Three-dimensional reconstruction method and quantitative analysis method of myocardial injury image

Abstract

The invention discloses a three-dimensional reconstruction method and a quantitative analysis method of a myocardial injury image, electronic equipment and a computer readable storage medium. The method comprises the following steps: firstly, calibrating a target area to obtain a CT image sequence, and after preprocessing such as anisotropic diffusion filtering, extracting pathological features including density variation and perfusion defect features in a section image by using a pre-trained neural network model; determining a myocardial tissue boundary and an injury area through specific sampling, gradient analysis and curve fitting, and performing multi-dimensional verification optimization on a reconstruction result; generating and optimizing a three-dimensional model on the reconstruction plane based on a correlation algorithm; and finally, extracting a model topological structure, and calculating parameters such as volume, deformation and perfusion to generate a quantitative analysis report. According to the method, the problems of low precision, insufficient analysis capability and the like of traditional two-dimensional image analysis are solved, and high-precision myocardial injury image three-dimensional reconstruction and comprehensive quantitative analysis are realized.

Description

Technical Field

[0001] The present invention belongs to the field of medical imaging technology, and in particular relates to a three-dimensional reconstruction and quantitative analysis method of myocardial injury images. Background Art

[0002] Accurate assessment of myocardial injury is crucial in the diagnosis and treatment of cardiovascular diseases. Traditional imaging analysis of myocardial injury primarily relies on two-dimensional cross-sectional images, which present numerous limitations. Firstly, two-dimensional images fail to intuitively and comprehensively depict the spatial structure and positional relationships of myocardial injury. This lacks a three-dimensional perspective when assessing the extent and severity of injury, which can easily lead to misdiagnosis or missed diagnosis. Secondly, existing technologies for extracting myocardial tissue pathological features rely on a single approach, failing to fully capture key information such as density variation and perfusion defects, resulting in inaccurate quantitative analysis of myocardial injury. Furthermore, traditional methods suffer from poor noise suppression during image preprocessing and are unable to effectively distinguish between changes in myocardial characteristics during different temporal phases (such as systole and diastole), further compromising the reliability and accuracy of myocardial injury assessment. With the advancement of medical technology, there is an urgent need for more accurate and efficient three-dimensional reconstruction and quantitative analysis of myocardial injury images to improve the diagnosis and treatment of cardiovascular diseases. Summary of the Invention

[0003] To this end, the present invention provides a three-dimensional reconstruction method and a quantitative analysis method for myocardial injury images.

[0004] A first aspect of the present invention provides a method for three-dimensional reconstruction of myocardial injury images, comprising the following steps:

[0005] S1. Demarcate the target area, obtain cross-sectional images of the target area along multiple axial sections, and obtain a CT image sequence of the target area;

[0006] S2. Preprocessing the CT image sequence, extracting pathological features of each section image in the CT image sequence using a pretrained neural network model, wherein the pathological features include at least density variation features and perfusion defect features of myocardial tissue;

[0007] S3. fitting the boundary features in each section image into a damage image including the myocardial damage area according to the extracted pathological features;

[0008] S4. Select a reconstruction plane, output a damaged area data set including a section image and a three-dimensional model, and associate the myocardial damaged area with one or more corresponding section images.

[0009] As a preferred embodiment, the preprocessing includes:

[0010] S21, performing a first preprocessing to suppress noise through anisotropic diffusion filtering;

[0011] S22, performing a second preprocessing to divide the CT image sequence into a systolic subsequence and a diastolic subsequence according to the temporal phase;

[0012] S23 , performing dynamic region cropping on each of the subsequences to generate a region image block of a preset size.

[0013] As a preferred embodiment, in S2, extracting pathological features includes the following steps:

[0014] S204, randomly selecting an initial sampling point in the selected target section image as a reference coordinate;

[0015] S205: Define a first axis direction and a second axis direction perpendicular to the first axis direction along the reference coordinates, obtaining a first slice image set and a second slice image set along the second axis direction, wherein each slice image in the first slice image set is parallel to the target slice image, and each slice image in the second slice image set is surrounded by the same axis as the centerline, and the axis is perpendicular to the second axis direction;

[0016] S206, respectively calculating the pixel intensity gradient change rate in the first slice image set and the second slice image set within the preset spherical sampling radius with the reference coordinate point as the origin;

[0017] S207, setting a double-threshold intensity interval for a standard myocardial structure, and determining that a gradient stable region in the interval [α1, α2] is the myocardium;

[0018] S208, establishing a first pixel band within the gradient stationary region, detecting a set of boundary points in the first pixel band whose gradient change rate is lower than a β threshold, and fitting the boundary points with a spline curve using a preset adjacent gradient change rate to obtain a myocardial tissue boundary;

[0019] S209, performing secondary random sampling within the fitted myocardial tissue boundary to obtain detection coordinates, calculating the average distribution of gradient change rates using the detection coordinates as reference points, and taking a set of points whose gradient change rates are within a preset threshold as a set of texture feature points;

[0020] S210, detecting a texture feature point set whose gradient change rate exceeds a γ threshold as an abnormal area and a damage candidate area;

[0021] S211 , fitting the texture feature points in each of the candidate damage areas to obtain a first complete damage portion and its boundary.

[0022] As a preferred embodiment, after S211, the following steps are further included:

[0023] S212, verifying a first complete lesion boundary obtained by fitting the first section image set and the second section image set to obtain a first fitting difference value of the complete lesion boundary;

[0024] S213, after selecting a new target section image, execute steps S204-S211 to obtain a second complete lesion and its boundary;

[0025] S214: Obtain a second fitting difference between the first intact damage portion boundary and the second intact damage portion boundary within a preset spherical sampling radius;

[0026] S215, obtaining all selected target section images along the same axis at a specified CT imaging depth, and comparing a first fitting difference and a second fitting difference between each target section image and another adjacent target section image;

[0027] S216, averaging the first fitting differences to obtain a first corrected difference, and averaging the second fitting differences to obtain a second corrected difference;

[0028] S217, writing the first fitting difference into the first complete lesion boundary obtained from the first section image to obtain a first lesion image;

[0029] The second fitting difference is written into the first damaged image to obtain a second damaged image.

[0030] As a preferred embodiment, after S211, the following steps are further included:

[0031] After executing steps S204 to S211 on the section images in the systolic period subsequence and the diastolic period subsequence respectively, the first lesion image and the second lesion image are written into the regional image block to obtain a diastolic period lesion portion image and a systolic period lesion portion image;

[0032] comparing a first deformation difference of a boundary of a complete lesion portion at an extreme position between the diastolic lesion portion image and the systolic lesion portion image;

[0033] Screening a sample whose average pixel intensity gradient is within a preset interval between the diastolic lesion image and the systolic lesion image from the regional image block, wherein the sample is the regional image block cut from the slice images in the first slice image set and the second slice image set;

[0034] The calibration sample is located in the same point set interval as the intact damaged part interval, and the second deformation difference of the point set interval is obtained;

[0035] The degree of damage to the intact damaged part is determined according to the position of the myocardial functional zone where the difference between the first deformation difference and the second deformation difference is located.

[0036] The perfusion defect characteristics under the current myocardial deformation are obtained according to the total blood inflow.

[0037] As a preferred method, in the reconstruction plane, perform the following steps:

[0038] The first lesion image is used as the main reconstruction data set, and the initial myocardial surface mesh model is generated based on the Delaunay triangulation algorithm;

[0039] The second lesion image is used as the reference image database, and voxel-level difference mapping is achieved between the main and reference images of the same anatomical site through feature space registration. The displacement vector field of the mesh vertices in the systolic-diastolic phase is calculated. When the local deformation difference of the displacement vector field is greater than the preset value, the process returns to step S2 for refitting, and the new first and second lesion images are used to reconstruct the three-dimensional model.

[0040] A second aspect of the present invention provides a method for quantitatively analyzing myocardial injury applied to the first aspect of the present invention, comprising the following steps:

[0041] Extracting the topological structure of the myocardial injury area in the three-dimensional model, and using the boundary points of each myocardial injury area as the topological connection points of the topological structure;

[0042] Calculate the volume quantitative parameters of the myocardial injury area, calculate the absolute injury volume based on the voxel integration method of the Delaunay triangular grid, and obtain the volume parameters

[0043] Calculate the displacement vector field during contraction and relaxation, measure the displacement at the grid vertices, and calculate the average radial deformation rate within the damaged area. Mark the area ratio of the deformation interval outside the preset value to obtain the deformation parameter.

[0044] The number of voxels in the lesion core area with CT values ​​less than the preset value was counted, and the abnormal perfusion index with gradient change rate outside the preset value was calculated to obtain the perfusion parameters;

[0045] Generate a quantitative analysis report including volume parameters, deformation parameters and perfusion parameters.

[0046] The third aspect of the present invention provides an electronic device for implementing the method of the first aspect or the second aspect of the present invention.

[0047] A fourth aspect of the present invention provides a storage medium for implementing the method of the first or second aspect of the present invention.

[0048] The above technical solution of the present invention has the following advantages over the prior art:

[0049] This method acquires CT image sequences from multiple axial sections of the target region and uses a pretrained neural network model to extract pathological features, including myocardial tissue density variation and perfusion defects, to comprehensively and accurately capture myocardial injury information. During feature extraction, specific sampling and gradient analysis methods are employed, such as acquiring a set of section images along an axis determined by reference coordinates and calculating the rate of change of pixel intensity gradients. Combined with methods such as dual-threshold intensity intervals and spline curve fitting, this method precisely determines myocardial tissue boundaries and damaged areas, effectively improving the accuracy of three-dimensional reconstruction of myocardial injury images.

[0050] The present invention performs multi-dimensional verification of the reconstruction results, including checking the lesion boundaries obtained by fitting different sets of cross-sectional images and comparing deformation differences in lesion images at different time stages (systole and diastole). These verification steps can promptly detect and correct errors in the reconstruction process, ensuring the reliability of the reconstruction results. Simultaneously, the reconstructed model is optimized within the reconstruction plane through feature space registration and displacement vector field analysis, further improving its accuracy and stability.

[0051] This invention not only enables 3D reconstruction of myocardial injury images but also provides a comprehensive set of quantitative analysis methods. By extracting the topological structure of the damaged myocardial area within the 3D model and calculating quantitative volumetric parameters, deformation parameters, and perfusion parameters, it enables quantitative assessment of myocardial injury from multiple perspectives, providing physicians with rich and accurate diagnostic information and facilitating the development of more scientific and rational treatment plans.

[0052] In the image preprocessing stage, the present invention suppresses noise through anisotropic diffusion filtering, segments the image sequence according to the temporal stage, and performs dynamic area cropping, which can effectively reduce the data processing volume and improve data processing efficiency, laying a good foundation for subsequent feature extraction and three-dimensional reconstruction, making the entire myocardial injury image processing process more efficient and faster. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 1 is a flow chart of the method provided in Example 1 of the present invention;

[0054] Figure 2 It is a structural diagram of an electronic device provided in the second embodiment of the present invention. DETAILED DESCRIPTION

[0055] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0056] Example 1

[0057] In a first aspect of the present disclosure, a method for three-dimensional reconstruction of myocardial injury images is provided, comprising the following steps:

[0058] S1. Demarcate the target area, obtain cross-sectional images of the target area along multiple axial sections, and obtain a CT image sequence of the target area;

[0059] S2. Preprocessing the CT image sequence, extracting pathological features of each section image in the CT image sequence using a pretrained neural network model, wherein the pathological features include at least density variation features and perfusion defect features of myocardial tissue;

[0060] S3. fitting the boundary features in each section image into a damage image including the myocardial damage area according to the extracted pathological features;

[0061] S4. Select a reconstruction plane, output a damaged area data set including a section image and a three-dimensional model, and associate the myocardial damaged area with one or more corresponding section images.

[0062] As a preferred embodiment, the preprocessing includes:

[0063] S21, performing a first preprocessing to suppress noise through anisotropic diffusion filtering;

[0064] S22, performing a second preprocessing to divide the CT image sequence into a systolic subsequence and a diastolic subsequence according to the temporal phase;

[0065] S23 , performing dynamic region cropping on each of the subsequences to generate a region image block of a preset size.

[0066] As a preferred embodiment, in S2, extracting pathological features includes the following steps:

[0067] S204, randomly selecting an initial sampling point in the selected target section image as a reference coordinate;

[0068] S205: Define a first axis direction and a second axis direction perpendicular to the first axis direction along the reference coordinates, obtaining a first slice image set and a second slice image set along the second axis direction, wherein each slice image in the first slice image set is parallel to the target slice image, and each slice image in the second slice image set is surrounded by the same axis as the centerline, and the axis is perpendicular to the second axis direction;

[0069] S206, respectively calculating the pixel intensity gradient change rate in the first slice image set and the second slice image set within the preset spherical sampling radius with the reference coordinate point as the origin;

[0070] S207, setting a double-threshold intensity interval for a standard myocardial structure, and determining that a gradient stable region in the interval [α1, α2] is the myocardium;

[0071] S208, establishing a first pixel band within the gradient stationary region, detecting a set of boundary points in the first pixel band whose gradient change rate is lower than a β threshold, and fitting the boundary points with a spline curve using a preset adjacent gradient change rate to obtain a myocardial tissue boundary;

[0072] S209, performing secondary random sampling within the fitted myocardial tissue boundary to obtain detection coordinates, calculating the average distribution of gradient change rates using the detection coordinates as reference points, and taking a set of points whose gradient change rates are within a preset threshold as a set of texture feature points;

[0073] S210, detecting a texture feature point set whose gradient change rate exceeds a γ threshold as an abnormal area and a damage candidate area;

[0074] S211 , fitting the texture feature points in each of the candidate damage areas to obtain a first complete damage portion and its boundary.

[0075] As a preferred embodiment, after S211, the following steps are further included:

[0076] S212, verifying a first complete lesion boundary obtained by fitting the first section image set and the second section image set to obtain a first fitting difference value of the complete lesion boundary;

[0077] S213, after selecting a new target section image, execute steps S204-S211 to obtain a second complete lesion and its boundary;

[0078] S214: Obtain a second fitting difference between the first intact damage portion boundary and the second intact damage portion boundary within a preset spherical sampling radius;

[0079] S215, obtaining all selected target section images along the same axis at a specified CT imaging depth, and comparing a first fitting difference and a second fitting difference between each target section image and another adjacent target section image;

[0080] S216, averaging the first fitting differences to obtain a first corrected difference, and averaging the second fitting differences to obtain a second corrected difference;

[0081] S217, writing the first fitting difference into the first complete lesion boundary obtained from the first section image to obtain a first lesion image;

[0082] The second fitting difference is written into the first damaged image to obtain a second damaged image.

[0083] As a preferred embodiment, after S211, the following steps are further included:

[0084] After executing steps S204 to S211 on the section images in the systolic period subsequence and the diastolic period subsequence respectively, the first lesion image and the second lesion image are written into the regional image block to obtain a diastolic period lesion portion image and a systolic period lesion portion image;

[0085] comparing a first deformation difference of a boundary of a complete lesion portion at an extreme position between the diastolic lesion portion image and the systolic lesion portion image;

[0086] Screening a sample whose average pixel intensity gradient is within a preset interval between the diastolic lesion image and the systolic lesion image from the regional image block, wherein the sample is the regional image block cut from the slice images in the first slice image set and the second slice image set;

[0087] The calibration sample is located in the same point set interval as the intact damaged part interval, and the second deformation difference of the point set interval is obtained;

[0088] The degree of damage to the intact damaged part is determined according to the position of the myocardial functional zone where the difference between the first deformation difference and the second deformation difference is located.

[0089] The perfusion defect characteristics under the current myocardial deformation are obtained according to the total blood inflow.

[0090] As a preferred method, in the reconstruction plane, perform the following steps:

[0091] The first lesion image is used as the main reconstruction data set, and the initial myocardial surface mesh model is generated based on the Delaunay triangulation algorithm;

[0092] The second lesion image is used as the reference image database, and voxel-level difference mapping is achieved between the main and reference images of the same anatomical site through feature space registration. The displacement vector field of the mesh vertices in the systolic-diastolic phase is calculated. When the local deformation difference of the displacement vector field is greater than the preset value, the process returns to step S2 for refitting, and the new first and second lesion images are used to reconstruct the three-dimensional model.

[0093] A second aspect of the present disclosure provides a method for quantitatively analyzing myocardial injury applied to the first aspect of the present disclosure, comprising the following steps:

[0094] Extracting the topological structure of the myocardial injury area in the three-dimensional model, and using the boundary points of each myocardial injury area as the topological connection points of the topological structure;

[0095] Calculate the volume quantitative parameters of the myocardial injury area, calculate the absolute injury volume based on the voxel integration method of the Delaunay triangular grid, and obtain the volume parameters

[0096] Calculate the displacement vector field during contraction and relaxation, measure the displacement at the grid vertices, and calculate the average radial deformation rate within the damaged area. Mark the area ratio of the deformation interval outside the preset value to obtain the deformation parameter.

[0097] The number of voxels in the lesion core area with CT values ​​less than the preset value was counted, and the abnormal perfusion index with gradient change rate outside the preset value was calculated to obtain the perfusion parameters;

[0098] Generate a quantitative analysis report including volume parameters, deformation parameters and perfusion parameters.

[0099] The disclosed embodiments acquire CT image sequences of multiple axial sections of the target region and utilize a pre-trained neural network model to extract pathological features, including myocardial tissue density variation and perfusion defect characteristics, to comprehensively and accurately capture myocardial injury information. During feature extraction, specific sampling and gradient analysis methods are employed, such as acquiring a set of section images along the axis determined by reference coordinates and calculating the rate of change of pixel intensity gradients. Combined with methods such as dual-threshold intensity intervals and spline curve fitting, myocardial tissue boundaries and damaged areas can be precisely determined, effectively improving the accuracy of three-dimensional reconstruction of myocardial injury images.

[0100] The disclosed embodiments perform multi-dimensional verification of the reconstruction results, including verification of the lesion boundaries obtained by fitting different sets of cross-sectional images and comparison of deformation differences between lesion images at different time stages (systole and diastole). These verification steps enable timely detection and correction of errors in the reconstruction process, ensuring the reliability of the reconstruction results. Furthermore, the reconstructed model is optimized within the reconstruction plane through feature space registration and displacement vector field analysis, further improving its accuracy and stability.

[0101] The disclosed embodiments not only achieve 3D reconstruction of myocardial injury images but also provide a comprehensive set of quantitative analysis methods. By extracting the topological structure of the myocardial injury region in the 3D model and calculating volumetric parameters, deformation parameters, and perfusion parameters, this method enables quantitative assessment of myocardial injury from multiple perspectives, providing physicians with rich and accurate diagnostic information and helping to develop more scientific and reasonable treatment plans.

[0102] In the image preprocessing stage, the disclosed embodiment performs noise suppression through anisotropic diffusion filtering, segments the image sequence according to the temporal stage, and performs dynamic region cropping. This can effectively reduce the amount of data processing, improve data processing efficiency, lay a good foundation for subsequent feature extraction and three-dimensional reconstruction, and make the entire myocardial injury image processing process more efficient and faster.

[0103] Example 2

[0104] Combine Figure 2 As shown, an embodiment of the present disclosure provides an electronic device including a processor 30 and a memory 31. Optionally, the electronic device may further include a communication interface 32 and a bus 33. The processor 30, communication interface 32, and memory 31 may communicate with each other via the bus 33. The communication interface 32 may be used for information transmission. The processor 30 may invoke logic instructions in the memory 31 to execute the method of the first embodiment described above.

[0105] An embodiment of the present disclosure provides a storage medium storing computer-executable instructions, wherein the computer-executable instructions are configured to execute the method of the first embodiment.

[0106] The aforementioned storage medium may be either a transient computer-readable storage medium or a non-transient computer-readable storage medium. Non-transient storage media include various media capable of storing program code, such as USB flash drives, external hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks. These media may also be transient storage media.

[0107] The above description and the accompanying drawings sufficiently illustrate the embodiments of the present disclosure to enable those skilled in the art to practice them. Other embodiments may include structural, logical, electrical, process and other changes. The embodiments represent only possible variations. Unless expressly required, individual components and functions are optional, and the order of operations may vary. Portions and features of some embodiments may be included in or replace portions and features of other embodiments. Moreover, the terms used in this application are only used to describe the embodiments and are not used to limit the claims. As used in the description of the embodiments and claims, the singular forms "a", "an" and "the" are intended to also include the plural forms unless the context clearly indicates otherwise. Similarly, the term "and / or" as used in this application means any and all possible combinations of one or more of the associated listings. In addition, when used in this application, the term "comprise" and its variations "comprises" and / or comprising refer to the presence of stated features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or groups thereof. In the absence of further restrictions, an element defined by the sentence "comprising a..." does not exclude the presence of other identical elements in the process, method or device that includes the element. In this article, each embodiment may focus on the differences from other embodiments, and the same and similar parts between the various embodiments can be referenced to each other. For the methods, products, etc. disclosed in the embodiments, if they correspond to the method part disclosed in the embodiments, then the relevant parts can be referred to the description of the method part.

[0108] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these effects are performed in hardware or software may depend on the specific application and design constraints of the technical solution. The technicians may use different methods for each specific application to implement the described effects, but such implementations should not be considered to exceed the scope of the embodiments of the present disclosure. The technicians will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described devices, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0109] The flowcharts and block diagrams in the accompanying drawings show the possible architectures, functions and operations of the devices, methods and computer program products according to the embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, program segment or part of the code, and the module, program segment or part of the code contains one or more executable instructions for implementing the specified logical functions. In some alternative implementations, the functions marked in the boxes can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed in parallel, or they can sometimes be executed in the opposite order, which can depend on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different boxes can also occur in an order different from that disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps can actually be executed in parallel, or they can sometimes be executed in the opposite order, which can depend on the functions involved. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented by dedicated hardware-based devices that perform the specified functions or actions, or may be implemented by a combination of dedicated hardware and computer instructions.

Claims

1. A three-dimensional reconstruction method for myocardial injury images, characterized in that: The steps include: S1. Demarcate the target area, obtain cross-sectional images of the target area along multiple axial sections, and obtain a CT image sequence of the target area; S2. Preprocessing the CT image sequence, extracting pathological features of each section image in the CT image sequence using a pretrained neural network model, wherein the pathological features include at least density variation features and perfusion defect features of myocardial tissue; S3. fitting the boundary features in each section image into a damage image including the myocardial damage area according to the extracted pathological features; S4. Select a reconstruction plane, output a damaged area data set including a section image and a three-dimensional model, and associate the myocardial damaged area with one or more corresponding section images.

2. The three-dimensional reconstruction method of myocardial injury images according to claim 1, characterized in that: S21, performing a first preprocessing to suppress noise through anisotropic diffusion filtering; S22, performing a second preprocessing to divide the CT image sequence into a systolic subsequence and a diastolic subsequence according to the temporal phase; S23 , performing dynamic region cropping on each of the subsequences to generate a region image block of a preset size.

3. The three-dimensional reconstruction method of myocardial injury images according to claim 2, characterized in that: In S2, extracting pathological features includes the following steps: S204, randomly selecting an initial sampling point in the selected target section image as a reference coordinate; S205: Define a first axis direction and a second axis direction perpendicular to the first axis direction along the reference coordinates, obtaining a first slice image set and a second slice image set along the second axis direction, wherein each slice image in the first slice image set is parallel to the target slice image, and each slice image in the second slice image set is surrounded by the same axis as the centerline, and the axis is perpendicular to the second axis direction; S206, respectively calculating the pixel intensity gradient change rate in the first slice image set and the second slice image set within the preset spherical sampling radius with the reference coordinate point as the origin; S207, setting a double-threshold intensity interval for a standard myocardial structure, and determining that a gradient stable region in the interval [α1, α2] is the myocardium; S208, establishing a first pixel band within the gradient stationary region, detecting a set of boundary points in the first pixel band whose gradient change rate is lower than a β threshold, and fitting the boundary points with a spline curve using a preset adjacent gradient change rate to obtain a myocardial tissue boundary; S209, performing secondary random sampling within the fitted myocardial tissue boundary to obtain detection coordinates, calculating the average distribution of gradient change rates using the detection coordinates as reference points, and taking a set of points whose gradient change rates are within a preset threshold as a set of texture feature points; S210, detecting a texture feature point set whose gradient change rate exceeds a γ threshold as an abnormal area and a damage candidate area; S211 , fitting the texture feature points in each of the candidate damage areas to obtain a first complete damage portion and its boundary.

4. The three-dimensional reconstruction method of myocardial injury images according to claim 3, characterized in that: After S211, the following steps are also included: S212, verifying a first complete lesion boundary obtained by fitting the first section image set and the second section image set to obtain a first fitting difference value of the complete lesion boundary; S213, after selecting a new target section image, execute steps S204-S211 to obtain a second complete lesion and its boundary; S214: Obtain a second fitting difference between the first intact damage portion boundary and the second intact damage portion boundary within a preset spherical sampling radius; S215, obtaining all selected target section images along the same axis at a specified CT imaging depth, and comparing a first fitting difference and a second fitting difference between each target section image and another adjacent target section image; S216, averaging the first fitting differences to obtain a first corrected difference, and averaging the second fitting differences to obtain a second corrected difference; S217, writing the first fitting difference into the first complete lesion boundary obtained from the first section image to obtain a first lesion image; The second fitting difference is written into the first damaged image to obtain a second damaged image.

5. The three-dimensional reconstruction method of myocardial injury images according to claim 4, characterized in that: After S211, the following steps are also included: After executing steps S204 to S211 on the section images in the systolic period subsequence and the diastolic period subsequence respectively, the first lesion image and the second lesion image are written into the regional image block to obtain a diastolic period lesion portion image and a systolic period lesion portion image; comparing a first deformation difference of a boundary of a complete lesion portion at an extreme position between the diastolic lesion portion image and the systolic lesion portion image; Screening a sample whose average pixel intensity gradient is within a preset interval between the diastolic lesion image and the systolic lesion image from the regional image block, wherein the sample is the regional image block cut from the slice images in the first slice image set and the second slice image set; The calibration sample is located in the same point set interval as the intact damaged part interval, and the second deformation difference of the point set interval is obtained; The degree of damage to the intact damaged part is determined according to the position of the myocardial functional zone where the difference between the first deformation difference and the second deformation difference is located. The perfusion defect characteristics under the current myocardial deformation are obtained according to the total blood inflow.

6. The three-dimensional reconstruction method of myocardial injury images according to claim 5, characterized in that: On the reconstruction plane, perform the following steps: The first lesion image is used as the main reconstruction data set, and the initial myocardial surface mesh model is generated based on the Delaunay triangulation algorithm; The second lesion image is used as the reference image database, and voxel-level difference mapping is achieved between the main and reference images of the same anatomical site through feature space registration. The displacement vector field of the mesh vertices in the systolic-diastolic phase is calculated. When the local deformation difference of the displacement vector field is greater than the preset value, the process returns to step S2 for refitting, and the new first and second lesion images are used to reconstruct the three-dimensional model.

7. The method for quantitative analysis of myocardial damage according to claim 6, characterized in that: The steps include: Extracting the topological structure of the myocardial injury area in the three-dimensional model, and using the boundary points of each myocardial injury area as the topological connection points of the topological structure; Calculate the volume quantitative parameters of the myocardial injury area, calculate the absolute injury volume based on the voxel integration method of the Delaunay triangular grid, and obtain the volume parameters Calculate the displacement vector field during contraction and relaxation, measure the displacement at the grid vertices, and calculate the average radial deformation rate within the damaged area. Mark the area ratio of the deformation interval outside the preset value to obtain the deformation parameter. The number of voxels in the lesion core area with CT values ​​less than the preset value was counted, and the abnormal perfusion index with gradient change rate outside the preset value was calculated to obtain the perfusion parameters; Generate a quantitative analysis report including volume parameters, deformation parameters and perfusion parameters.

8. An electronic device, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method according to any one of claims 1 to 7 is implemented.

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

Images