative sum of the characteristic dispersion coefficients of all clusters to the global dispersion coefficient, combined with the negative correlation coefficient of the proportion of clusters contained in the clustering result.

[0020] Preferably, the method of dividing the area to be analyzed on each plane layer according to the angle distribution of the area extension line between the area to be analyzed on each plane layer and the area to be analyzed on the adjacent plane layer and the area distance distribution to obtain the suspected lesion area on each plane layer specifically includes:

[0021] According to the distance distribution between the centroid of each area to be analyzed in each plane and the centroid of each area to be analyzed in the adjacent plane, combined with the distribution of regional edge extension, the extension degree of the lesion between the area to be analyzed in each plane and the adjacent plane is obtained;

[0022] The area to be analyzed on each plane is divided to obtain the initial lesion area on each plane. The degree of lesion extension between the initial lesion area on each plane and the initial lesion area on the adjacent plane is calculated based on the lesion possibility index to determine the optimality of each division result.

[0023] The initial lesion area in the segmentation result corresponding to the maximum value of the segmentation preference degree is used as the suspected lesion area in each plane layer.

[0024] Preferably, the distance distribution between the centroid of each area to be analyzed in each plane and the centroid of each area to be analyzed in the adjacent plane is combined with the distribution of the extension of the regional edges to obtain the extension degree of the lesion between the area to be analyzed in each plane and the adjacent plane, specifically including:

[0025] Any area to be analyzed in any plane layer is used as the first analysis area, and any area to be analyzed in the plane layer adjacent to the first analysis area is used as the second analysis area;

[0026] Determine the first parameter based on the spatial distance between the centroid of the first analysis area and the centroid of the second analysis area; determine the second parameter based on the angle between the straight line containing the long side of the minimum circumscribed rectangle of the first analysis area and the straight line containing the long side of the minimum circumscribed rectangle of the second analysis area;

[0027] The negative correlation coefficient of the Euclidean distance between the first parameter and the second parameter is determined as the extent of lesion extension between the first analysis region and the second analysis region.

[0028] Preferably, the division preference degree of each division result is obtained based on the lesion extension degree between the initial lesion area of ​​each layer plane and the initial lesion area of ​​the adjacent layer plane in each division result, combined with the lesion possibility index, specifically including:

[0029] For any division result, the product of the lesion possibility index between each initial lesion region in each layer plane and each initial lesion region in the adjacent layer plane is obtained as the first characteristic factor, and the second characteristic factor is determined based on the difference in the lesion possibility index between each initial lesion region in each layer plane and each initial lesion region in the adjacent layer plane;

[0030] The ratio of the first characteristic factor to the second characteristic factor and the product of the lesion extension degree are used as the probability index of the initial lesion area corresponding to each two adjacent planes; the probability indexes of all initial lesion areas corresponding to all adjacent planes are combined to obtain the division preference degree of the division result.

[0031] Preferably, performing edge matching on the suspected lesion area of ​​each plane layer and the adjacent plane layer according to the distance distribution and relative position distribution between the edge pixel points of the suspected lesion area of ​​each plane layer and the edge pixel points of the suspected lesion area of ​​the adjacent plane layer specifically includes:

[0032] According to the distance distribution and position distribution between the edge pixels of the suspected lesion area of ​​each plane and the edge pixels of the suspected lesion area of ​​the adjacent plane, the edge correlation degree between the edge pixels of the suspected lesion area of ​​each plane and the adjacent plane is obtained;

[0033] The edge pixels between the suspected lesion areas of each two adjacent planes are matched one by one to obtain different matching methods;

[0034] According to the lesion extension degree between each two adjacent layers of suspected lesion regions and the edge correlation degree between edge pixels in the suspected lesion regions under each matching method, the matching effect evaluation of each matching method for each suspected lesion region is obtained;

[0035] Based on the matching method corresponding to the maximum value of the matching effect evaluation, edge matching is performed on the suspected lesion area of ​​each layer plane and the adjacent layer plane.

[0036] Preferably, obtaining the edge correlation degree between the edge pixels in the suspected lesion area of ​​each plane and the suspected lesion area of ​​the adjacent plane based on the distance distribution and position distribution between the edge pixels in the suspected lesion area of ​​each plane and the edge pixels in the suspected lesion area of ​​the adjacent plane specifically includes:

[0037] Any edge pixel point in any suspected lesion area on any plane is used as the first pixel point of the first lesion area, the suspected lesion area on the plane adjacent to the first lesion area is used as the reference lesion area, and any edge pixel point in any reference lesion area is used as the second pixel point of the second lesion area;

[0038] Obtain a line segment connecting a first pixel point and a second pixel point; use the length of the line segment as a first coefficient; use the angle between the projection line of the line segment on the plane where the first lesion area is located and the straight line containing the long side of the minimum circumscribed rectangle of the first lesion area as a first angle; use the angle between the projection line of the line segment on the plane where the second lesion area is located and the straight line containing the long side of the minimum circumscribed rectangle of the second lesion area as a second angle; and calculate the normalized mean of the first angle and the second angle to obtain a second coefficient;

[0039] Negative correlation processing is performed on the L2 norms of the first coefficient and the second coefficient to obtain the edge correlation degree between the first pixel point of the first lesion area and the second pixel point of the second lesion area.

[0040] Preferably, obtaining the matching effect evaluation of each matching method according to the lesion extension degree between each two adjacent layers of suspected lesion areas and the edge correlation degree between edge pixels in the suspected lesion areas under each matching method specifically includes:

[0041] For any matching method between the first lesion area and the second lesion area, the negative correlation coefficient between the edge association degree corresponding to each matching pixel group and the lesion extension degree between the first lesion area and the second lesion area is calculated to obtain the matching coefficient of each matching pixel group, and the average value of the matching coefficients of all matching pixel groups in the matching method is calculated to obtain the matching effect evaluation between the first lesion area and the second lesion area in the matching method.

[0042] The embodiments of the present invention have at least the following beneficial effects:

[0043] The present invention first collects CT images of different planes to provide a data basis for the subsequent combination of multiple planes for three-dimensional reconstruction. Then, considering that there are other tissue structures in the brain that are close to the density of thrombus, which have certain interference in the division of thrombus lesion areas, the present invention mainly progressively screens the areas where thrombus lesions may exist in each CT image from three levels, and performs corresponding matching on the possible areas with matching correspondences in the images of adjacent planes. On the one hand, with respect to the shape and grayscale distribution of the closed area in each CT image, the degree of expression of the cord sign in the closed area in the CT image is analyzed, and the closed area is preliminarily screened to obtain the area to be analyzed. On the other hand, the venous thrombus part preliminarily screened out is distributed along the actual veins in the multi-plane CT images, and the single-layer possibility and the cross-plane extension probability are integrated. The accuracy of the thrombus area division is improved by the preliminary feature association of the cross-plane CT images, and the suspected lesion area of ​​each plane is determined. Thirdly, in order to solve the fitting distortion problem caused by the misalignment of anatomical structures between layers, a matching relationship between adjacent layer CT images is established for the divided suspected lesion areas, combined with the venous extension and distribution characteristics, ultimately achieving better results in the three-dimensional reconstruction of cranial CT. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0045] Figure 1 This is a flowchart of the steps of a multi-planar brain CT reconstruction method for patients with cerebral venous thrombosis provided by the present invention;

[0046] Figure 2 This is a flow chart of a method for obtaining the area to be analyzed on each plane provided by the present invention;

[0047] Figure 3 This is a flow chart of the method for obtaining the suspected lesion area on each plane provided by the present invention;

[0048] Figure 4 is a flowchart of the steps of the method for obtaining the extent of lesion extension provided by the present invention;

[0049] Figure 5 This is a flow chart of the method provided by the present invention for edge matching of suspected lesion areas on each plane layer and adjacent plane layers. DETAILED DESCRIPTION

[0050] To further illustrate the technical means and effectiveness of the present invention in achieving its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effectiveness of a multi-planar cranial CT reconstruction method for patients with cerebral venous thrombosis proposed in accordance with the present invention. In the following description, references to different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0051] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0052] The following describes in detail a specific scheme of a multi-plane brain CT reconstruction method for patients with cerebral venous thrombosis provided by the present invention with reference to the accompanying drawings.

[0053] See also Figure 1 , which shows a flowchart of a multi-planar cranial CT reconstruction method for patients with cerebral venous thrombosis provided by one embodiment of the present invention, the method comprising the following steps:

[0054] Step S100 : obtaining a CT image of each plane of a patient's CT scan and extracting a closed area of ​​each CT image.

[0055] First, a computed tomography (CT) scan is performed to acquire a CT image of the patient's brain. The grayscale of a CT image is based on Hounsfield units (HUs). Different tissues have different HU ranges, such as air, fat, water, soft tissue, blood, and bone. This CT range (e.g., -1000 HU to +3000 HU) is compressed to the grayscale range supported by the display, typically 256 levels, with specific grayscale intervals corresponding to different tissues.

[0056] CT images corresponding to several planes in the scanning direction from the top of the skull to the base of the skull are collected. It can be understood that each layer of image can be a cross-section corresponding to a certain thickness of the brain, and there is a certain distance between adjacent layer planes. The images are transmitted to the computer for subsequent operation and processing.

[0057] Next, each slice of the CT image is preprocessed using the Canny edge detection algorithm. Considering that some CT images may contain subtle noise, resulting in small gaps in some closed edges, a morphological algorithm is applied to the edge-detected CT images to close these small gaps. Finally, the closed edge contours in each CT image are extracted to form the closed region. It is understood that image preprocessing methods, such as edge detection, morphological processing, and contour extraction, are well-known techniques and will not be further elaborated upon here.

[0058] Taking into account the existence of other tissue structures in the brain with a density close to that of thrombus, which interferes with the division of thrombus lesion areas, this embodiment mainly progressively screens the areas in each CT image that may contain thrombus lesions from three levels, and matches the possible areas with matching correspondences in the images of adjacent planes. On the one hand, referring to step S200, the shape and grayscale distribution of the closed area in each CT image are analyzed, and the degree of the cord sign in the closed area of ​​the CT image is analyzed, and the closed area is preliminarily screened. On the other hand, referring to step S300, the venous thrombus part preliminarily screened out is distributed along the actual veins in the multi-plane CT image, and the single-layer possibility weight and the cross-plane extension probability are integrated to improve the accuracy of the thrombus area division through the preliminary feature association of the cross-plane CT image. On the third hand, in order to solve the fitting distortion problem caused by the misalignment of the anatomical structure between layers, the matching relationship of the adjacent layer CT images is established for the divided thrombus area and non-thrombus area, combined with the venous direction extension distribution characteristics.

[0059] Step S200 , screening the closed regions according to the shape distribution and grayscale distribution of each closed region in each CT image to obtain the region to be analyzed in each plane layer.

[0060] Acute cerebral venous thrombosis is primarily composed of red blood cells and fibrin. Its characteristic appearance on plain CT scans is a high-density, cord-like shadow in cross-section, with CT values ​​ranging from +50HU to +70HU. Its cause is closely related to the direction of the vein's course and the angle between the scan planes. Because the cerebral veins are three-dimensional and tortuous, when the angle between the thrombosed venous segment and the CT scan cross-section is less than 90 degrees, the vessel projection on the scan plane appears as a linear or ribbon-like high-density shadow, a sign known as the "cord sign." Compared to other physiological or pathological components of the brain parenchyma, the cord sign of venous thrombosis presents a distinct, strip-like appearance.

[0061] Based on this feature, the expression level of the thin strip features extending along the vein in each closed area in the single-layer CT image is analyzed, and then the preliminary screening operation is completed by clustering. In some embodiments, Figure 2 As shown, the method for obtaining the area to be analyzed on each plane layer can be implemented by steps S201 to S204.

[0062] Step S201 : obtaining the feature expression level of each closed area in each CT image according to the shape distribution of the minimum circumscribed rectangle of each closed area in each CT image.

[0063] Specifically, a first ratio coefficient is determined based on the area ratio of each closed area in each CT image to the corresponding minimum circumscribed rectangle, and a second ratio coefficient is determined based on the aspect ratio of the minimum circumscribed rectangle of each closed area in each CT image. The first ratio coefficient and the second ratio coefficient are combined to obtain the characteristic expression degree of each closed area in each CT image.

[0064] As a specific example, the first step is to obtain the minimum bounding rectangle of each closed area in each CT image. The ratio between the total number of pixels contained in each closed area in each CT image and the total number of pixels contained in the minimum bounding rectangle of the closed area is used as the first ratio coefficient for each closed area in each CT image. This ratio reflects the ratio of the area of ​​each closed area to the area of ​​the minimum bounding rectangle. The value of the first ratio coefficient reflects the fit between the current closed area shape and the minimum bounding rectangle. The larger the value of the first ratio coefficient, the greater the degree of feature expression of the corresponding closed area, that is, the greater the possibility of striped appearance.

[0065] In the second step, the ratio of the length to the width of the minimum bounding rectangle of each closed region in each CT image is used as the second ratio coefficient for each closed region in each CT image. Both the length and width of the minimum bounding rectangle can be obtained by counting the number of pixels contained in the side length. The second ratio coefficient reflects the shape distribution of the length-to-width ratio of the closed region. A larger value indicates a greater degree of striped characteristics in the closed region, that is, the longer the long side of the closed region, the narrower the width.

[0066] In the third step, the product of the first ratio coefficient and the second ratio coefficient is used as the feature expression degree of each closed area in each CT image, that is, the feature expression degree represents the degree to which the corresponding closed area exhibits strip-like features.

[0067] Step S202 : determining a lesion possibility index for each closed area in each CT image based on the grayscale value mean of all pixels in each closed area in each CT image and the degree of feature expression.

[0068] Specifically, the mean grayscale value of all pixels in each closed area in each CT image is calculated, and then the product of the mean grayscale value and the characteristic expression degree of the corresponding closed area is used as the lesion possibility index of each closed area in each CT image, reflecting the possibility that the closed area is preliminarily suspected to be a venous thrombosis part, and fully considering the color characteristics and morphological characteristics of venous thrombosis compared with other normal tissues.

[0069] Thus, the strip-like features corresponding to each closed area of ​​the single-layer CT image are quantified, and based on the quantification results, a preliminary screening operation is performed on all closed areas in the single-layer CT image.

[0070] In some embodiments, for any CT image, a clustering algorithm can be used to screen all closed areas based on the lesion likelihood index of the closed areas to obtain the area to be analyzed on the slice plane where the CT image is located. Specifically, a K-means clustering algorithm is used, and the cluster distance is measured by the difference distance of the lesion likelihood index between different closed areas. The elbow method is used to determine the number of clusters, that is, the value of K. Furthermore, for each cluster in the clustering result, all closed areas corresponding to the maximum mean value of the lesion likelihood index of all closed areas in the same cluster are the area to be analyzed on the slice plane where the CT image is located.

[0071] In some embodiments, to ensure more accurate preliminary screening results, different clustering results are evaluated to identify the best clustering method. This allows for the initial screening of closed areas with characteristic venous thrombosis features as areas to be analyzed, providing a data foundation for subsequent feature analysis. This process can be implemented in steps S203 and S204.

[0072] Step S203 , clustering all closed regions in each CT image, and obtaining an evaluation index for each clustering result based on fluctuations in the lesion possibility index of closed regions within the same cluster in each clustering result.

[0073] In this embodiment, the screening process of closed regions in each CT image is exactly the same, so the CT image corresponding to any plane is used as an example for description. For any CT image, a clustering algorithm is used to obtain clustering results corresponding to different numbers of clusters.

[0074] The clustering algorithm can adopt the K-means algorithm, using the difference distance of the lesion possibility index between different closed areas as the metric distance of the clustering process, and obtaining different clustering results by respectively determining the value of different cluster numbers K. As a specific example, the number of clusters can be determined by the elbow method to obtain a reference number, and the number interval of the preset length is obtained with the reference number value as the center. The number interval of the cluster number K can be expressed as , and the number of clusters is an integer, and the preset length is 2N+1, For reference, N may be 5 by way of example. In other embodiments, the value may be determined by the implementer according to the specific implementation scenario.

[0075] Furthermore, for any clustering result, the characteristic dispersion coefficient of each cluster is determined based on the dispersion degree of the lesion possibility index of all closed areas in the same cluster; the global dispersion coefficient of the clustering result is determined based on the dispersion degree of the mean value of the lesion possibility index of all closed areas in each cluster; the evaluation index of the clustering result is determined based on the ratio of the cumulative sum of the characteristic dispersion coefficients of all clusters to the global dispersion coefficient, combined with the negative correlation coefficient of the proportion of the number of clusters contained in the clustering result.

[0076] As a specific example, the number of clusters Taking the corresponding clustering results as an example, the evaluation index of the current clustering results can be expressed as: ,in Indicates the number of clusters The corresponding evaluation index of the clustering results is Indicates the clustering result The variance of the lesion possibility index of all closed areas in the cluster, that is, The characteristic dispersion coefficient of the clusters, Indicates the number of clusters in the current clustering result. Indicates the maximum number of clusters in all clustering results. Indicates the global dispersion coefficient of the current clustering result.

[0077] The specific method for obtaining the global dispersion coefficient of the current clustering result is to first calculate the mean of the lesion probability index of the closed area in each cluster, and use the variance of the mean of the lesion probability index of all clusters as the global dispersion coefficient. The optimal clustering result should make the difference in the possibility of the closed areas in each cluster as small as possible, and the difference in the possibility means between different clusters as large as possible. Based on this feature, The larger the value of is, and the smaller the number of clusters is, the larger the corresponding evaluation index value is, the better the clustering effect of the corresponding clustering result is, and the better the effect of distinguishing different types of closed areas is.

[0078] Step S204 , obtaining the clustering result corresponding to the maximum value of the evaluation index, and determining all closed areas corresponding to the maximum value of the balance of the lesion possibility index of all closed areas in the cluster as the area to be analyzed on the corresponding layer plane of the CT image.

[0079] The clustering result corresponding to the maximum value of the evaluation index represents the clustering result with the best clustering effect. The mean value of the lesion possibility index of all closed areas in the same cluster is used as the measurement indicator of the balance situation. The cluster corresponding to the maximum value of the mean value of the lesion possibility index of all clusters is obtained, and all closed areas in the cluster are used as the areas to be analyzed on the plane of the CT image.

[0080] At this point, the region to be analyzed represents the result of a preliminary screening of all closed regions based on their cord-like features, providing a preliminary representation of the area suspected of venous thrombosis. This preliminary demarcation of closed regions within single-plane CT images avoids misidentification that can occur when distinguishing and identifying regions directly based on grayscale or HU values. It also avoids the difficulty of screening and redundant identification processes caused by the large number of closed regions across different planes, thereby improving the accuracy and efficiency of venous thrombosis identification.

[0081] Step S300 , dividing the area to be analyzed on each plane layer according to the angle distribution of the area extension lines between the area to be analyzed on each plane layer and the area to be analyzed on adjacent plane layers, and obtaining the suspected lesion area on each plane layer.

[0082] Acute venous thrombosis is influenced by both vascular trajectory and pathological progression, and its imaging characteristics exhibit a continuous distribution pattern in three-dimensional space. Based on preliminary screening results, considering that when the angle between the thrombosed venous segment and the CT scan cross-section is less than 90 degrees, the blood vessels projected onto the tomographic plane appear as linear or ribbon-like high-density shadows, and that venous thrombosis has a continuous distribution effect within space, the characteristic manifestations of venous thrombosis in each plane can be measured by integrating cross-plane geometric correlation information with morphological consistency assessment. Based on these characteristic manifestations, areas suspected of venous thrombosis can be further screened.

[0083] In some embodiments, as Figure 3 As shown, the method for obtaining the suspected lesion area of ​​each plane layer can be implemented by steps S301 to S303.

[0084] Step S301 : Based on the distance distribution between the centroid of each area to be analyzed in each plane layer and the centroid of each area to be analyzed in the adjacent plane layer, combined with the regional edge extension distribution, the extension degree of the lesion between the area to be analyzed in each plane layer and the adjacent plane layer is obtained.

[0085] More specifically, if Figure 4 As shown, the method for obtaining the extension extent of the lesion can be implemented by steps S3011 to S3014.

[0086] Step S3011: any area to be analyzed in any plane layer is used as a first analysis area, and any area to be analyzed in a plane layer adjacent to the first analysis area is used as a second analysis area.

[0087] It should be noted that the method for obtaining the extension degree of the lesion in the area to be analyzed contained in each layer of the planar CT image is exactly the same. Here, the CT image of any layer of the plane is used as an example for explanation. It can be understood that this embodiment performs feature analysis on each layer of the plane and the adjacent next layer of the plane, and does not perform feature analysis on the adjacent next layer of the plane that cannot be obtained.

[0088] At the same time, in order to facilitate spatial feature analysis of different planes, it is necessary to map all plane images to the same spatial coordinate system. Specifically, the corresponding pixel points of the two-dimensional plane image are mapped to the three-dimensional spatial coordinate system. Each plane corresponds to a fixed Z-axis coordinate (that is, the position of each layer), and the X-axis and Y-axis coordinates correspond to the image pixel position.

[0089] Step S3012: determining a first parameter based on a spatial distance between the centroid of the first analysis area and the centroid of the second analysis area.

[0090] First, the coordinates of the center of mass of the first analysis area and the center of mass of the second analysis area in the spatial coordinate system are obtained. Then, based on the coordinates, the line segment distance between the center of mass of the first analysis area and the center of mass of the second analysis area can be calculated. As the spatial distance measurement result, the line segment distance calculated based on the coordinates is used as the first parameter. The smaller the value, the higher the possibility that the two areas to be analyzed in two adjacent planes show thrombus.

[0091] Step S3013 : determining a second parameter based on the angle between the straight line where the long side of the minimum circumscribed rectangle of the first analysis area lies and the straight line where the long side of the minimum circumscribed rectangle of the second analysis area lies.

[0092] The straight line on which the long side of the area to be analyzed lies reflects the approximate direction of the strip features within the corresponding area. When the angle between the strip features of the areas to be analyzed in two adjacent planes is small, it indicates that the strip features between the two may extend to each other. In other words, the greater the degree of correlation between the venous thrombosis features of the areas to be analyzed in two adjacent planes, the greater the possibility of common thrombosis.

[0093] As a specific example, the angle between the line containing the longest side of the minimum bounding rectangle of the first analysis area and the line containing the longest side of the minimum bounding rectangle of the second analysis area is normalized and used as the second parameter. The normalization method is well known and will not be described in detail here. It is understood that the angle between the two lines generally ranges from [0° to 90°].

[0094] Step S3014: determining the negative correlation coefficient of the Euclidean distance between the first parameter and the second parameter as the extension degree of the lesion between the first analysis region and the second analysis region.

[0095] The first parameter represents the distribution of the straight-line distance between two analyzed regions, corresponding to two different planes. The second parameter represents the likelihood of mutual extension between the two analyzed regions, corresponding to two different planes. The smaller the value, the higher the probability that the two analyzed regions, corresponding to two different planes, are mutually extended. Based on this, both the first and second parameters are negatively correlated with the extent of lesion extension.

[0096] As a specific example, the L2 norm of the first parameter and the second parameter is calculated, and the inverse of the L2 norm is used as the degree of extension of the lesion between the first analysis area and the second analysis area, which represents that the greater the possibility that the strip features of the first analysis area and the second analysis area belonging to the characteristics of venous thrombosis extend to each other, that is, the greater the degree of correlation between the thrombosis characteristics of the corresponding areas to be analyzed.

[0097] It should be noted that the above method for obtaining the extent of lesion extension is described in detail using an arbitrary region to be analyzed in a slice plane and an arbitrary region to be analyzed in an adjacent slice plane as an example. It is understood that to avoid computational redundancy due to an excessive number of regions to be analyzed, the subsequent analysis of the degree of correlation of thrombus characteristics will use the minimum bounding rectangle of any region to be analyzed in the current slice plane as the range, and obtain the regions to be analyzed in adjacent slice planes whose centroids fall within this range for correlation analysis.

[0098] Among them, when the centroid of the area to be analyzed on the adjacent layer plane is not within the range, it means that the planar distance difference between the area to be analyzed on the current layer plane and the adjacent layer plane is large, and it is unlikely that there will be a feature of mutual extension, so the correlation feature analysis is not performed.

[0099] In step S302, the area to be analyzed of each plane layer is divided to obtain the initial lesion area of ​​each plane layer. According to the degree of lesion extension between the initial lesion area of ​​each plane layer and the initial lesion area of ​​the adjacent plane layer in each division result, combined with the lesion possibility index, the division preference degree of each division result is obtained.

[0100] First, for any plane layer, different areas to be analyzed are used as initial lesion areas to obtain different segmentation results. Then, the effectiveness of the segmentation results is evaluated based on the degree of feature expression of the initial lesion area in each segmentation result.

[0101] Furthermore, for any division result, the product of the lesion possibility index between each initial lesion area of ​​each layer plane and each initial lesion area of ​​the adjacent layer plane is obtained as the first characteristic factor, and the second characteristic factor is determined based on the difference in lesion possibility index between each initial lesion area of ​​each layer plane and each initial lesion area of ​​the adjacent layer plane; the ratio of the first characteristic factor and the second characteristic factor and the product of the lesion extension degree are used as the probability index of the initial lesion area corresponding to each two adjacent layers planes; the probability indicators of all initial lesion areas corresponding to all adjacent layers planes are comprehensively used to obtain the division preference degree of the division result.

[0102] As a specific example, taking any partitioning method as an example, the calculation method of the partitioning preference degree can be expressed by the formula:

[0103] ;

[0104] in, Indicates the degree of partition preference of the current partitioning method corresponding to the partitioning result. Indicates the total number of layer planes, Indicates the number of initial lesion areas contained in the current segmentation result of the CT image of the i-th plane. Indicates the total number of initial lesion regions corresponding to the mth initial lesion region in the adjacent layer planes in the current segmentation result of the CT image of the i-th layer plane, Represents the lesion possibility index of the mth initial lesion area in the current segmentation result of the CT image of the i-th plane, The lesion possibility index of the mth initial lesion area in the current segmentation result of the CT image of the i-th plane and the rth initial lesion area corresponding to the adjacent i+1-th plane is represented. Indicates the lesion extension degree between the mth initial lesion region in the current segmentation result of the CT image of the i-th plane and the rth initial lesion region corresponding to the adjacent i+1-th plane.

[0105] is the first characteristic factor, As the second characteristic factor, it can avoid the influence of the calculation result when the difference of the lesion possibility index is 0. The ratio of the first characteristic factor to the second characteristic factor represents that the quantitative process of the evaluation result of the division method needs to encourage the combination of regions with high probability and small difference, and suppress the combination with large difference in probability. At the same time, combined with the extension degree of the lesion in the initial lesion area corresponding to the adjacent layer planes, it represents that the quantitative process of the evaluation result of the division method needs to strengthen the regional correlation with the consistent extension direction of the regional strip-like features. The division preference degree of the division result represents the degree of preference of the effect of the corresponding division method.

[0106] Step S303 : taking the initial lesion region in the division result corresponding to the maximum value of the division preference degree as the suspected lesion region of each plane layer.

[0107] The greater the value of the preferred degree of the division results corresponding to different division methods, the greater the possibility that venous thrombosis may exist in the initial lesion area of ​​the corresponding division method, which in turn reflects the better division effect of the corresponding division method. Therefore, the division results corresponding to the maximum value of the preferred degree of division in all division methods are determined as the final suspected lesion area in each layer plane. After completing the secondary screening operation of all closed areas, it can effectively distinguish between venous thrombosis and the surrounding normal tissue parts with similar grayscale values. The suspected lesion area represents the part of the CT image of the corresponding layer plane suspected of containing venous thrombosis.

[0108] At this point, by analyzing the differences in the extension angles of the long sides of the minimum circumscribed rectangles between different adjacent planes in different areas to be analyzed, the similarities and differences between the dense shadow trends in different areas to be analyzed are obtained, and the probability of thrombus extension, that is, the degree of lesion extension corresponding to the areas to be analyzed, is obtained. Combined with the lesion possibility index of the area to be analyzed to show the possibility of thrombus existence in the corresponding area, the area to be analyzed is screened twice, avoiding the problem of difficulty in segmenting venous thrombosis from normal areas due to certain differences in the CT manifestations of actual thrombus areas in different periods, different positions, and different planes, thereby improving the accuracy of venous thrombosis identification and reconstruction.

[0109] Step S400: Perform edge matching on the suspected lesion area of ​​each plane layer and the adjacent plane layer based on the distance distribution and relative position distribution between the edge pixel points of the suspected lesion area of ​​each plane layer and the edge pixel points of the suspected lesion area of ​​the adjacent plane layer; perform three-dimensional reconstruction of the patient's brain CT based on the matching results.

[0110] During the reconstruction process, it is usually necessary to fit the areas showing venous thrombosis in two adjacent planes. However, due to the differences in the actual location and size of the areas showing venous thrombosis in the two adjacent planes, and the existence of a sampling interval length between the two adjacent planes, fitting directly based on the venous area can easily lead to excessive sharpening of the edges of the reconstructed graphics, resulting in errors in the imaging contours. Considering that there are cord-like features and similar extension trends between the suspected lesion areas corresponding to adjacent planes, and the degree of correlation between the trends of the corresponding areas has been analyzed, in the process of matching features in different areas, when the extension trend between the corresponding matched edge pixels is close to the extension trend of the entire area, the matching method of the corresponding edge pixels is consistent with the feature expression.

[0111] Based on this feature, feature matching analysis of edge pixels is performed on each layer plane and the suspected lesion area corresponding to the adjacent layer plane, such as Figure 5 As shown, the method of performing edge matching on the suspected lesion area of ​​each layer plane and the adjacent layer plane can be implemented by steps S401 to S403.

[0112] Step S401, based on the distance distribution and position distribution between the edge pixel points of the suspected lesion area of ​​each layer plane and the edge pixel points of the suspected lesion area of ​​the adjacent layer plane, obtain the edge correlation degree between the edge pixel points in the suspected lesion area of ​​each layer plane and the adjacent layer plane.

[0113] Analogously to the method for obtaining the extension degree of the lesion between the areas to be analyzed on two adjacent planes, the correlation of the extension trend under the corresponding matching relationship of each pixel point on the edge of the region is refined through the angle representation of the extension trend between the corresponding edge pixels on the region and the distance distribution between the corresponding edge pixels.

[0114] Specifically, taking any suspected lesion region in any two adjacent planes as an example, in the first step, any edge pixel point in any suspected lesion region in any plane is used as the first pixel point of the first lesion region, the suspected lesion region in the plane adjacent to the first lesion region is used as the reference lesion region, and any edge pixel point in any reference lesion region is used as the second pixel point of the second lesion region. The second lesion region belongs to the next plane adjacent to the plane where the first lesion region is located.

[0115] The second step is to obtain a line segment connecting the first pixel point and the second pixel point; use the length of the line segment as the first coefficient; use the angle between the projection line of the line segment on the plane where the first lesion area is located and the straight line where the long side of the minimum circumscribed rectangle of the first lesion area is located as the first angle, and use the angle between the projection line of the line segment on the plane where the second lesion area is located and the straight line where the long side of the minimum circumscribed rectangle of the second lesion area is located as the second angle, and calculate the normalized mean of the first angle and the second angle to obtain the second coefficient.

[0116] It should be noted that the length of the line segment can be calculated using the corresponding coordinate values ​​of the first pixel and the second pixel in the three-dimensional coordinate system. Furthermore, the first angle and the second angle are both intended to obtain the distribution of angles between the projection line of the line connecting the two edge pixels on the corresponding plane and the extension of the suspected lesion area in each plane, respectively, and are analyzed using the same processing method as in step S3013.

[0117] In the third step, a negative correlation process is performed on the L2 norm of the first coefficient and the second coefficient to obtain the edge correlation degree between the first pixel point in the first lesion area and the second pixel point in the second lesion area. In this embodiment, the reciprocal of the L2 norm of the first coefficient and the second coefficient is used as the corresponding edge correlation degree.

[0118] Venous thrombi are distributed in a continuous three-dimensional pattern along the course of the blood vessels, and their CT images must be projected on different scanning slices to maintain spatial topological consistency. By calculating the distance and directional offset angle between edge pixels across slices, the degree of edge correlation can quantitatively assess the geometric correlation between edge pixels in thrombus regions on adjacent slices.

[0119] Because adjacent CT slices have different Z coordinates in the spatial coordinate system, if the suspected lesion areas in two adjacent CT slices are completely matched, the line connecting the corresponding edge pixels is perpendicular to the XOY axis plane. In this case, the length of the line segment connecting the two edge pixels with the optimal matching relationship is the shortest, and the angle between the projection of this line segment in the two adjacent slices and the corresponding extension of the slice area is the smallest. Based on this feature, the smaller the distance between the first pixel and the second pixel, and the smaller the average angle between the first angle and the second angle, the greater the probability that the two are in corresponding positions in the adjacent slices, and the greater the probability that the first pixel and the second pixel are a match.

[0120] In step S402, the edge pixel points between each two adjacent layers of the suspected lesion area are matched one by one to obtain different matching methods; according to the lesion extension degree between each two adjacent layers of the suspected lesion area and the edge correlation degree between the edge pixel points in the suspected lesion area under each matching method, the matching effect evaluation of each matching method for each suspected lesion area is obtained.

[0121] Taking the first lesion area and the corresponding reference lesion area as an example, the edge pixel points in the first lesion area and each reference lesion area are matched one by one, and the situation where there is no cross-matching is preserved. Several matching methods between the edge pixel points of the first lesion area and each reference lesion area can be obtained, and each matching method corresponds to several groups of matching pixel groups.

[0122] It should be noted that, taking any reference lesion area as an example, considering that the areas of the first lesion area and the second lesion area may be different, the area ratio between the first lesion area and the CT image of the layer plane where it is located is obtained as the first area ratio, and the area ratio between the second lesion area and the CT image of the layer plane where it is located is obtained as the second area ratio. The ratio between the first area ratio and the second area ratio is used as the area ratio coefficient. The area ratio coefficient can reflect the scaling of the number of edge pixels in the first lesion area.

[0123] When the area ratio coefficient is greater than 1, it means that the relative area of ​​the first lesion area is greater than the relative area of ​​the second lesion area. At this time, each edge pixel point of the second lesion area needs to correspond to multiple edge pixels of the first lesion area. At this time, the maximum number of matching edges of any edge pixel point in the second lesion area is for , Indicates the total number of edge pixels in the second lesion area, Represents the area ratio coefficient between the first lesion area and the second lesion area.

[0124] When the area ratio coefficient is less than 1, it means that the relative area of ​​the first lesion area is smaller than the relative area of ​​the second lesion area. At this time, each edge pixel point in the first lesion area needs to correspond to multiple edge pixels in the first lesion area. At this time, the maximum number of matching edges of any edge pixel point in the first lesion area is for , Represents the total number of edge pixels in the first lesion area. By calculating the maximum number of matches, full edge matching of the two areas can be achieved without cross matching.

[0125] Furthermore, for the matching mode between the first lesion region and each corresponding reference lesion region, a matching effect evaluation corresponding to a matching mode can be obtained between the first lesion region and each reference lesion region.

[0126] Specifically, for any matching method between the first lesion area and the second lesion area, the negative correlation coefficient between the edge association degree corresponding to each matching pixel group and the lesion extension degree between the first lesion area and the second lesion area is calculated to obtain the matching coefficient of each matching pixel group, and the average value of the matching coefficients of all matching pixel groups in the matching method is calculated to obtain the matching effect evaluation between the first lesion area and the second lesion area in the matching method.

[0127] The matching pixel group refers to a combination of two edge pixel points having a matching relationship in any two adjacent layer planes in any matching mode, and the two edge pixel points belong to two adjacent layer planes respectively.

[0128] It can be understood that for the first pixel point in the first lesion area, under any matching method, there is a matching relationship between the first pixel point of the first lesion area and the edge pixel point of the second lesion area, that is, the first pixel point may correspond to multiple matching pixel groups. First, feature analysis is performed on the matching pixel group corresponding to each edge pixel point of the first lesion area, and then the matching analysis results of all edges on the first lesion area are combined to obtain the matching effect evaluation between the first lesion area and the second lesion area.

[0129] As a specific example, the calculation formula for evaluating the matching effect between the first lesion region and the second lesion region can be expressed as:

[0130] ;

[0131] in, represents the matching effect evaluation between the first lesion area and the second lesion area, Indicates the number of edge pixels contained in the first lesion area, represents the number of matching pixel groups corresponding to the a-th edge pixel point in the first lesion area, Indicates the edge correlation degree of the bth matching pixel group corresponding to the ath pixel point in the first lesion area, represents the extent of lesion extension between the first lesion area and the second lesion area, is the normalization function, Represents an exponential function with the natural constant e as its base.

[0132] When the ratio of the edge correlation feature between edge pixels and the extension trend correlation feature between the corresponding suspected lesion areas is The closer it is to 1, the greater the degree of geometric consistency between the characteristic trend corresponding to the edge pixel point and the characteristic trend of the entire region. At this time, the edge matching effect between the two is better. Therefore, by performing negative correlation processing on the difference, the matching effect evaluation corresponding to the edge matching result between the two suspected lesion areas is determined. The larger the value of the matching effect evaluation, the better the corresponding matching method.

[0133] Step S403 : Based on the matching mode corresponding to the maximum value of the matching effect evaluation, edge matching is performed on the suspected lesion area of ​​each plane layer and the adjacent plane layers.

[0134] For the first lesion region, the matching method for the reference lesion region corresponding to the maximum matching effect evaluation is obtained, and edge matching is performed on the first lesion region and the reference lesion region. The same method can be used to edge match each suspected lesion region in each slice and adjacent slices. This avoids the problem of oversharpening of the reconstructed image caused by the difference in the location and size of the region representing venous thrombosis in two adjacent slices, thereby improving the accuracy of the final reconstructed image.

[0135] Finally, based on the matching results, the patient's brain CT is three-dimensionally reconstructed. It is understandable that this embodiment only performs feature analysis on the venous thrombosis part that is prone to misjudgment, and does not limit the matching method for other normal parts. The implementer can choose the corresponding well-known technology for processing according to the specific implementation scenario. The optimized matching relationship in this embodiment can be used as a control point for triangulation or surface fitting to generate a smooth blood vessel model, effectively avoiding sharpening distortion. In other embodiments, suspected lesion areas can also be marked as suspected thrombosis for review by professional doctors.

[0136] It should be noted that 3D reconstruction can be performed using the CT machine's built-in post-processing workstation. Reconstruction methods include multiplanar reformation (MPR), volume rendering (VR), and maximum intensity projection (MIP). MPR reconstruction includes coronal, sagittal, and curved planar reformation (CPR) views, with a slice thickness of 2.0 mm, an interval of 2.0 mm, a window width of 250 Hu, and a window position of 50 Hu. CT 3D reconstruction methods are well known and will not be further elaborated here.

[0137] In summary, this embodiment performs a preliminary division of tissue regions within a single-slice CT image based on the color and morphological differences between venous vessels and thrombus regions compared to other normal tissues, avoiding misdiagnosis and missed diagnosis caused by direct differentiation and identification based on grayscale values ​​(Hu values), thereby improving the accuracy of venous thrombus region identification. Based on the likelihood of thrombus manifestation, all closed regions within different planes are analyzed for preliminary screening, avoiding the difficulty of screening and redundancy in the identification process caused by the large number of closed regions within different planes, thereby improving the efficiency of venous thrombus identification. Furthermore, by analyzing the differences in the extended angles of the long sides of the minimum circumscribed rectangles between adjacent planes between different closed regions, similarities and differences between the dense shadow trends of different closed regions are obtained, and the probability of thrombus extension is obtained. A secondary screening is performed based on the likelihood of thrombus manifestation, avoiding misdiagnosis and missed diagnosis of thrombus regions, thereby improving the accuracy of venous thrombus identification and reconstruction.

[0138] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A multi-planar cranial CT reconstruction method for patients with cerebral venous thrombosis, characterized in that: The method comprises the following steps: Obtaining CT images of each plane of the patient's CT scan and extracting closed areas of each CT image; According to the shape distribution and grayscale distribution of each closed area in each CT image, the closed area is screened to obtain the area to be analyzed in each plane layer; According to the angle distribution of the extension lines between the area to be analyzed on each plane and the area to be analyzed on the adjacent planes, as well as the area distance distribution, the area to be analyzed on each plane is divided to obtain the suspected lesion area on each plane; Based on the distance distribution and relative position distribution between the edge pixels of the suspected lesion area on each plane and the edge pixels of the suspected lesion area on the adjacent plane, edge matching is performed on the suspected lesion area on each plane and the adjacent plane; based on the matching results, a three-dimensional reconstruction of the patient's brain CT is performed; The method for obtaining the suspected lesion area includes: According to the distance distribution between the centroid of each area to be analyzed in each plane and the centroid of each area to be analyzed in the adjacent plane, combined with the distribution of regional edge extension, the extension degree of the lesion between the area to be analyzed in each plane and the adjacent plane is obtained; The area to be analyzed of each plane layer is divided to obtain an initial lesion area of ​​each plane layer, and the division preference of each division result is obtained based on the lesion extension degree between the initial lesion area of ​​each plane layer and the initial lesion area of ​​the adjacent plane layer in each division result, combined with the lesion possibility index; the lesion possibility index is obtained by: obtaining the characteristic expression degree of each closed area in each CT image based on the shape distribution of the minimum circumscribed rectangle of each closed area in each CT image; and determining the lesion possibility index of each closed area in each CT image based on the mean grayscale value of all pixels in each closed area in each CT image and the characteristic expression degree; The initial lesion area in the division result corresponding to the maximum value of the division preference degree is used as the suspected lesion area of ​​each plane layer; The edge matching of the suspected lesion area of ​​each plane and the adjacent plane specifically includes: According to the distance distribution and position distribution between the edge pixels of the suspected lesion area of ​​each plane and the edge pixels of the suspected lesion area of ​​the adjacent plane, the edge correlation degree between the edge pixels of the suspected lesion area of ​​each plane and the adjacent plane is obtained; The edge pixels between the suspected lesion areas of each two adjacent planes are matched one by one to obtain different matching methods; According to the lesion extension degree between each two adjacent layers of suspected lesion regions and the edge correlation degree between edge pixels in the suspected lesion regions under each matching method, the matching effect evaluation of each matching method for each suspected lesion region is obtained; Based on the matching method corresponding to the maximum value of the matching effect evaluation, edge matching is performed on the suspected lesion area of ​​each layer plane and the adjacent layer plane.

2. The multi-planar brain CT reconstruction method for patients with cerebral venous thrombosis according to claim 1, characterized in that: The closed regions are screened based on the shape distribution and grayscale distribution of each closed region in each CT image to obtain the region to be analyzed on each plane layer, specifically including: All closed areas in each CT image are clustered, and the evaluation index of each clustering result is obtained according to the fluctuation of the lesion possibility index of the closed areas in the same cluster in each clustering result; In the clustering results corresponding to the maximum value of the evaluation index, all closed areas corresponding to the maximum value of the balance of the lesion possibility index of all closed areas in the cluster are determined as the areas to be analyzed on the corresponding layer plane of the CT image.

3. The multi-planar brain CT reconstruction method for patients with cerebral venous thrombosis according to claim 2, characterized in that: The step of obtaining the characteristic expression degree of each closed area in each CT image according to the shape distribution of the minimum circumscribed rectangle of each closed area in each CT image specifically includes: Based on the area ratio of each closed area in each CT image to the corresponding minimum circumscribed rectangle, a first ratio coefficient is determined. Based on the aspect ratio of the minimum circumscribed rectangle of each closed area in each CT image, a second ratio coefficient is determined. The first ratio coefficient and the second ratio coefficient are combined to obtain the characteristic expression degree of each closed area in each CT image.

4. The multi-planar brain CT reconstruction method for patients with cerebral venous thrombosis according to claim 2, characterized in that: The evaluation index of each clustering result is obtained according to the fluctuation of the lesion possibility index of the closed area in the same cluster in each clustering result, specifically including: For any clustering result, the characteristic dispersion coefficient of each cluster is determined based on the dispersion degree of the lesion possibility index of all closed areas in the same cluster; The global dispersion coefficient of the clustering result is determined based on the dispersion degree of the mean value of the lesion possibility index of all closed areas in each cluster; The evaluation index of the clustering result is determined based on the ratio of the cumulative sum of the characteristic dispersion coefficients of all clusters to the global dispersion coefficient, combined with the negative correlation coefficient of the proportion of clusters contained in the clustering result.

5. The multi-planar brain CT reconstruction method for patients with cerebral venous thrombosis according to claim 1, characterized in that: The method of obtaining the extent of lesion extension between the areas to be analyzed on each plane and the adjacent planes according to the distance distribution between the centroid of each area to be analyzed on each plane and the centroid of each area to be analyzed on the adjacent planes, combined with the area edge extension distribution, specifically includes: Any area to be analyzed in any plane layer is used as the first analysis area, and any area to be analyzed in the plane layer adjacent to the first analysis area is used as the second analysis area; Determine the first parameter based on the spatial distance between the centroid of the first analysis area and the centroid of the second analysis area; determine the second parameter based on the angle between the straight line containing the long side of the minimum circumscribed rectangle of the first analysis area and the straight line containing the long side of the minimum circumscribed rectangle of the second analysis area; The negative correlation coefficient of the Euclidean distance between the first parameter and the second parameter is determined as the extent of lesion extension between the first analysis region and the second analysis region.

6. The multi-planar brain CT reconstruction method for patients with cerebral venous thrombosis according to claim 1, characterized in that: The method of obtaining the division preference of each division result based on the extension degree of the lesion between the initial lesion area of ​​each plane and the initial lesion area of ​​the adjacent plane in each division result and the lesion possibility index specifically includes: For any division result, the product of the lesion possibility index between each initial lesion region in each layer plane and each initial lesion region in the adjacent layer plane is obtained as the first characteristic factor, and the second characteristic factor is determined based on the difference in the lesion possibility index between each initial lesion region in each layer plane and each initial lesion region in the adjacent layer plane; The ratio of the first characteristic factor to the second characteristic factor and the product of the lesion extension degree are used as the probability index of the initial lesion area corresponding to each two adjacent planes; the probability indexes of all initial lesion areas corresponding to all adjacent planes are combined to obtain the division preference degree of the division result.

7. The multi-planar brain CT reconstruction method for patients with cerebral venous thrombosis according to claim 1, characterized in that: The step of obtaining the edge correlation degree between the edge pixels in the suspected lesion area of ​​each plane and the suspected lesion area of ​​the adjacent plane based on the distance distribution and position distribution between the edge pixels in the suspected lesion area of ​​each plane and the edge pixels in the suspected lesion area of ​​the adjacent plane specifically includes: Any edge pixel point in any suspected lesion area on any plane is used as the first pixel point of the first lesion area, the suspected lesion area on the plane adjacent to the first lesion area is used as the reference lesion area, and any edge pixel point in any reference lesion area is used as the second pixel point of the second lesion area; Obtain a line segment connecting a first pixel point and a second pixel point; use the length of the line segment as a first coefficient; use the angle between the projection line of the line segment on the plane where the first lesion area is located and the straight line containing the long side of the minimum circumscribed rectangle of the first lesion area as a first angle; use the angle between the projection line of the line segment on the plane where the second lesion area is located and the straight line containing the long side of the minimum circumscribed rectangle of the second lesion area as a second angle; and calculate the normalized mean of the first angle and the second angle to obtain a second coefficient; Negative correlation processing is performed on the L2 norms of the first coefficient and the second coefficient to obtain the edge correlation degree between the first pixel point of the first lesion area and the second pixel point of the second lesion area.

8. The multi-planar brain CT reconstruction method for patients with cerebral venous thrombosis according to claim 7, characterized in that: The matching effect evaluation of each matching method is obtained according to the lesion extension degree between each two adjacent layers of suspected lesion areas and the edge correlation degree between edge pixels in the suspected lesion areas under each matching method, specifically including: For any matching method between the first lesion area and the second lesion area, the negative correlation coefficient between the edge association degree corresponding to each matching pixel group and the lesion extension degree between the first lesion area and the second lesion area is calculated to obtain the matching coefficient of each matching pixel group, and the average value of the matching coefficients of all matching pixel groups in the matching method is calculated to obtain the matching effect evaluation between the first lesion area and the second lesion area in the matching method.

Images