Aortic dissection postoperative distal remodeling risk prediction model based on multi-modal data fusion

Abstract

This invention relates to the field of personal health risk assessment technology, specifically to a multimodal data fusion model for predicting distal remodeling risk after aortic dissection. This model, through the cooperation of multiple modules, achieves the following steps: acquiring a target CT cross-sectional image and vessel wall pulsation pressure difference of the target patient after aortic dissection, and identifying the true lumen region and false lumen region; determining the thrombus density block value; determining the pressure source node set and the bearing end node set; determining the local fluid conduction resistance; thereby determining the total radial osmotic resistance matrix; and using the total radial osmotic resistance matrix and vessel wall pulsation pressure difference to assist in predicting distal remodeling risk. This invention, through the fusion of multimodal data such as image data and vessel wall pulsation pressure difference, achieves assisted prediction of distal remodeling risk after aortic dissection, improves the rationality of assisted prediction of distal remodeling risk after aortic dissection, and thus improves the auxiliary effect.

Description

Technical Field

[0001] This invention relates to the field of personal health risk assessment technology, specifically to a multimodal data fusion model for predicting the risk of distal remodeling after aortic dissection surgery. Background Technology

[0002] With the development of technology, the application of assisting doctors in risk prediction is becoming increasingly widespread. For example, it can be used to assist doctors in predicting the risk of distal remodeling after aortic dissection. Currently, the common method used to assist in predicting the risk of distal remodeling after aortic dissection is to use the thrombus filling status in the false lumen area in CT (Computed Tomography) images to assist in predicting the risk of distal remodeling after aortic dissection.

[0003] However, when using the thrombus filling status within the false lumen area in CT images to assist in predicting the risk of distal remodeling after aortic dissection, the following technical problems often arise: In reality, although the false lumen of some patients is morphologically filled with thrombus, its diameter continues to expand, eventually leading to distal rupture. The reason for this phenomenon is that the degree of organization of thrombus in the lumen varies among different patients, and the resulting fluid permeability resistance often differs as well, thus affecting the distal remodeling effect after aortic dissection. Therefore, if only the thrombus filling status in the false lumen area is considered when assisting in the prediction of distal remodeling risk after aortic dissection, it may lead to poor rationality in the prediction of distal remodeling risk after aortic dissection, resulting in poor auxiliary effect. Summary of the Invention

[0004] To address the technical problem of poor auxiliary effects due to the poor rationality of the prediction of distal remodeling risk after aortic dissection, this invention proposes a multimodal data fusion model for predicting the risk of distal remodeling after aortic dissection.

[0005] In a first aspect, the present invention provides a multimodal data fusion model for predicting the risk of distal remodeling after aortic dissection, the model comprising: The data acquisition and region identification module is used to acquire the target CT cross-sectional image and the pulsation pressure difference of the vessel wall of the target patient after aortic dissection surgery, and to identify the true lumen region and the false lumen region from the target CT cross-sectional image. The thrombus density block value determination module is used to determine the thrombus density block value corresponding to each pixel in the false lumen region based on the CT value corresponding to each pixel in the false lumen region. The intersection set determination module is used to determine the pressure source node set and the bearing end node set based on the distribution of intersection points between a preset number of rays constructed from the center point of the true cavity region and the false cavity region. The local fluid conduction resistance determination module is used to determine the local fluid conduction resistance between each two pixels in the false lumen region based on the distance between each two pixels in the false lumen region, as well as the thrombus density retardation value and grayscale gradient modulus corresponding to these two pixels. The radial permeation resistance matrix determination module is used to determine the radial permeation resistance matrix based on the local fluid conduction resistance between different pixels on the path formed between the pixels in the pressure source node set and the pixels in the bearing end node set. The remote remodeling risk prediction module is used to assist in predicting the risk of remote remodeling based on the radial permeation resistance matrix and the pipe wall pulsation pressure difference.

[0006] In conjunction with the first aspect above, in one possible implementation, determining the thrombus density block value corresponding to each pixel within the false lumen region based on the CT value corresponding to each pixel within the false lumen region includes: The mean CT value of all pixels in the true cavity region is determined as the reference baseline for the true cavity blood pool. Based on the difference between the true cavity blood pool reference benchmark and the preset solid tissue benchmark, and the difference between the CT value corresponding to each pixel in the false cavity region and the preset solid tissue benchmark, the density reference factor corresponding to each pixel in the false cavity region is determined. Based on the density reference factor corresponding to each pixel in the false cavity region, determine the relative density factor corresponding to each pixel in the false cavity region; The thrombus density block value corresponding to each pixel in the false lumen region is determined by the square of the relative density factor corresponding to each pixel in the false lumen region.

[0007] In conjunction with the first aspect above, in one possible implementation, determining the relative density factor corresponding to each pixel within the pseudo-cavity region based on the density reference factor corresponding to each pixel within the pseudo-cavity region includes: Any pixel within the false cavity region is designated as the marked pixel. If the density reference factor corresponding to the marked pixel is greater than a constant 1, then the relative density factor corresponding to the marked pixel is set to a constant 1. If the density reference factor corresponding to the marked pixel is less than a constant 0, then the relative density factor corresponding to the marked pixel is set to a constant 0. If the density reference factor corresponding to the marked pixel is greater than or equal to a constant 0 and less than or equal to a constant 1, then the relative density factor corresponding to the marked pixel is set as its corresponding density reference factor.

[0008] In conjunction with the first aspect above, in one possible implementation, determining the pressure source node set and the bearing end node set based on the distribution of intersection points between a predetermined number of rays constructed from the center point of the true cavity region and the false cavity region includes: With the center point of the true cavity region as the endpoint, draw a predetermined number of rays in different directions, which are denoted as the initial rays; If the initial ray intersects with both the intima and the outer wall of the blood vessel, then the initial ray is identified as the target ray. The pressure source nodes and bearing end nodes are selected from each target ray; All pressure source nodes are combined into a pressure source node set, and all bearing end nodes are combined into a bearing end node set.

[0009] In conjunction with the first aspect above, in one possible implementation, the step of screening out the pressure source node and the bearing end node from each target ray includes: Any target ray is designated as the marker ray, and the center point of the true cavity region is designated as the marker reference point; The intersection point closest to the marked reference point is selected from all intersection points of the marked ray and the inner diaphragm and denoted as the pressure source node; From all the intersections between the marked ray and the outer wall of the blood vessel, select the intersection point that is farthest from the marked reference point and denote it as the bearing end node.

[0010] In conjunction with the first aspect above, in one possible implementation, determining the local fluid conduction resistance between every two pixels within the false lumen region, based on the distance between each pair of pixels within the false lumen region, and the thrombus density retardation value and grayscale gradient modulus corresponding to those two pixels, includes: The material density resistance weight between each two pixels in the false lumen region is determined based on the distance between each two pixels in the false lumen region and the mean value between the thrombus density resistance values ​​corresponding to each two pixels in the false lumen region. The relative gradient intensity of each pixel in the false cavity region is determined based on the gray-level gradient magnitude of each pixel in the false cavity region and the maximum value of the gray-level gradient magnitude of all pixels in the false cavity region. The layered interface penetration coefficient corresponding to each pixel in the false cavity region is determined based on the relative gradient intensity of each pixel in the false cavity region. The local fluid conduction resistance between each pair of pixels in the false cavity region is determined based on the material density resistance weight between each pair of pixels in the false cavity region and the average value of the layered interface permeability coefficient corresponding to each pair of pixels in the false cavity region.

[0011] In conjunction with the first aspect above, in one possible implementation, determining the total radial permeation resistance matrix based on the local fluid conduction resistance between different pixels along the path formed by the pixels in the pressure source node set and the pixels in the bearing end node set includes: Based on whether each pixel in the false cavity region and its preset neighboring pixels belong to the false cavity region, adaptive connections are made between each pixel in the false cavity region and its preset neighboring pixels to obtain the undirected connection edges corresponding to each pixel in the false cavity region. A topological network is constructed by taking all pixels within the false cavity region as nodes and all their corresponding undirected edges as nodes. Using the local fluid conduction resistance between every two connected nodes in the topology network as the path length between these two nodes, the Dijkstra algorithm is used to obtain the total path length corresponding to the shortest connected path in the topology network for each pixel in the pressure source node set and each pixel in the bearing end node set, which is denoted as the total radial permeation resistance between each pixel in the pressure source node set and each pixel in the bearing end node set. The total radial permeation resistance matrix is ​​formed by the total radial permeation resistance between all pixels in the pressure source node set and all pixels in the bearing end node set.

[0012] In conjunction with the first aspect above, in one possible implementation, the step of assisting in the prediction of distal remodeling risk based on the total radial permeation resistance matrix and the pipe wall pulsation pressure difference includes: Based on the radial permeation resistance matrix, the bearing end nodes that match each pixel in the pressure source node set are selected from the bearing end node set and used as the bearing end matching points corresponding to each pixel in the pressure source node set. The overall barrier capacity value of the false cavity is determined based on the radial permeation resistance between the pixels in the pressure source node set and their corresponding bearing end matching points. The distal expansion risk index is determined based on the overall barrier capacity value of the false lumen and the pulsating pressure difference of the vessel wall; The risk of remote restructuring is predicted using the remote expansion risk index.

[0013] In conjunction with the first aspect above, in one possible implementation, determining the overall barrier capacity value of the false cavity based on the total radial permeation resistance between pixels in the pressure source node set and their corresponding bearing end matching points includes: If the radial permeation resistance between a pixel in the pressure source node set and its corresponding bearing end matching point is equal to a preset maximum value, then the pixel in the pressure source node set is determined to be an invalid pixel. Pixels in the pressure source node set other than invalid pixels are identified as valid pixels. If the number of valid pixels is a constant of 0, the overall barrier capability value of the false cavity is set to a preset maximum value. If the number of effective pixels is greater than the constant 0, the overall barrier capacity value of the false cavity is determined based on the number of effective pixels and the average of the radial permeation resistance between all effective pixels and their corresponding bearing end matching points.

[0014] In conjunction with the first aspect above, in one possible implementation, the step of using the remote expansion risk index to assist in predicting remote reshaping risk includes: If the remote expansion risk index is less than or equal to the preset risk threshold, the remote remodeling is deemed to be good. If the distal expansion risk index is greater than the preset risk threshold, it indicates that the false cavity has a high risk of continuous expansion or rupture.

[0015] Secondly, the present invention provides a method for predicting the risk of distal remodeling after aortic dissection using a multimodal data fusion model, the method comprising: Acquire target CT cross-sectional images and vessel wall pulsation pressure difference of target patients after aortic dissection surgery, and identify the true lumen region and false lumen region from the target CT cross-sectional images; Based on the CT value corresponding to each pixel within the false lumen region, determine the thrombus density block value corresponding to each pixel within the false lumen region. Based on the distribution of the intersection points between a predetermined number of rays constructed from the center point of the true cavity region and the false cavity region, the set of pressure source nodes and the set of bearing end nodes are determined. Based on the distance between every two pixels in the false lumen region, as well as the thrombus density retardation value and grayscale gradient modulus corresponding to these two pixels, the local fluid conduction resistance between every two pixels in the false lumen region is determined. The total radial permeation resistance matrix is ​​determined based on the local fluid conduction resistance between different pixels on the path formed between the pixels in the pressure source node set and the pixels in the bearing end node set. Based on the radial permeation resistance matrix and the wall pulsation pressure difference, the risk of remote remodeling is predicted.

[0016] Thirdly, a server is provided, including a memory and a processor. The memory stores executable program code, and the processor calls and runs the executable program code from the memory, enabling the device to perform the aforementioned multimodal data fusion method for predicting the risk of distal remodeling after aortic dissection.

[0017] Fourthly, a computer program product is provided, comprising: computer program code, which, when run on a computer, causes the computer to execute the aforementioned method for predicting the risk of distal remodeling after aortic dissection using multimodal data fusion.

[0018] Fifthly, a computer-readable storage medium is provided that stores computer program code, which, when executed on a computer, causes the computer to perform the aforementioned method for predicting the risk of distal remodeling after aortic dissection using multimodal data fusion.

[0019] The present invention has the following beneficial effects: The multimodal data fusion model for predicting the risk of distal remodeling after aortic dissection in this invention achieves auxiliary prediction of the risk of distal remodeling after aortic dissection by fusing multimodal data such as image data and vessel wall pulsation pressure difference. This solves the technical problem of poor auxiliary effect due to the poor rationality of the auxiliary prediction of distal remodeling risk after aortic dissection, and improves the rationality of the auxiliary prediction of distal remodeling risk after aortic dissection, thereby improving the auxiliary effect. Specifically, this invention quantifies multiple indicators related to fluid osmotic resistance, such as thrombus density resistance and local fluid conduction resistance, by analyzing the grayscale and gradient of CT cross-sectional images, thereby quantifying the total radial osmotic resistance matrix. Combined with vessel wall pulsation pressure difference, this matrix assists in the prediction of distal remodeling risk, achieving auxiliary prediction of the risk of distal remodeling after aortic dissection and improving the rationality of the auxiliary prediction of distal remodeling risk after aortic dissection, thus improving the auxiliary effect. Attached Figure Description

[0020] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a schematic diagram of the structure of a multimodal data fusion model for predicting the risk of distal remodeling after aortic dissection according to the present invention. Figure 2 This is a flowchart of a method for predicting the risk of distal remodeling after aortic dissection using multimodal data fusion, according to the present invention. Figure 3 This is a schematic diagram of the structure of a computer device according to the present invention. Detailed Implementation

[0022] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the specific implementation methods, structures, features, and effects of the technical solution proposed according to the present invention are described in detail below with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0023] Unless otherwise defined, 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 pertains.

[0024] Aortic dissection surgery can be used to treat aortic dissection, with TEVAR (Thoracic Endovascular Aortic Repair) being a common procedure. After thoracic endovascular repair of the aortic dissection, the distal remodeling effect of the false lumen often depends on the balance between the degree of organization of the intraluminal thrombus and the hemodynamic load. Ideally, the false lumen is remodeled into a dense, organized thrombus that acts as a strong physical barrier, effectively blocking blood pressure from reaching the vessel wall, thus promoting the shrinkage of the false lumen. However, clinically, it is often observed that in some patients, although the false lumen is morphologically filled with thrombus, its diameter continues to expand, eventually leading to distal rupture. This phenomenon is caused by the fact that current assessment methods mainly rely on CT imaging to measure changes in the diameter or volume of the false lumen. This method only focuses on the external geometry, ignoring the microstructural differences within the false lumen filler, and relies on the measurement of the false lumen diameter to assess the remodeling status, only observing the results after macroscopic deformation of the vessel wall, exhibiting significant lag. In fact, thrombi that appear to have similar densities on imaging may contain layered structures formed at different times. The binding forces at the interfaces of these layers are weak, making them highly susceptible to forming invisible, latent flow channels under blood pressure. This results in the false lumen walls thickening without forming an effective pressure barrier. While existing computational fluid dynamics methods can simulate flow fields, they are difficult to accurately define boundary conditions and are computationally expensive. Therefore, this invention provides a method to quantify the impact of the internal layered structure of thrombi on fluid permeability resistance based on conventional CT images and to predict remodeling risk by combining systemic blood pressure indicators.

[0025] refer to Figure 1 This diagram illustrates the structure of a multimodal data fusion-based model for predicting the risk of distal remodeling after aortic dissection according to the present invention. The multimodal data fusion-based model for predicting the risk of distal remodeling after aortic dissection includes: The data acquisition and region identification module 101 is used to acquire the target CT cross-sectional image and the pulsation pressure difference of the aortic wall after the aortic dissection surgery of the target patient, and to identify the true lumen region and the false lumen region from the target CT cross-sectional image.

[0026] The target patients can be those undergoing distal remodeling risk assessment after aortic dissection surgery. Aortic dissection surgery can be any surgical procedure to treat aortic dissection, with TEVAR (Thoracic Endovascular Aortic Repair) being a common example. The target CT cross-sectional image can be an enhanced CT cross-sectional image of the aorta acquired after the target patient's aortic dissection surgery. Enhanced CT cross-sectional images of the aorta can be acquired using a multi-slice spiral CT (Computed Tomography) scanner (MSCT). In practice, an enhanced CT cross-sectional image of the aorta refers to a cross-sectional image of the aorta perpendicular to the long axis (horizontal plane) of the human body obtained through a CT scan; "enhanced" refers to the injection of a contrast agent (usually iodine contrast agent) into the blood vessel before the scan, making the blood in the aorta appear bright white (high CT value) on the CT image, thus clearly distinguishing the true lumen, false lumen, intima-tie, and thrombus within the false lumen.

[0027] The pressure difference caused by pulsation in the blood vessel walls can be equal to the difference between systolic and diastolic blood pressure. Systolic blood pressure, also known as high pressure, refers to the highest pressure exerted on the arterial walls when the heart contracts, pumping blood from the left ventricle into the aorta. Diastolic blood pressure, also known as low pressure, refers to the lowest pressure exerted on the arterial walls when the heart relaxes, the aortic valves close, the heart stops ejecting blood, and the arteries elastically recoil, allowing blood to continue flowing. Systolic and diastolic blood pressure can be obtained using a sphygmomanometer.

[0028] The true cavity region can be obtained using existing techniques such as region growing algorithms, threshold segmentation techniques, or neural network techniques. For example, threshold segmentation techniques can be used to identify the connected regions with the highest contrast agent filling degree in the target CT cross-sectional image and record them as the true cavity region.

[0029] False cavity regions can be obtained using existing techniques such as threshold segmentation or neural network techniques. For example, based on a target CT cross-sectional image, regions adjacent to the true cavity region and separated by an inner membrane sheet can be identified and defined as false cavity regions.

[0030] The thrombus density block value determination module 102 is used to determine the thrombus density block value corresponding to each pixel in the false lumen region based on the CT value corresponding to each pixel in the false lumen region.

[0031] As an example, determining the thrombus density block value corresponding to each pixel within the false lumen region may include the following steps: The first step is to determine the mean CT value of all pixels in the true cavity region as the reference baseline for the true cavity blood pool.

[0032] It should be noted that the CT value of the true lumen blood pool (containing contrast agent) of the aorta is extremely high, often exceeding 200 HU.

[0033] The second step is to determine the density reference factor for each pixel in the false cavity region based on the difference between the true cavity blood pool reference benchmark and the preset solid tissue reference benchmark, as well as the difference between the CT value of each pixel in the false cavity region and the preset solid tissue reference benchmark.

[0034] The preset solid tissue benchmark can be an empirical CT value representing a fully organized thrombus or soft tissue, which can be pre-set according to the actual situation, such as 40 HU.

[0035] For example, the formula for determining the density reference factor corresponding to a pixel within the pseudo-cavity region can be: ; in, It is the density reference factor corresponding to the i-th pixel within the pseudo-cavity region. i is the index of the pixel within the pseudo-cavity region. It is the CT value corresponding to the i-th pixel within the false cavity region. It is a preset solid tissue reference. It is the reference standard for the true cavity blood pool.

[0036] It should be noted that due to differences in tube voltage settings among different CT scanning devices and variations in contrast agent metabolism rates within patients, the absolute CT value (HU value) of an image often cannot be directly used as a physical quantity for cross-patient comparisons. To eliminate this systematic bias, a dynamic reference scale based on the patient's own blood density can be established; therefore, [the following is introduced]. and . It can characterize relative density to some extent.

[0037] The third step, determining the relative density factor for each pixel within the pseudo-cavity region based on the density reference factor, may include the following sub-steps: The first sub-step involves identifying any pixel within the false cavity region as the marked pixel.

[0038] In the second sub-step, if the density reference factor corresponding to the above-mentioned marked pixel is greater than a constant 1, it means that the density is higher than that of blood. Then, the relative density factor corresponding to the above-mentioned marked pixel is set to a constant 1.

[0039] In the third sub-step, if the density reference factor corresponding to the above-mentioned marked pixel is less than a constant 0, it means that the density is lower than that of soft tissue. Then, the relative density factor corresponding to the above-mentioned marked pixel is set to a constant 0.

[0040] The fourth sub-step is to set the relative density factor of the marked pixel to its corresponding density reference factor if the density reference factor is greater than or equal to constant 0 and less than or equal to constant 1.

[0041] It should be noted that, in order to ensure the physical meaning of the numerical values, the calculation results are truncated so that the relative density factor ranges from 0 to 1.

[0042] The fourth step is to determine the thrombus density block value corresponding to each pixel in the false lumen region based on the square of the relative density factor corresponding to each pixel in the false lumen region.

[0043] For example, the formula for determining the thrombus density block value corresponding to a pixel within the false lumen region can be: ; in, It is the thrombus density block value corresponding to the i-th pixel within the false lumen region. i is the index of the pixel within the false lumen region. It is the relative density factor corresponding to the i-th pixel within the pseudo-cavity region. It is an adjustment factor that is preset according to the actual situation, mainly used to prevent the denominator from being 0. For example, it can be 0.001.

[0044] It should be noted that all variables in the embodiments of the present invention, especially the denominator, can be adjusted by adding corresponding adjustment factors according to the actual situation, so as to adjust their value range or prevent the denominator from being 0.

[0045] It should be noted that the filling medium within the false lumen is typically composed of a mixture of liquid blood, semi-coagulated thrombus, and organized thrombus. The density of this material directly determines the microscopic resistance to fluid permeation. Higher density and lower porosity make fluid permeation more difficult. Therefore, it is necessary to convert medical grayscale information into physical barrier information. The larger the value, the denser the material represented by the i-th pixel in the false cavity region, and the stronger the resistance to fluid flow.

[0046] The intersection set determination module 103 is used to determine the pressure source node set and the bearing end node set based on the distribution of the intersection points between a preset number of rays constructed from the center point of the true cavity region and the false cavity region.

[0047] The center point of the true cavity region can be represented by the geometric centroid of the true cavity region. The preset quantity can be a number pre-set according to the actual situation, such as 360. The number of elements in the pressure source node set can be the same as the number of elements in the bearing end node set.

[0048] It should be noted that in real physiological environments, the pulsating pressure difference in the vessel wall tends to radiate outwards from the true lumen. To simulate this radial impact, it is often necessary to establish a one-to-one spatial correspondence between the pressure inlets on the inner membrane side and the pressure outlets on the outer membrane side to ensure that the permeation path calculated subsequently conforms to the geometric characteristics of pressure conduction. Therefore, a set of pressure source nodes representing the set of pressure inlets on the inner membrane side and a set of bearing end nodes representing the set of pressure outlets on the outer membrane side can be selected to facilitate subsequent calculations.

[0049] As an example, determining the set of pressure source nodes and the set of bearing end nodes may include the following steps: The first step is to draw a predetermined number of rays in different directions, with the center point of the true cavity region as the endpoint, and denoted as the initial rays.

[0050] For example, taking the center point of the true cavity region as the endpoint, 360 rays with different directions can be drawn, and the angle between the directions of adjacent rays can be 1°, and each ray obtained is recorded as the initial ray.

[0051] The second step is to identify the initial ray as the target ray if it intersects with both the intima and the outer wall of the blood vessel.

[0052] It should be noted that if the initial ray does not simultaneously intersect with both the intima and the outer wall of the blood vessel, i.e., fails to penetrate both the intima and the outer wall at the same time, then the ray can be considered invalid and not recorded as a target ray.

[0053] The third step, screening out the pressure source nodes and bearing end nodes from each target ray, may include the following sub-steps: The first sub-step involves identifying any target ray as the marker ray and identifying the center point of the true cavity region as the marker reference point.

[0054] The second sub-step involves selecting the intersection point closest to the aforementioned reference point from all intersection points between the marked ray and the inner membrane, and denoting it as the pressure source node.

[0055] Among them, the pressure source node can be located on the boundary line between the true and false cavities (inner diaphragm), that is, the innermost boundary of the false cavity.

[0056] The third sub-step involves selecting the intersection point farthest from the aforementioned reference point from all intersection points between the marked rays and the outer wall of the blood vessel, and denoting it as the bearing end node.

[0057] Among them, the receiving end node can be located on the outer boundary of the false lumen (the outer wall of the blood vessel), that is, the outermost boundary of the false lumen.

[0058] The fourth step is to form a pressure source node set by combining all pressure source nodes and a bearing end node set by combining all bearing end nodes.

[0059] The local fluid conduction resistance determination module 104 is used to determine the local fluid conduction resistance between each two pixels in the false lumen region based on the distance between each two pixels in the false lumen region, as well as the thrombus density retardation value and grayscale gradient modulus corresponding to these two pixels.

[0060] The gray-level gradient magnitude, also known as the gray-level gradient value, can be obtained by: performing Gaussian smoothing on the target CT cross-sectional image to suppress imaging noise, such as using a 3×3 Gaussian kernel with a standard deviation σ=0.5; then, using the Sobel operator or Canny operator to calculate the gray-level gradient magnitude of each pixel in the image.

[0061] As an example, determining the local fluid conduction resistance between every two pixels within the pseudo-cavity region may include the following steps: The first step is to determine the material density resistance weight between each pair of pixels in the false lumen region based on the distance between each pair of pixels in the false lumen region and the mean value between the thrombus density resistance values ​​corresponding to each pair of pixels in the false lumen region.

[0062] The distance between two pixels can be represented by the Euclidean distance between them, and orthogonal adjacent pixels can be assigned a value of 1, while diagonally adjacent pixels can be assigned a value of 1. Top, bottom, left, and right are orthogonally adjacent. Top left, bottom left, top right, and bottom right are diagonally adjacent.

[0063] For example, the formula for determining the material density drag weight between two pixels within the pseudo-cavity region can be: ; in, It represents the material density drag weight between the i-th and j-th pixels within the pseudo-cavity region. i and j are the indices of the different pixels within the pseudo-cavity region. It is the thrombus density block value corresponding to the i-th pixel in the false cavity region. It is the thrombus density block value corresponding to the j-th pixel point within the false cavity region. It is the distance between the i-th pixel and the j-th pixel within the false cavity region.

[0064] It should be noted that, It often only reflects the static resistance of material porosity to fluid flow, and often does not include the effect of structural stratification.

[0065] The second step is to determine the relative gradient intensity of each pixel in the pseudo-cavity region based on the gray-level gradient magnitude of each pixel in the pseudo-cavity region and the maximum value of the gray-level gradient magnitude of all pixels in the pseudo-cavity region.

[0066] For example, the formula for determining the relative gradient intensity of a pixel within the pseudo-cavity region can be: ; in, It represents the relative gradient intensity corresponding to the i-th pixel within the pseudo-cavity region. i is the index of the pixel within the pseudo-cavity region. It is the grayscale gradient magnitude corresponding to the i-th pixel within the pseudo-cavity region. It is the maximum value of the grayscale gradient magnitude corresponding to all pixels within the pseudo-cavity region. It is an adjustment factor that is preset according to the actual situation, mainly used to prevent the denominator from being 0. For example, it can be 0.001.

[0067] It should be noted that, It can characterize the relative gradient intensity obtained by normalizing the gradient magnitude in order to eliminate the influence of scanning parameters from different devices.

[0068] The third step is to determine the layered interface penetration coefficient corresponding to each pixel in the false cavity region based on the relative gradient intensity of each pixel in the false cavity region.

[0069] For example, the formula for determining the layered interface permeability coefficient corresponding to a pixel within the pseudo-cavity region can be: ; in, It is the layered interface penetration coefficient corresponding to the i-th pixel within the false cavity region. i is the index of the pixel within the false cavity region. It is an exponential function with the natural constant as its base. It is an interface sensitivity coefficient preset according to the actual situation, used to control the penalty intensity for weak connection structures. Its value range can be set from 1.5 to 2.5. In this embodiment... You can choose 2. It is the relative gradient intensity corresponding to the i-th pixel within the pseudo-cavity region.

[0070] It should be noted that the thrombus layering interface within the false lumen appears on imaging as a dramatic change in local grayscale; these high-gradient regions correspond to physically weak structural surfaces. To simulate the slippage effect of fluids at these weak surfaces, it is often necessary to calculate a correction factor that reduces the underlying drag, i.e., the layering interface permeability coefficient. When a pixel is inside a homogeneous thrombus, the gradient... Approaching 0, coefficient Approaching 1, meaning the resistance is not discounted; when a pixel is at a layered interface, the gradient... Larger, coefficient It is significantly less than 1, thereby reducing the fluid conduction resistance at that location.

[0071] The fourth step is to determine the local fluid conduction resistance between each pair of pixels in the pseudo-cavity region based on the material density resistance weight between each pair of pixels in the pseudo-cavity region and the average value of the layered interface permeability coefficient corresponding to each pair of pixels in the pseudo-cavity region.

[0072] For example, the formula for determining the local fluid conduction resistance between two pixels within the pseudo-cavity region can be: ; in, It represents the local fluid conduction resistance between the i-th and j-th pixels within the pseudo-cavity region. i and j are the indices of the different pixels within the pseudo-cavity region. It is the material density drag weight between the i-th pixel and the j-th pixel within the pseudo-cavity region. It is the layered interface penetration coefficient corresponding to the i-th pixel within the false cavity region. It is the layered interface penetration coefficient corresponding to the j-th pixel point within the false cavity region.

[0073] It should be noted that real fluid permeation pathways tend to propagate along composite channels with the lowest material density and weakest structural bonding. Therefore, material density properties can be combined with interfacial structural properties to generate the final local fluid conduction resistance. Even if a thrombus has a high density at a certain location, i.e. Large, but if it is located at the layered interface, i.e. and Small, its final local transmission resistance It will still be significantly lowered, thus simulating the pathological characteristics of "dense but layered" thrombi that are prone to leakage.

[0074] The radial permeation resistance matrix determination module 105 is used to determine the radial permeation resistance matrix based on the local fluid conduction resistance between different pixels on the path formed between the pixels in the pressure source node set and the pixels in the bearing end node set.

[0075] As an example, determining the total radial permeability resistance matrix may include the following steps: The first step is to adaptively connect each pixel in the false cavity region and its preset neighboring pixels to obtain the undirected connection edge corresponding to each pixel in the false cavity region, based on whether each pixel in the false cavity region and its preset neighboring pixels belong to the false cavity region.

[0076] The preset neighborhood can be a pre-set neighborhood, which can be an 8-neighborhood.

[0077] For example, any pixel within the false cavity region can be designated as a marker pixel. Pixels belonging to the false cavity region can be selected from the preset neighborhood corresponding to the marker pixel and used as temporary pixels. The marker pixel and the temporary pixel can be connected, and the undirected edge between the marker pixel and the temporary pixel can be recorded as the undirected connection edge corresponding to the marker pixel.

[0078] The second step is to construct a topological network by using all pixels within the false cavity region as nodes and all their corresponding undirected edges as nodes.

[0079] It should be noted that the movement of fluid in a pseudo-cavity porous medium is often a non-jumping, continuous process, and often needs to be simulated within a space strictly constrained by anatomical boundaries. To prevent the calculated permeation path from passing through true cavities or non-pseudo-cavity regions such as the spine, a topological network can be constructed.

[0080] The third step involves using the local fluid conduction resistance between any two connected nodes in the topology network as the path length between these two nodes. Using Dijkstra's algorithm, the total path length corresponding to the shortest connected path in the topology network for each pixel in the pressure source node set and each pixel in the bearing end node set is obtained and denoted as the total radial permeation resistance between each pixel in the pressure source node set and each pixel in the bearing end node set.

[0081] For example, any pixel in the set of pressure source nodes can be designated as the marked pressure source point, and any pixel in the set of bearing end nodes can be designated as the marked bearing end point. The total radial permeation resistance between the marked pressure source point and the marked bearing end point can be equal to the sum of the local fluid conduction resistances between all adjacent nodes on the shortest connected path from the marked pressure source point to the marked bearing end point in the topology network. The shortest connected path can be the connected path with the smallest sum of local fluid conduction resistances among the connected paths between the marked pressure source point and the marked bearing end point.

[0082] It should be noted that if there is no connecting path between the marked pressure source point and the marked bearing endpoint, the total radial permeation resistance between the marked pressure source point and the marked bearing endpoint can be set to a preset maximum value. This preset maximum value can be a maximum value pre-set based on the actual situation; for example, it could be... .

[0083] The fourth step is to construct a radial permeation resistance matrix by combining the radial permeation resistance between all pixels in the pressure source node set and all pixels in the bearing end node set.

[0084] In this matrix, a single row of elements can be composed of the total radial permeation resistance between a single pixel in the pressure source node set and all pixels in the bearing end node set. Similarly, a single column of elements in the total radial permeation resistance matrix can be composed of the total radial permeation resistance between a single pixel in the bearing end node set and all pixels in the pressure source node set.

[0085] It should be noted that the total radial permeation resistance matrix can include all possible point-to-point transmission costs from the inner membrane pressure inlet to the outer membrane pressure outlet.

[0086] The remote remodeling risk prediction module 106 is used to assist in predicting the remote remodeling risk based on the radial permeation resistance matrix and the pipe wall pulsation pressure difference.

[0087] As an example, the following steps may be included to assist in predicting the risk of distal remodeling based on the total radial permeability resistance matrix and the wall pulsation pressure difference: The first step is to select, based on the total radial permeation resistance matrix, the bearing end node that matches each pixel in the pressure source node set from the bearing end node set, and use it as the bearing end matching point corresponding to each pixel in the pressure source node set.

[0088] For example, if the radial permeation resistance matrix is ​​considered as the weight matrix of a bipartite graph, a minimum weight perfect matching algorithm, such as the KM algorithm or the Hungarian algorithm, can be applied to find a one-to-one mapping relationship between nodes, such that every pixel in the pressure source node set is matched with a unique pixel in the bearing end node set, and the sum of the resistance of all matching paths is minimized. At this time, the bearing end node that matches each pixel in the pressure source node set can be denoted as the bearing end matching point corresponding to each pixel in the pressure source node set.

[0089] The second step, determining the overall barrier capacity value of the pseudo-cavity based on the radial permeation resistance between the pixels in the pressure source node set and their corresponding bearing end matching points, may include the following sub-steps: In the first sub-step, if the radial permeation resistance between a pixel in the pressure source node set and its corresponding bearing end matching point is equal to a preset maximum value, then the pixel in the pressure source node set is determined as an invalid pixel.

[0090] It should be noted that invalid pixels are often pixels that are not connected to the receiving end node, which often indicates that the corresponding pseudo cavity wall may have been calcified or completely blocked, and there is no risk of fluid conduction.

[0091] The second sub-step involves identifying all pixels in the pressure source node set except for invalid pixels as valid pixels. If the number of valid pixels is a constant of 0, then the overall barrier capability value of the false cavity is set to a preset maximum value.

[0092] It should be noted that effective pixels often represent pixels at the site of soft thrombi. When the number of effective pixels is a constant of 0, circumferential calcification is likely, and the risk index for distal expansion is close to 0. Therefore, the overall barrier capacity value of the false lumen can be set to a preset maximum value.

[0093] In the third sub-step, if the number of effective pixels is greater than the constant 0, the overall barrier capacity value of the false cavity is determined based on the number of effective pixels and the average of the radial permeation resistance between all effective pixels and their corresponding bearing end matching points.

[0094] For example, the formula for determining the overall barrier capacity value of the pseudo-cavity can be: ; Where E is the overall barrier capability value of the pseudo-cavity. N is the number of valid pixels in the pressure source node set. b is the index of the valid pixel in the pressure source node set. It is the total radial permeation resistance between the b-th effective pixel in the pressure source node set and its corresponding bearing end matching point.

[0095] It should be noted that the false lumen patch after aortic dissection is subjected to uniformly distributed blood pressure throughout the entire circumference under physiological conditions. Fluid does not seek to penetrate a single point, but rather tends to exert tension on the vessel wall using all possible low-resistance channels. Therefore, assessing the safety of the false lumen wall cannot rely solely on a single shortest path, but often requires evaluating its comprehensive ability to resist circumferential synergistic infiltration. A larger N value often indicates a potentially larger soft thrombus area, meaning a potentially larger low-resistance infiltration network. It can characterize the average resistance. E can characterize the effective fluid resistance that the false lumen thrombus layer as a whole structure can provide on average per unit area under the most unfavorable fluid conduction mode; the lower the value, the more likely there is a widely connected low-resistance seepage network in the false lumen, and the worse the overall defense capability.

[0096] The third step is to determine the distal expansion risk index based on the overall barrier capacity value of the false lumen and the pressure difference of the vessel wall pulsation.

[0097] It should be noted that the failure of distal remodeling of the false lumen, i.e., diameter expansion or rupture, is essentially a "load-strength" interference process. When the mechanical driving force exerted by the blood in the true lumen exceeds the structural barrier capacity of the false lumen wall, the outer wall of the false lumen will continuously bear tension and undergo plastic deformation.

[0098] For example, the formula for determining the risk index of remote expansion can be: ; F represents the risk index for remote expansion. It is the pressure difference due to pulsation in the vessel wall. E is the overall barrier capacity value of the false lumen. It is an adjustment factor that is preset according to the actual situation, mainly used to prevent the denominator from being 0. For example, it can be 0.00001.

[0099] It should be noted that F intuitively quantifies the ratio of the current hemodynamic load to the defensive capacity of the false lumen wall: when F is low, it often indicates that the organized thrombus in the false lumen is dense and structurally intact (large E), which is often sufficient to resist current blood pressure fluctuations. Fluid has difficulty penetrating to the outer wall, which often indicates a good remodeling outcome; when F is high, it often indicates either poor blood pressure control in the patient ( A large false lumen (F) or severe interlaminar fissures within the thrombus (Small F) often allow fluid pressure to be transmitted to the outer wall of the blood vessel with almost no loss, indicating a high risk of continued expansion or rupture of the false lumen. Therefore, doctors can use F to determine whether to strengthen antihypertensive treatment or perform secondary surgical intervention.

[0100] The fourth step, using the remote expansion risk index to assist in predicting remote restructuring risks, may include the following sub-steps: In the first sub-step, if the remote expansion risk index is less than or equal to the preset risk threshold, the remote remodeling is deemed to be good.

[0101] The preset risk threshold can be a threshold set in advance based on the actual situation. It can be determined by obtaining the distal expansion risk index of patients with good and failed distal remodeling in the historical sample library and training the optimal classification boundary value using machine learning classification algorithms such as support vector machine or logistic regression. For example, it can be 1.5.

[0102] The second sub-step is that if the distal expansion risk index is greater than the preset risk threshold, it indicates that the false cavity has a high risk of continuous expansion or rupture.

[0103] refer to Figure 2 Based on the same inventive concept as the above-described method embodiments, this invention provides a method for predicting the risk of distal remodeling after aortic dissection using multimodal data fusion, comprising the following steps: Step S1: Obtain the target CT cross-sectional image and the pulsation pressure difference of the aortic wall after aortic dissection surgery, and identify the true lumen region and the false lumen region from the target CT cross-sectional image.

[0104] Step S2: Determine the thrombus density block value corresponding to each pixel in the false lumen region based on the CT value corresponding to each pixel in the false lumen region.

[0105] Step S3: Based on the distribution of the intersection points between the preset number of rays constructed from the center point of the true cavity region and the false cavity region, determine the pressure source node set and the bearing end node set.

[0106] Step S4: Determine the local fluid conduction resistance between every two pixels in the false lumen region based on the distance between every two pixels in the false lumen region, as well as the thrombus density retardation value and grayscale gradient modulus corresponding to these two pixels.

[0107] Step S5: Determine the total radial permeation resistance matrix based on the local fluid conduction resistance between different pixels on the path formed between the pixels in the pressure source node set and the pixels in the bearing end node set.

[0108] Step S6: Based on the radial permeation resistance matrix and the pipe wall pulsation pressure difference, assist in predicting the risk of remote remodeling.

[0109] Figure 3 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. For example, as shown... Figure 3 As shown, the computer device 300 includes: a memory 301, a processor 302, and a computer program 303 stored in the memory 301 and running on the processor 302. When the processor 302 executes the computer program 303, the computer device can execute the aforementioned method for predicting the risk of distal remodeling after aortic dissection using multimodal data fusion.

[0110] Based on the same inventive concept as the above-described method embodiments, the present invention provides a server, including a memory and a processor. The memory is used to store executable program code, and the processor is used to call and run the executable program code from the memory, causing the device to execute the above-described method for predicting the risk of distal remodeling after aortic dissection using multimodal data fusion.

[0111] Based on the same inventive concept as the above-described method embodiments, the present invention provides a computer program product comprising: computer program code, which, when run on a computer, causes the computer to execute the above-described method for predicting the risk of distal remodeling after aortic dissection using multimodal data fusion.

[0112] Based on the same inventive concept as the above-described method embodiments, the present invention provides a computer-readable storage medium storing computer program code, which, when executed on a computer, causes the computer to perform the above-described method for predicting the risk of distal remodeling after aortic dissection using multimodal data fusion.

[0113] In summary, this invention quantifies multiple indicators related to fluid osmotic resistance, such as thrombus density retardation value and local fluid conduction resistance, by analyzing the grayscale and gradient of CT cross-sectional images. This quantifies the total radial osmotic resistance matrix and, combined with the vessel wall pulsation pressure difference, assists in predicting the risk of distal remodeling. This enables auxiliary prediction of the risk of distal remodeling after aortic dissection and improves the rationality of the auxiliary prediction of the risk of distal remodeling after aortic dissection, thereby enhancing the auxiliary effect.

[0114] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A multimodal data fusion model for predicting the risk of distal remodeling after aortic dissection surgery, characterized in that, The model includes: The data acquisition and region identification module is used to acquire the target CT cross-sectional image and the pulsation pressure difference of the vessel wall of the target patient after aortic dissection surgery, and to identify the true lumen region and the false lumen region from the target CT cross-sectional image. The thrombus density block value determination module is used to determine the thrombus density block value corresponding to each pixel in the false lumen region based on the CT value corresponding to each pixel in the false lumen region. The intersection set determination module is used to determine the pressure source node set and the bearing end node set based on the distribution of intersection points between a preset number of rays constructed from the center point of the true cavity region and the false cavity region. The local fluid conduction resistance determination module is used to determine the local fluid conduction resistance between each two pixels in the false lumen region based on the distance between each two pixels in the false lumen region, as well as the thrombus density retardation value and grayscale gradient modulus corresponding to these two pixels. The radial permeation resistance matrix determination module is used to determine the radial permeation resistance matrix based on the local fluid conduction resistance between different pixels on the path formed between the pixels in the pressure source node set and the pixels in the bearing end node set. The remote remodeling risk prediction module is used to assist in predicting the risk of remote remodeling based on the radial permeation resistance matrix and the pipe wall pulsation pressure difference.

2. The multimodal data fusion-based distal remodeling risk prediction model after aortic dissection surgery according to claim 1, characterized in that, The step of determining the thrombus density block value corresponding to each pixel in the false lumen region based on the CT value of each pixel in the false lumen region includes: The mean CT value of all pixels in the true cavity region is determined as the reference baseline for the true cavity blood pool. Based on the difference between the true cavity blood pool reference benchmark and the preset solid tissue benchmark, and the difference between the CT value corresponding to each pixel in the false cavity region and the preset solid tissue benchmark, the density reference factor corresponding to each pixel in the false cavity region is determined. Based on the density reference factor corresponding to each pixel in the false cavity region, determine the relative density factor corresponding to each pixel in the false cavity region; The thrombus density block value corresponding to each pixel in the false lumen region is determined by the square of the relative density factor corresponding to each pixel in the false lumen region.

3. The multimodal data fusion-based distal remodeling risk prediction model after aortic dissection surgery according to claim 1, characterized in that, The step of determining the relative density factor corresponding to each pixel in the pseudo-cavity region based on the density reference factor corresponding to each pixel in the pseudo-cavity region includes: Any pixel within the false cavity region is designated as the marked pixel. If the density reference factor corresponding to the marked pixel is greater than a constant 1, then the relative density factor corresponding to the marked pixel is set to a constant 1. If the density reference factor corresponding to the marked pixel is less than a constant 0, then the relative density factor corresponding to the marked pixel is set to a constant 0. If the density reference factor corresponding to the marked pixel is greater than or equal to a constant 0 and less than or equal to a constant 1, then the relative density factor corresponding to the marked pixel is set as its corresponding density reference factor.

4. The multimodal data fusion-based distal remodeling risk prediction model after aortic dissection surgery according to claim 1, characterized in that, The determination of the pressure source node set and the bearing end node set based on the distribution of intersection points between a predetermined number of rays constructed according to the center point of the true cavity region and the false cavity region includes: With the center point of the true cavity region as the endpoint, draw a predetermined number of rays in different directions, which are denoted as the initial rays; If the initial ray intersects with both the intima and the outer wall of the blood vessel, then the initial ray is identified as the target ray. The pressure source nodes and bearing end nodes are selected from each target ray; All pressure source nodes are combined into a pressure source node set, and all bearing end nodes are combined into a bearing end node set.

5. The multimodal data fusion-based distal remodeling risk prediction model after aortic dissection surgery according to claim 4, characterized in that, The process of screening out pressure source nodes and bearing end nodes from each target ray includes: Any target ray is designated as the marker ray, and the center point of the true cavity region is designated as the marker reference point; The intersection point closest to the marked reference point is selected from all intersection points of the marked ray and the inner diaphragm and denoted as the pressure source node; From all the intersections between the marked ray and the outer wall of the blood vessel, select the intersection point that is farthest from the marked reference point and denote it as the bearing end node.

6. The multimodal data fusion-based distal remodeling risk prediction model after aortic dissection surgery according to claim 1, characterized in that, The determination of the local fluid conduction resistance between every two pixels within the false lumen region, based on the distance between each pair of pixels and the corresponding thrombus density resistance value and grayscale gradient modulus of those two pixels, includes: The material density resistance weight between each two pixels in the false lumen region is determined based on the distance between each two pixels in the false lumen region and the mean value between the thrombus density resistance values ​​corresponding to each two pixels in the false lumen region. The relative gradient intensity of each pixel in the false cavity region is determined based on the gray-level gradient magnitude of each pixel in the false cavity region and the maximum value of the gray-level gradient magnitude of all pixels in the false cavity region. The layered interface penetration coefficient corresponding to each pixel in the false cavity region is determined based on the relative gradient intensity of each pixel in the false cavity region. The local fluid conduction resistance between each pair of pixels in the false cavity region is determined based on the material density resistance weight between each pair of pixels in the false cavity region and the average value of the layered interface permeability coefficient corresponding to each pair of pixels in the false cavity region.

7. The multimodal data fusion-based distal remodeling risk prediction model after aortic dissection surgery according to claim 1, characterized in that, The determination of the total radial permeation resistance matrix based on the local fluid conduction resistance between different pixels on the path formed by the pixels in the pressure source node set and the pixels in the bearing end node set includes: Based on whether each pixel in the false cavity region and its preset neighboring pixels belong to the false cavity region, adaptive connections are made between each pixel in the false cavity region and its preset neighboring pixels to obtain the undirected connection edges corresponding to each pixel in the false cavity region. A topological network is constructed by taking all pixels within the false cavity region as nodes and all their corresponding undirected edges as nodes. Using the local fluid conduction resistance between every two connected nodes in the topology network as the path length between these two nodes, the Dijkstra algorithm is used to obtain the total path length corresponding to the shortest connected path in the topology network for each pixel in the pressure source node set and each pixel in the bearing end node set, which is denoted as the total radial permeation resistance between each pixel in the pressure source node set and each pixel in the bearing end node set. The total radial permeation resistance matrix is ​​formed by the total radial permeation resistance between all pixels in the pressure source node set and all pixels in the bearing end node set.

8. The multimodal data fusion-based distal remodeling risk prediction model after aortic dissection surgery according to claim 7, characterized in that, The method of predicting distal remodeling risk based on the radial permeability resistance matrix and the pipe wall pulsation pressure difference includes: Based on the radial permeation resistance matrix, the bearing end nodes that match each pixel in the pressure source node set are selected from the bearing end node set and used as the bearing end matching points corresponding to each pixel in the pressure source node set. The overall barrier capacity value of the false cavity is determined based on the radial permeation resistance between the pixels in the pressure source node set and their corresponding bearing end matching points. The distal expansion risk index is determined based on the overall barrier capacity value of the false lumen and the pulsating pressure difference of the vessel wall; The risk of remote restructuring is predicted using the remote expansion risk index.

9. The multimodal data fusion-based distal remodeling risk prediction model after aortic dissection surgery according to claim 8, characterized in that, The step of determining the overall barrier capacity value of the pseudo-cavity based on the total radial permeation resistance between pixels in the pressure source node set and their corresponding bearing end matching points includes: If the radial permeation resistance between a pixel in the pressure source node set and its corresponding bearing end matching point is equal to a preset maximum value, then the pixel in the pressure source node set is determined to be an invalid pixel. Pixels in the pressure source node set other than invalid pixels are identified as valid pixels. If the number of valid pixels is a constant of 0, the overall barrier capability value of the false cavity is set to a preset maximum value. If the number of effective pixels is greater than the constant 0, the overall barrier capacity value of the false cavity is determined based on the number of effective pixels and the average of the radial permeation resistance between all effective pixels and their corresponding bearing end matching points.

10. The multimodal data fusion-based distal remodeling risk prediction model after aortic dissection surgery according to claim 8, characterized in that, The method of predicting remote reshaping risk based on the remote expansion risk index includes: If the remote expansion risk index is less than or equal to the preset risk threshold, the remote remodeling is deemed to be good. If the distal expansion risk index is greater than the preset risk threshold, it indicates that the false cavity has a high risk of continuous expansion or rupture.

Images