Cerebrovascular aging index determination method and device, medium and product

Abstract

The invention discloses a cerebrovascular aging index determination method and device, a medium and a product, and relates to the field of image data processing, and the method comprises the steps: employing Voronoi covariance to extract a central axis of a cerebrovascular segmentation image, and employing a lambda-maximum section tangent method to determine a tangent and a normal plane corresponding to each voxel on the central axis; fitting a maximum circle in a corresponding normal plane for each voxel, and determining a corresponding diameter; performing three-dimensional reconstruction on the tubular tree-shaped structure by combining the central axis and the diameter, extracting multi-level standardized imaging features based on a three-dimensional reconstructed image, and inputting the multi-level standardized imaging features into a cerebrovascular biological age prediction model to obtain a cerebrovascular biological age prediction result; and calculating a cerebrovascular aging index in combination with a cerebrovascular biological age prediction result and the actual age of the user. According to the invention, quantitative and accurate evaluation of cerebrovascular aging can be realized.

Description

Technical Field

[0001] The present application relates to the field of image data processing, and in particular to a method, device, medium and product for determining a cerebrovascular aging index. Background Art

[0002] Due to the special nature of intracranial blood vessels being wrapped by the skull, many methods used to assess the elasticity and stiffness of blood vessels in peripheral blood vessels cannot be implemented in the brain. Compared with extracranial blood vessels, the structure of intracranial blood vessels is very special, such as thin outer membrane, no external elastic layer and nutrient vessels. In addition, the characteristics of intracranial atherosclerotic plaques are different from those of neck blood vessels. The nature of intracranial hemorrhage and calcified plaques is lower than that of extracranial blood vessels.

[0003] Abnormalities of the cerebral vascular wall (such as thickening, sclerosis, and plaque formation) are important risk factors for stroke, but conventional imaging cannot quantitatively analyze the characteristics of the intracranial vascular wall and lumen. There is no imaging method for evaluating the intracranial cavity that can comprehensively reflect the characteristics of the intracranial vascular lumen structure and vascular wall from multiple perspectives, and thus cannot achieve quantitative and accurate assessment of cerebrovascular aging. Summary of the invention

[0004] The purpose of this application is to provide a method, device, medium and product for determining a cerebrovascular aging index, which can achieve quantitative and accurate evaluation of cerebrovascular aging.

[0005] To achieve the above objectives, this application provides the following solutions: In a first aspect, the present application provides a method for determining a cerebrovascular aging index, comprising: Acquiring a magnetic resonance image of the brain of the user to be determined; Obtaining a cerebral vascular segmentation image based on a magnetic resonance image; The Voronoi covariance is used to extract the central axis of the cerebral vascular segmentation image; The tangent corresponding to each voxel on the central axis is determined by using the λ-maximum segment tangent method, and the normal plane on each voxel is determined based on the tangent; For each voxel on the central axis, fit the largest circle in the corresponding normal plane and determine the diameter of the circle corresponding to the voxel; Combining the central axis and the diameter, the tubular tree structure in the cerebral blood vessel segmentation image is reconstructed in three dimensions to obtain a cerebral blood vessel segmentation three-dimensional image; Extract multi-level standardized imaging features based on 3D images of cerebral vascular segmentation; Get the age of the user to be determined; Constructing a cerebrovascular biological age prediction model; the cerebrovascular biological age prediction model is a neural network model constructed based on a generalized linear model; Input the multi-level standardized imaging features into the cerebrovascular biological age prediction model to obtain the cerebrovascular biological age prediction results; The cerebrovascular aging index was obtained based on the cerebrovascular biological age prediction results and age.

[0006] Optionally, obtaining a cerebral blood vessel segmentation image based on the magnetic resonance image includes: Preprocessing the magnetic resonance image to obtain a preprocessed image; Constructing a cerebrovascular tissue segmentation model; the cerebrovascular tissue segmentation model is a network model constructed and trained by introducing an attention gating module into the skip connection of a 3D-UNet convolutional neural network; Inputting the preprocessed image into the cerebrovascular tissue segmentation model to obtain a segmented image; The segmented image is post-processed to obtain the cerebral blood vessel segmented image.

[0007] Optionally, preprocessing the magnetic resonance image to obtain a preprocessed image includes: Performing brain tissue-specific identification and segmentation on the magnetic resonance image to obtain a brain tissue structure image; By using the linear rigid body transformation and spline interpolation method, the brain tissue structure image is transformed to the same anatomical coordinates to obtain the transformed image; Performing image denoising processing on the converted image using an adaptive non-local means algorithm to obtain a denoised image; Using a signal intensity remapping method after Gaussian deconvolution, magnetic resonance image artifacts caused by B0 field inhomogeneity in the denoised image are removed to obtain an intermediate image; Processing the intermediate image using a contrast-limited adaptive histogram equalization method to obtain a first signal-enhanced image; Using the Sobel convolution method, identifying edge positions where a sudden change in signal intensity occurs in the first signal-enhanced image, and obtaining a second signal-enhanced image based on the identified edge positions; The second signal enhanced image is processed by a gamma correction method to obtain the preprocessed image.

[0008] Optionally, post-processing the segmented image to obtain the cerebral blood vessel segmented image includes: A three-dimensional conditional random field model is used to perform penalty processing on adjacent points with similar signal strengths on the segmented image to obtain a three-dimensional space segmented image; the adjacent points with similar signal strengths refer to pixel points whose signal strength difference is within a set range; The three-dimensional maximum connected component is found on the three-dimensional space segmentation image, and the unconnected signal noise in the three-dimensional space segmentation image is removed based on the three-dimensional maximum connected component to obtain the cerebral blood vessel segmentation image.

[0009] Optionally, the process of extracting the basic features of the cerebral blood vessels based on the cerebral blood vessel segmentation three-dimensional image includes: Counting the number of voxels with positive markers in the cerebral blood vessel segmentation three-dimensional image, and obtaining the absolute length of the cerebral blood vessel based on the number of voxels; Acquiring the individual brain tissue volume, and obtaining the relative cerebral blood vessel length based on the absolute cerebral blood vessel length and the individual brain tissue volume, and using the relative cerebral blood vessel length as the cerebral blood vessel length; Determine the voxel volume based on the image resolution, and obtain the absolute volume of the cerebral blood vessels in the cerebral blood vessel segmentation three-dimensional image based on the statistical number of the voxels; A relative cerebral vascular volume is obtained based on the absolute cerebral vascular volume and the individual brain tissue volume, and the relative cerebral vascular volume is used as the volume of the cerebral blood vessels.

[0010] Optionally, the process of extracting cerebral blood vessel geometric features based on the cerebral blood vessel segmentation three-dimensional image includes: In the three-dimensional image of cerebral vascular segmentation, the curvature radius and bending degree of each voxel on the central axis in the three-dimensional space are determined according to the tangent vector flip angle and the Frenet frame; Determine the average vascular curvature radius and the average vascular tortuosity in each brain region of the cerebral vascular segmentation three-dimensional image; The cerebral vascular curvature radius is obtained based on the curvature radius of each voxel in three-dimensional space and the average vascular curvature radius of the brain region; The curvature of cerebral blood vessels is obtained based on the curvature of each voxel in three-dimensional space and the average curvature of blood vessels in the brain region.

[0011] Optionally, the process of extracting cerebral vascular topological features based on the cerebral vascular segmentation three-dimensional image includes: Based on the cerebral vascular segmentation three-dimensional image, a tree structure is used to characterize the adjacency relationship of each voxel on the central axis, bifurcation points are obtained, and the number of bifurcations and the spatial distribution of bifurcation points are counted; Determine the angle of the tangent vectors of the adjacent voxels at the bifurcation point; Based on the number of bifurcations and the spatial distribution of bifurcation points, determining an average bifurcation angle and a cerebrovascular bifurcation density in each brain region of the cerebrovascular segmented three-dimensional image; Based on the angle between the tangent vectors of adjacent voxels and the average bifurcation angle, the overall bifurcation angle characteristics of cerebral vessels were determined; The process of extracting cerebrovascular morphological features based on the cerebrovascular segmentation three-dimensional image includes: Acquire the cerebrovascular segmentation images of healthy people, and iteratively register the cerebrovascular segmentation images of healthy people to obtain the cerebrovascular distribution density map of the group; Nonlinearly registering the cerebrovascular segmentation three-dimensional image with the population cerebrovascular distribution density map in a nonlinear registration manner, and obtaining a transformed Jacobian matrix; The cerebral blood vessel segmentation three-dimensional image after the Jacobian matrix standardization registration is used to obtain the distribution density of the cerebral blood vessels in the standard atlas space.

[0012] In a second aspect, the present application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-mentioned method for determining a cerebrovascular aging index.

[0013] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-mentioned method for determining a cerebrovascular aging index.

[0014] In a fourth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the above-mentioned method for determining a cerebrovascular aging index.

[0015] According to the specific embodiments provided in this application, this application has the following technical effects: The present application provides a method, device, medium and product for determining a cerebrovascular aging index, which predicts the biological age of cerebrovascular vessels by integrating a multi-feature fusion index based on cerebrovascular vessels (i.e., multi-level standardized imaging features including basic cerebrovascular features, cerebrovascular geometric features, cerebrovascular topological features and cerebrovascular morphological features), which is a qualitative improvement over the cerebrovascular aging form that only reflects one aspect. By defining the output of the model as the biological age of cerebrovascular vessels and comparing it with the actual physical age of the user to obtain the cerebrovascular aging index, the degree of cerebrovascular aging in the user's imaging can be quantitatively and objectively measured, thereby achieving an accurate assessment of cerebrovascular aging. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 A schematic diagram of a flow chart of a method for determining a cerebrovascular aging index provided in one embodiment of the present application; Figure 2 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0019] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0020] In an exemplary embodiment, the present application provides a method for determining a cerebrovascular aging index, which is executed by a computer device, specifically, a terminal or a server, or a terminal and a server. In the present application, the method is applied to a server as an example for illustration. Figure 1 As shown, the method includes: Step 100: Obtain a magnetic resonance image of the brain of the user to be determined.

[0021] Step 101: Obtain a cerebral blood vessel segmentation image based on a magnetic resonance image.

[0022] Step 102: Use Voronoi covariance to extract the central axis of the cerebrovascular segmentation image. For example, the centerline of the cerebrovascular segmentation image is first estimated, and the centerline of the cerebrovascular segmentation is stably extracted based on the Voronoi covariance measure (VCM), thereby reducing the center point recognition drift and voxel discontinuity problems at the cerebrovascular bifurcation.

[0023] Step 103, using the λ-maximal segment tangent method to determine the tangent corresponding to each voxel on the central axis, and based on the tangent, determine the normal plane on each voxel. For example, based on the extraction of the central axis of the cerebral blood vessels, the λ-maximal segment tangent method (λ-MST) is used to calculate the tangent corresponding to each voxel on the central axis, so as to further calculate the normal plane on each central axis voxel, and the square axis method of weighted average of the normal planes of adjacent voxels is used to obtain a robust estimate of the normal plane of each voxel on the central line.

[0024] Step 104: For each voxel on the central axis, fit a maximum circle in the corresponding normal plane, and determine the diameter of the circle corresponding to the voxel.

[0025] Step 105: Perform three-dimensional reconstruction of the tubular tree structure in the cerebral vascular segmentation image by combining the central axis and the diameter to obtain a three-dimensional cerebral vascular segmentation image. This step can effectively prevent the problem that local small errors that may be introduced by the structure segmentation algorithm are cascaded and amplified in the vascular feature estimation, and can ensure the robustness of the three-dimensional reconstruction.

[0026] Step 106, extracting multi-level standardized imaging features based on the cerebrovascular segmentation three-dimensional image. The multi-level standardized imaging features include basic cerebrovascular features, cerebrovascular geometric features, cerebrovascular topological features and cerebrovascular morphological features. Basic cerebrovascular features include: the length and volume of cerebrovascular and the diameter and length of local brain regions. Local brain regions are determined based on voxels. Cerebrovascular geometric features include: the radius of curvature of cerebrovascular and the degree of curvature of cerebrovascular. Cerebrovascular topological features include: the angle of cerebrovascular bifurcation points and the density of cerebrovascular bifurcation. Cerebrovascular morphological features include the distribution density of cerebrovascular in the standard atlas space.

[0027] Step 107: Obtain the age of the user to be determined.

[0028] Step 108: construct a cerebrovascular biological age prediction model. The cerebrovascular biological age prediction model is a neural network model constructed based on a generalized linear model.

[0029] Step 109: Input the multi-level standardized imaging features into the cerebrovascular biological age prediction model to obtain the cerebrovascular biological age prediction result.

[0030] Step 110: Obtain a cerebrovascular aging index based on the cerebrovascular biological age prediction result and age.

[0031] In another exemplary embodiment of the present application, taking the acquired magnetic resonance image as a TOF-MRA (Time-of-Flight Magnetic Resonance Angiography) magnetic resonance image as an example, the implementation process of the above step 101 is described, including: 1. Preprocess the TOF-MRA magnetic resonance image to obtain a preprocessed image, for example: 11. Perform brain tissue-specific identification and segmentation on the TOF-MRA magnetic resonance image, remove non-brain tissue structures such as the scalp, skull, facial features and neck, and obtain a brain tissue structure image. The purpose of this step is to eliminate the interference of non-brain tissue structure image signals on subsequent processing.

[0032] 12. Use linear rigid body transformation and spline interpolation to transform the brain tissue structure image to the same anatomical coordinates to obtain a transformed image. For example, for TOF-MRA magnetic resonance images, a linear rigid body transformation with 6 degrees of freedom and spline interpolation are used to align the T1-weighted imaging of the same individual and resample to the same image space. The purpose of this step is to transform each sequence image at different voxel coordinates to the same anatomical coordinates to facilitate the integration of cross-sequence signal information.

[0033] 13. Use the adaptive non-local means algorithm to perform image denoising on the converted image to obtain a denoised image to remove Gaussian noise that may exist in the TOF-MRA magnetic resonance image.

[0034] 14. The signal intensity remapping method after Gaussian deconvolution is used to remove the magnetic resonance image artifacts caused by B0 field inhomogeneity in the denoised image, and an intermediate image is obtained to remove the common artifact signals in TOF-MRA magnetic resonance images and improve the image signal-to-noise ratio.

[0035] 15. Use the limited contrast adaptive histogram equalization method to process the intermediate image to obtain a first signal enhanced image.

[0036] Due to the principle limitation of vascular imaging magnetic resonance sequences such as TOF-MRA, the signal enhancement amplitude of small blood vessels is significantly lower than that of large blood vessels, making it difficult to identify small blood vessels. In order to solve this problem, step 15 is performed so that cerebral blood vessel signals of different scales can obtain similar enhancement amplitudes.

[0037] 16. Due to the existence of partial volume effect, the edge signal of the blood vessel is significantly lower than the center signal of the blood vessel, and therefore is not easy to identify. To solve this problem, this embodiment uses the Sobel convolution method to identify the edge position where the signal intensity mutation occurs in the first signal enhanced image, achieve specific enhancement of the edge signal of the blood vessel, and improve the contrast between the blood vessel and the surrounding non-vascular tissue. A second signal enhanced image is obtained based on the identified edge position. This step allows the cerebral blood vessel signals at different distances from the center line to obtain similar enhancements.

[0038] 17. The second signal enhancement image is processed by the Gamma correction method to obtain a preprocessed image. The Gamma correction method nonlinearly increases the vascular tissue signal (high signal) and suppresses the non-vascular tissue signal (low signal) of the magnetic resonance image by adjusting the transformation coefficient Gamma. Among them, the transformation coefficient Gamma is used as a processing hyperparameter, and the relative amplitude of the enhancement and suppression of the signal can be controlled according to the nature of the processed image to obtain better image contrast.

[0039] 2. Construct a cerebrovascular tissue segmentation model. The cerebrovascular tissue segmentation model is a trained network model constructed by introducing an attention gating module into the skip connection of the 3D-UNet convolutional neural network.

[0040] In actual application, the specific implementation methods of constructing the cerebrovascular tissue segmentation model may include: 21. Build training and testing datasets.

[0041] Collect T1-weighted imaging and TOF-MRA paired data of multiple (for example, 100) healthy people acquired by different equipment using routine clinical scanning sequences. After preprocessing the image data according to steps 11 to 17 above, multiple (for example, two) neuroradiologists with rich experience in image annotation can independently annotate the cerebrovascular tissue signals in the magnetic resonance images, and at least one senior neuroradiologist can conduct a comprehensive examination and evaluation of the annotation results to complete the annotation of cerebrovascular tissue signals in the multimodal magnetic resonance image data, ensure the consistency and accuracy of the data annotation, and finally construct a training and testing dataset.

[0042] 22. Build a deep neural network architecture.

[0043] The basic structure of 3D-UNet and attention gating module is used to build a deep neural network model as the initial cerebrovascular tissue segmentation model. The overall 3D-UNet convolutional neural network adopts an adaptive design, which automatically adjusts hyperparameters such as batch size and network depth width according to the spatial resolution and size of the input image. At the same time, the attention gating module is introduced in the jump connection of the 3D-UNet convolutional neural network, and its basic form can be expressed as: .

[0044] In the formula, and Represent the feature tensors of the encoding module and decoding module connected by the jump, represents the RELU activation function, represents the Sigmoid activation function, , and represents the linear transformation coefficient, as well as represents the bias coefficient, represents the processed image, Represents the SAct image after resampling.

[0045] Finally, the resampled SAct image is combined with the feature tensor on the decoding module Multiplying together can fuse feature information at different scales and reduce background noise to a great extent.

[0046] 23. Training and testing of the initial cerebrovascular tissue segmentation model.

[0047] The initial cerebrovascular tissue segmentation model is trained and tested based on the training and testing data set constructed in step 21 above by using a five-fold cross-validation method. In order to cope with the differences in specificity and sensitivity of vascular segmentation results at different scales during vascular segmentation, the Tversky function is introduced as the loss function of the initial cerebrovascular tissue segmentation model. On the basis of the Tversky function, a weighted Tversky loss is further proposed, that is, the inverse of the Euclidean distance of the voxel from the center line is used as the mismatch loss weight, which strengthens the topological structure restrictions of the segmentation results, so that the vascular segmentation results have a more consistent performance at all scales. This algorithm selects the initial cerebrovascular tissue segmentation model with the best result as the final output model (i.e., the cerebrovascular tissue segmentation model) by default, and the user can also choose to deploy the average model.

[0048] 3. Input the preprocessed image into the cerebral vascular tissue segmentation model to obtain a segmented image.

[0049] 4. Post-process the segmented image to obtain a cerebral vascular segmentation image, for example: 41. In the post-processing of the segmented image, a three-dimensional conditional random field model is used to penalize adjacent points with similar signal strength on the segmented image to obtain a three-dimensional spatial segmented image. Adjacent points with similar signal strength refer to pixel points whose signal strength difference is within a set range.

[0050] The three-dimensional conditional random field model takes into account the continuity of vascular tissue in space. By establishing potential functions between pairs and on individual points and penalizing adjacent points with similar signal intensities on magnetic resonance images with different segmentation labels on the segmented images, a more continuous and reasonable segmentation image in three-dimensional space (i.e., a three-dimensional spatial segmentation image) can be obtained.

[0051] 42. Find the three-dimensional maximum connected component on the three-dimensional space segmentation image (i.e., the cerebrovascular segmentation image after three-dimensional conditional random field post-processing), and remove the unconnected small blocks of signal noise in the three-dimensional space segmentation image based on the three-dimensional maximum connected component to obtain the cerebrovascular segmentation image.

[0052] In another exemplary embodiment of the present application, the basic features of cerebral blood vessels are mainly obtained by extracting the length and volume of individual tubular structures and the diameter and length of local brain regions. Based on this, the process of extracting the basic features of cerebral blood vessels based on the cerebral blood vessel segmentation three-dimensional image in step 106 includes: 11) Count the number of voxels with positive labels in the 3D image of cerebral vascular segmentation, and obtain the absolute length of cerebral vessels based on the number of voxels.

[0053] 12) Obtain the individual brain tissue volume, and obtain the relative cerebral blood vessel length based on the absolute cerebral blood vessel length and the individual brain tissue volume, and define the relative cerebral blood vessel length as the cerebral blood vessel length. Considering that the size of individual brain tissue can significantly affect the length of cerebral blood vessel structure, this embodiment divides the absolute length by the individual brain tissue volume to obtain the relative cerebral blood vessel length.

[0054] 13) Determine the voxel volume based on the voxel resolution, and obtain the absolute volume of the cerebral blood vessels in the cerebral blood vessel segmentation three-dimensional image based on the statistical number of voxels.

[0055] 14) Based on the absolute volume of cerebral blood vessels and the volume of individual brain tissue, the relative cerebral blood vessel volume is obtained, and the relative cerebral blood vessel volume is used as the volume of cerebral blood vessels. For example, considering that the size of individual brain tissue can also significantly affect the volume of cerebral blood vessel structure, this embodiment can also divide the absolute volume of cerebral blood vessels by the size of individual brain tissue to obtain the relative volume, so as to reflect the average distribution density of cerebral blood vessel tissue in the whole brain tissue.

[0056] Furthermore, in the process of determining the length and diameter of cerebral blood vessels in local brain regions, the MNI standard brain template and the cerebral vascular magnetic resonance image can be nonlinearly registered using a two-step registration method, and the Harvard-Oxford brain partition map in the MNI standard brain space can be converted to the magnetic resonance angiography image space at the individual level. The local blood vessel length and diameter are calculated in each brain partition according to the above steps 11) to 14) to reflect the heterogeneity of the basic characteristics of blood vessels in the spatial distribution of each brain region.

[0057] In another exemplary embodiment of the present application, the process of extracting cerebral blood vessel geometric features based on the cerebral blood vessel segmentation three-dimensional image in step 106 includes: 21) In the three-dimensional image of cerebral vascular segmentation, the curvature radius and bending degree of each voxel on the central axis in the three-dimensional space are determined according to the tangent vector flip angle and the Frenet frame.

[0058] 22) Determine the average vascular curvature radius and the average vascular tortuosity in each brain region of the 3D cerebral vascular segmentation image (or each brain region corresponding to the Harvard-Oxford brain partition map).

[0059] 23) The radius of curvature of cerebral blood vessels is obtained based on the radius of curvature of each voxel in three-dimensional space and the average radius of curvature of blood vessels in the brain region.

[0060] 24) Based on the curvature of each voxel in three-dimensional space and the average curvature of blood vessels in the brain region, the curvature of cerebral blood vessels is obtained to measure the geometric brain distribution characteristics of cerebral blood vessels.

[0061] In another exemplary embodiment of the present application, the process of extracting cerebral vascular topological features based on the cerebral vascular segmentation three-dimensional image in step 106 includes: 31) Based on the 3D image of cerebral vascular segmentation, the tree structure is used to characterize the adjacency relationship of each voxel on the central axis, the bifurcation points are obtained, the number of bifurcations and the spatial distribution of bifurcation points are counted, so as to realize the automatic identification of bifurcation points in the vascular tree structure.

[0062] 32) Determine the angle between the tangent vectors of the adjacent voxels at the bifurcation point. This angle is the bifurcation angle of the bifurcation point.

[0063] 33) Based on the number of bifurcations and the spatial distribution of bifurcation points, the average bifurcation angle and cerebrovascular bifurcation density are determined in each brain region of the cerebrovascular segmentation three-dimensional image.

[0064] 34) Based on the angle between the tangent vectors of adjacent voxels and the average bifurcation angle, the overall bifurcation angle characteristics of cerebral blood vessels are determined to reflect the distribution characteristics of vascular bifurcation statistics in the brain region space.

[0065] In another exemplary embodiment of the present application, the process of extracting Voxel based morphometry (VBM) features based on the 3D image of cerebral blood vessel segmentation in step 106 includes: 41) Obtain cerebrovascular segmentation images of healthy people, and iteratively align the cerebrovascular segmentation images of healthy people to obtain a population cerebrovascular distribution density map.

[0066] 42) The cerebrovascular segmentation three-dimensional image is nonlinearly registered with the group cerebrovascular distribution density map by nonlinear registration, and the transformed Jacobian matrix is ​​obtained.

[0067] 43) Using the Jacobian matrix to standardize and register the 3D images of cerebral vascular segmentation, the distribution density of cerebral vessels in the standard atlas space is obtained. This distribution density can quantitatively evaluate the differences in the density of individual vascular distribution networks and the local or global levels between groups.

[0068] In another exemplary embodiment of the present application, the implementation process of the above step 108 may include: Step 1: Organize data and control quality: The multi-level standardized imaging feature group data provided above are aligned with the basic demographic indicators such as gender and age of the subjects. Outliers are detected based on the Median Absolute Deviation (MAD) and outlier data are removed. Finally, the data is structured into a data frame for subsequent downstream analysis.

[0069] Step 2: Construct a cerebrovascular biological age prediction model: In order to ensure the interpretability of the cerebrovascular biological age prediction model and realize the direct clinical application value of the cerebrovascular biological age prediction model, this embodiment uses the generalized linear model as the basic framework of the cerebrovascular biological age prediction model. On this basis, in order to prevent the overfitting of the cerebrovascular biological age prediction model, the lasso method is used to perform L1 norm penalty on the generalized linear model parameters. The specific formula cost function form is as follows: .

[0070] In the formula, Representative About The cost function of are the parameters in the generalized linear model, represents the outcome variable of the ith sample (i.e., physical age), Represents the imaging feature group data of the i-th sample, T Represents a matrix transpose operation. Represents the penalty hyperparameter, which needs to be set manually to control the feature L1 norm The degree of punishment.

[0071] In order to fully train the constructed initial model with data, the ensemble learning method can be used to perform Bootstrap sampling on the training set data constructed in step one to obtain multiple subsets of the training set, perform initial model training on each subset, and obtain multiple trained linear models (i.e., cerebrovascular biological age prediction model). Finally, the output results of these linear models are weighted averaged to obtain the final output result.

[0072] Among them, the data set is randomly divided into a training set and a test set, and multiple models are trained with ten-fold cross validation in the training set. The optimal model is selected as the cerebrovascular biological age prediction model (the model accepts individual-level imaging feature group data and basic demographic indicators such as gender and age as model input), and the model is verified in the test set. The output of the model is the biological age of the cerebrovascular, and the difference between it and the individual's physical age is the cerebrovascular aging index proposed in this application, which can quantitatively and objectively measure the degree of cerebrovascular aging in patients' imaging. At the same time, the effect coefficient corresponding to each imaging feature is obtained in the cerebrovascular biological age prediction model. This coefficient can quantitatively evaluate the contribution of the imaging feature to the overall cerebrovascular aging index, indicating the imaging feature that has the greatest impact on vascular aging, which has a strong clinical indication significance.

[0073] Based on the above description, compared with the current technology of evaluating cerebrovascular diseases based on single image analysis, this application integrates a multi-feature fusion index based on cerebrovascular diseases. Since aging is not reflected in only one aspect, by adopting this multi-feature fusion index, its accuracy is qualitatively improved compared with the cerebrovascular aging form that only reflects one aspect. By defining the model output as the biological age of cerebrovascular diseases and comparing it with the physical age of the individual, the cerebrovascular aging index is obtained, which can quantitatively and objectively measure the degree of cerebrovascular aging in patients' imaging, providing clinicians with a new tool to assess the vascular health of patients.

[0074] Furthermore, the present application combines generalized linear models (GLM), LASSO regularization, ensemble learning and other technologies, and has significant innovation and advantages. Among them, the generalized linear model (GLM), as a classic statistical model, has strong interpretability. Through LASSO regularization, the generalized linear model can automatically select the most important imaging features, and each feature has a corresponding effect coefficient. This allows the present application to clearly know which imaging features have the greatest impact on the cerebrovascular aging index, thereby providing intuitive explanations and guidance for clinicians. LASSO regularization not only prevents the generalized linear model from overfitting, but also quantifies the contribution of each imaging feature to the cerebrovascular aging index through the effect coefficient. This quantitative evaluation helps to identify the imaging features that have the greatest impact on vascular aging, provides a basis for exploring the potential mechanisms of vascular aging, and provides important clues for subsequent clinical research and optimization of treatment plans.

[0075] In addition, the generalized linear model has a simple structure, is easy to implement and deploy, and is suitable for promotion and application in different medical institutions. Compared with deep learning models, generalized linear models and LASSO regularization provide a more transparent model structure, allowing doctors to understand the decision-making process of the model, which is particularly important in the medical field because doctors need to have full trust in the prediction results of the model. In this way, the model can be more widely used in clinical practice to avoid the uncertainty of the "black box" problem.

[0076] Furthermore, this application uses a basic structure that combines 3D-UNet and an attention gating module to construct a deep neural network model, which can not only achieve multi-scale end-to-end segmentation, but also improve the segmentation performance of the model. The multi-scale feature extraction capability of 3D-UNet combined with the dynamic weight adjustment of the attention gating module can better handle small or fuzzy boundaries while maintaining the global structure, thereby improving the accuracy of segmentation. The attention gating module can help the network reduce its dependence on noise and irrelevant areas, making the model more robust when facing low-quality or uneven medical images, reducing the risk of mis-segmentation. Attention mechanisms can help the network converge to the optimal solution faster because they can guide the network to focus on important features and optimize feature extraction efficiency.

[0077] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 2 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface and the input / output interface are connected to the system bus together. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data for determining the cerebrovascular aging index. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for determining the cerebrovascular aging index is implemented.

[0078] Those skilled in the art will understand that Figure 2The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components. In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0079] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0080] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0081] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0082] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0083] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited thereto. The processors involved in each embodiment provided in this application include but are not limited to general-purpose processors, central processing units, graphics processors, digital signal processors, programmable logic devices (PLDs), data processing logic based on quantum computing, etc.

[0084] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0085] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for determining a cerebrovascular aging index, characterized in that: include: Obtaining a magnetic resonance image of the brain of the user to be determined; Obtaining a cerebral vascular segmentation image based on a magnetic resonance image; The Voronoi covariance is used to extract the central axis of the cerebral vascular segmentation image; The tangent corresponding to each voxel on the central axis is determined by using the λ-maximum segment tangent method, and the normal plane on each voxel is determined based on the tangent; For each voxel on the central axis, fit the largest circle in the corresponding normal plane and determine the diameter of the circle corresponding to the voxel; Combining the central axis and the diameter, the tubular tree structure in the cerebral blood vessel segmentation image is reconstructed in three dimensions to obtain a cerebral blood vessel segmentation three-dimensional image; Extract multi-level standardized imaging features based on 3D images of cerebral vascular segmentation; Get the age of the user to be determined; Constructing a cerebrovascular biological age prediction model; the cerebrovascular biological age prediction model is a neural network model constructed based on a generalized linear model; Input the multi-level standardized imaging features into the cerebrovascular biological age prediction model to obtain the cerebrovascular biological age prediction results; The cerebrovascular aging index was obtained based on the cerebrovascular biological age prediction results and age.

2. The method for determining the cerebrovascular aging index according to claim 1, characterized in that: Obtaining a cerebral blood vessel segmentation image based on the magnetic resonance image includes: Preprocessing the magnetic resonance image to obtain a preprocessed image; Constructing a cerebrovascular tissue segmentation model; the cerebrovascular tissue segmentation model is a network model constructed and trained by introducing an attention gating module into the skip connection of a 3D-UNet convolutional neural network; Inputting the preprocessed image into the cerebrovascular tissue segmentation model to obtain a segmented image; The segmented image is post-processed to obtain the cerebral blood vessel segmented image.

3. The method for determining the cerebrovascular aging index according to claim 2, characterized in that: Preprocessing the magnetic resonance image to obtain a preprocessed image includes: Performing brain tissue-specific identification and segmentation on the magnetic resonance image to obtain a brain tissue structure image; By using the linear rigid body transformation and spline interpolation method, the brain tissue structure image is transformed to the same anatomical coordinates to obtain the transformed image; Performing image denoising processing on the converted image using an adaptive non-local means algorithm to obtain a denoised image; Using a signal intensity remapping method after Gaussian deconvolution, magnetic resonance image artifacts caused by B0 field inhomogeneity in the denoised image are removed to obtain an intermediate image; Processing the intermediate image using a contrast-limited adaptive histogram equalization method to obtain a first signal-enhanced image; Using the Sobel convolution method, identifying edge positions where a sudden change in signal intensity occurs in the first signal-enhanced image, and obtaining a second signal-enhanced image based on the identified edge positions; The second signal enhanced image is processed by a gamma correction method to obtain the preprocessed image.

4. The method for determining the cerebrovascular aging index according to claim 2, characterized in that: Post-processing the segmented image to obtain the cerebral blood vessel segmented image includes: A three-dimensional conditional random field model is used to perform penalty processing on adjacent points with similar signal strengths on the segmented image to obtain a three-dimensional space segmented image; the adjacent points with similar signal strengths refer to pixel points whose signal strength difference is within a set range; The three-dimensional maximum connected component is found on the three-dimensional space segmentation image, and the unconnected signal noise in the three-dimensional space segmentation image is removed based on the three-dimensional maximum connected component to obtain the cerebral blood vessel segmentation image.

5. The method for determining the cerebrovascular aging index according to claim 1, characterized in that: The process of extracting the basic features of the cerebral blood vessels based on the cerebral blood vessel segmentation three-dimensional image includes: Counting the number of voxels with positive markers in the cerebral blood vessel segmentation three-dimensional image, and obtaining the absolute length of the cerebral blood vessel based on the number of voxels; Acquiring the individual brain tissue volume, and obtaining the relative cerebral blood vessel length based on the absolute cerebral blood vessel length and the individual brain tissue volume, and using the relative cerebral blood vessel length as the cerebral blood vessel length; Determine the voxel volume based on the image resolution, and obtain the absolute volume of the cerebral blood vessels in the cerebral blood vessel segmentation three-dimensional image based on the statistical number of the voxels; A relative cerebral vascular volume is obtained based on the absolute cerebral vascular volume and the individual brain tissue volume, and the relative cerebral vascular volume is used as the volume of the cerebral blood vessels.

6. The method for determining the cerebrovascular aging index according to claim 1, characterized in that: The process of extracting cerebral blood vessel geometric features based on the cerebral blood vessel segmentation three-dimensional image includes: In the three-dimensional image of cerebral vascular segmentation, the curvature radius and bending degree of each voxel on the central axis in the three-dimensional space are determined according to the tangent vector flip angle and the Frenet frame; Determine the average vascular curvature radius and the average vascular tortuosity in each brain region of the cerebral vascular segmentation three-dimensional image; The cerebral vascular curvature radius is obtained based on the curvature radius of each voxel in three-dimensional space and the average vascular curvature radius of the brain region; The curvature of cerebral blood vessels is obtained based on the curvature of each voxel in three-dimensional space and the average curvature of blood vessels in the brain region.

7. The method for determining the cerebrovascular aging index according to claim 1, characterized in that: The process of extracting cerebrovascular topological features based on the cerebrovascular segmentation three-dimensional image includes: Based on the cerebral vascular segmentation three-dimensional image, a tree structure is used to characterize the adjacency relationship of each voxel on the central axis, bifurcation points are obtained, and the number of bifurcations and the spatial distribution of bifurcation points are counted; Determine the angle of the tangent vectors of the adjacent voxels at the bifurcation point; Based on the number of bifurcations and the spatial distribution of bifurcation points, determining an average bifurcation angle and a cerebrovascular bifurcation density in each brain region of the cerebrovascular segmented three-dimensional image; Based on the angle between the tangent vectors of adjacent voxels and the average bifurcation angle, the overall bifurcation angle characteristics of cerebral vessels were determined; The process of extracting cerebrovascular morphological features based on the cerebrovascular segmentation three-dimensional image includes: Acquire the cerebrovascular segmentation images of healthy people, and iteratively register the cerebrovascular segmentation images of healthy people to obtain the cerebrovascular distribution density map of the group; Nonlinearly registering the cerebrovascular segmentation three-dimensional image with the population cerebrovascular distribution density map in a nonlinear registration manner, and obtaining a transformed Jacobian matrix; The cerebral blood vessel segmentation three-dimensional image after the Jacobian matrix standardization registration is used to obtain the distribution density of the cerebral blood vessels in the standard atlas space.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for determining the cerebrovascular aging index according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for determining the cerebrovascular aging index according to any one of claims 1 to 7 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for determining the cerebrovascular aging index according to any one of claims 1 to 7 is implemented.

Images