Blood vessel analysis method and system based on multi-contrast magnetic resonance image

Abstract

The invention provides a blood vessel analysis method and system based on a multi-contrast magnetic resonance image, and relates to the technical field of medical image processing. In the embodiment of the invention, the multi-contrast brain magnetic resonance scanning sequence (including T1W and TOF), T1W and TOF can provide various contrast information, the problem that manual preprocessing needs to be carried out on the skull part when blood vessel parameter extraction is carried out only by using a single TOF image in the prior art is solved, and a transformation result in a standard brain space can be additionally provided. In addition, compared with a manual and semi-automatic measurement method adopted at present, the embodiment of the invention can realize a full-automatic parameter extraction process, more manpower and time are saved, artificial factor intervention is avoided, and the result is more stable. In addition, multiple blood vessel parameters of multiple blood vessel groups can be provided, and the blood vessel variation degree can be quantified according to the transformation result in the standard brain space.

Description

Technical Field

[0001] The present invention relates to the technical field of medical image processing, and particularly to a vascular analysis method and system based on multi-contrast magnetic resonance images. Background Art

[0002] Currently, the methods for evaluating and analyzing brain vascular images are manual, that is, experienced personnel evaluate the images through visual inspection and manual measurement. However, this method has problems such as strong subjectivity, large consumption of time and resources. Some researchers have also proposed some automatic and semi-automatic programs for analyzing physiological parameters of brain vascular images. Based on computer algorithms and libraries, automated analysis and evaluation of large-scale brain vascular images can be achieved.

[0003] iCafe is a semi-automatic arterial modeling tool developed for TOF images, which integrates modules such as vascular centerline tracking algorithm, arterial anatomical structure marking, automatic vascular segmentation and analysis, and vascular structure visualization. Related research has proposed a quantification process for the visualization degree of intracranial small blood vessels. Through operations such as image preprocessing, threshold segmentation, and skeleton extraction, physiological parameters such as blood vessel radius can be obtained. Some researchers have also dedicated themselves to constructing statistical atlases using morphological parameters of intracranial arteries to improve the understanding of the distribution and subtle morphological changes of human brain arteries. Related research has also proposed a method for processing intracranial vascular statistical atlases. This method uses a linear registration method to register TOF images to a standard brain template and calculates statistical atlases including signal intensity, vascular probability, average vascular diameter, etc.

[0004] It can be seen that the current algorithms mainly focus on parameter extraction from single-contrast images and do not utilize the information of multi-contrast images. In addition, most of the current algorithms are manual or semi-automatic parameter extraction (such as in the preprocessing process of the skull part of TOF images and the classification process of different vascular groups), which consume time and manpower. Some parameter selections are affected by subjective factors, making it difficult to perform standardized processing on large batches of data. In addition, the current algorithms mainly focus on the extraction of a single parameter of blood vessel radius and cannot provide multiple parameters of multiple vascular groups. Summary of the Invention

[0005] The present invention provides a vascular analysis method and system based on multi-contrast magnetic resonance images to solve at least part of the above problems.

[0006] In a first aspect of the present invention, a vascular analysis method based on multi-contrast magnetic resonance images is provided. The method includes: Obtain multi-contrast magnetic resonance images in DICOM format, perform format conversion and standardization processing on the multi-contrast magnetic resonance images in DICOM format to obtain an image file to be processed, and the image file to be processed includes: T1W image data and TOF image data; Perform registration preprocessing operations on the image file to be processed using general brain image processing tools, obtain a brain mask in the TOF original space, perform TOF skull stripping operations, and standard brain space registration operations; Perform structured processing on the TOF image data in the original space to obtain a vascular segmentation image; convert the vascular segmentation image into a three-dimensional skeleton, convert the three-dimensional skeleton into a graph structure, and generate a topological structure file containing nodes and edges. The topological structure file characterizes the connectivity and original structure of the skeleton. Determine the blood vessel length, curvature, and surface area based on the topological structure file; obtain the segmentation results of the bilateral internal carotid arteries, bilateral middle cerebral arteries, bilateral anterior cerebral arteries, bilateral posterior cerebral arteries, and basilar artery based on the vascular segmentation image and a pre-trained classification network; Determine the blood vessel radius, length, curvature, surface area, and volume of different branch arteries based on the segmentation results of the bilateral internal carotid arteries, bilateral middle cerebral arteries, bilateral anterior cerebral arteries, bilateral posterior cerebral arteries, and basilar artery; for the statistical atlas in the standard brain space, perform radius result allocation on the vascular segmentation image, search for the centerline closest to a point in the mask, obtain the radius value of the centerline point, and obtain a blood vessel radius file; based on the registration preprocessing operation, transform the blood vessel radius file and the vascular segmentation image to the standard brain template to generate a statistical atlas of blood vessel radius distribution.

[0007] Optionally, the registration preprocessing operation includes: Obtain a non-linear transformation file from the T1W image in the original space to the standard brain template, obtain a linear transformation file for linearly registering the TOF image in the original space to the T1W space in the original space, and use the non-linear transformation file to jointly transform the TOF post-processing parameter file with the linear transformation file.

[0008] Optionally, performing structured processing on the TOF image data in the original space to obtain a vascular segmentation image includes: Analyze the gray histogram of the TOF image data to determine the maximum peak. After this maximum peak, the frequency value rapidly decreases as the brightness increases and finally approaches the x-axis; Select two points at the maximum peak and the maximum brightness frequency and draw a straight line. Take the point farthest from this straight line in the gray curve as the segmentation threshold; Calculate the volume of all connected components based on the segmentation threshold, and retain the connected components with a volume greater than 95% of the total volume to obtain a vascular segmentation image.

[0009] Optionally, determining the blood vessel length, curvature, and surface area based on the topological structure file includes: Traverse all edges, obtain the vascular segmentation chunks around each skeletal node on the edge, and calculate the radius of the largest inscribed sphere around the skeletal node as the corresponding vascular radius on the skeletal node; Obtain the vascular skeleton to be calculated and its topological connection structure, and determine the vascular length, curvature, and surface area point by point according to the position information and radius information of the skeletal nodes on each edge.

[0010] Optionally, determining the vascular length, curvature, and surface area point by point according to the position information and radius information of the skeletal nodes on each edge includes: ; ; ; wherein, ( x i ，y i ，z i ) and ( x i+1 ，y i+1 ，z i+1 ) respectively represent the position information of two adjacent points on the edge to be calculated; length represents the vascular length, surface represents the vascular surface area, radius represents the vascular radius, tortuosity represents the curvature, norm represents the straight-line distance between the starting point and the ending point of the blood vessel.

[0011] Optionally, perform format conversion and standardization processing on the DICOM format multi-contrast magnetic resonance image to obtain a to-be-processed image file, including: Use the dcm2niix conversion tool to convert the original Dicom format file to Nifti format; Normalize the obtained Nifti file using the median value of the file.

[0012] The second aspect of the present invention provides a vascular analysis system based on multi-contrast magnetic resonance images, and the vascular analysis system based on multi-contrast magnetic resonance images includes: A file management module, configured to obtain a DICOM format multi-contrast magnetic resonance image, perform format conversion and standardization processing on the DICOM format multi-contrast magnetic resonance image to obtain a to-be-processed image file, and the to-be-processed image file includes: T1W image data and TOF image data; A registration preprocessing module, configured to perform registration preprocessing operations on the to-be-processed image file using general brain image processing tools, obtain a brain mask in the TOF original space, perform TOF skull stripping operations, and standard brain space registration operations; A structuring processing module, configured to perform structuring processing on the TOF image data in the original space to obtain a vascular segmentation image; convert the vascular segmentation image into a three-dimensional skeleton, convert the three-dimensional skeleton into a graph structure, generate a topological structure file including nodes and edges, where the topological structure file represents the connectivity and original structure of the skeleton, determine the vascular length, curvature, and surface area based on the topological structure file; obtain the segmentation results of bilateral internal carotid arteries, bilateral middle cerebral arteries, bilateral anterior cerebral arteries, bilateral posterior cerebral arteries, and basilar artery based on the vascular segmentation image and a pre-trained classification network; A physiological parameter generation module, configured to determine the vascular radius, length, curvature, surface area, and volume of different branch arteries based on the segmentation results of bilateral internal carotid arteries, bilateral middle cerebral arteries, bilateral anterior cerebral arteries, bilateral posterior cerebral arteries, and basilar artery; for a statistical atlas in the standard brain space, perform radius result allocation on the vascular segmentation image, search for the centerline closest to a point in the mask, obtain the radius value of the centerline point, and obtain a vascular radius file; based on the registration preprocessing operations, transform the vascular radius file and the vascular segmentation image to a standard brain template to generate a statistical atlas of vascular radius distribution.

[0013] A third aspect of the present invention provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, where when the processor executes, it implements the vascular analysis method based on multi-contrast magnetic resonance images as described in the first aspect of the present invention.

[0014] A fourth aspect of the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the vascular analysis method based on multi-contrast magnetic resonance images as described in the first aspect of the present invention.

[0015] A fifth aspect of the present invention provides a computer program product, including a computer program / instructions, where when the computer program / instructions are executed by a processor, it implements the steps in the vascular analysis method based on multi-contrast magnetic resonance images as described in the first aspect of the present invention.

[0016] In the embodiments of the present invention, the multi-contrast brain magnetic resonance scanning sequences (including the brain structure image T1W and the cerebrovascular image TOF) T1W and TOF can provide various contrast information, solve the problem of manual preprocessing of the skull part when extracting vascular parameters using only a single TOF image in the past, and can additionally provide the transformation results in the standard brain space. In addition, compared with the currently used manual and semi-automatic measurement methods, the embodiments of the present invention can realize a fully automatic parameter extraction process, which saves more manpower and time, has no artificial factor intervention, and the results are more stable. In addition, the embodiments of the present invention can provide various vascular parameters of multiple vascular groups, and can quantify the degree of vascular variation according to the transformation results in the standard brain space.

[0017] The vascular analysis method based on multi-contrast magnetic resonance images proposed in the embodiments of the present invention can realize the parameter extraction of fully automatic brain vascular images, has more stable performance than traditional algorithms, and can batch process large-scale data. And, the embodiments of the present invention can provide various parameters including the whole cerebral blood vessels and branch arteries based on multi-contrast images, and provide the quantitative parameter distribution in the standard brain to realize a comprehensive evaluation of the vascular images. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the present invention, the drawings required for the description of the present invention will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0019] Figure 1 is the step flowchart of the vascular analysis method based on multi-contrast magnetic resonance images provided by the present invention; Figure 2 is the detailed step flowchart of the vascular analysis method based on multi-contrast magnetic resonance images provided by the present invention; Figure 3 is the sample image of the multi-contrast brain magnetic resonance scanning sequence in the vascular analysis method based on multi-contrast magnetic resonance images provided by the present invention; Figure 4 is the structural block diagram of the vascular analysis system based on multi-contrast magnetic resonance images provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below with reference to the drawings and specific embodiments.

[0021] The diagnosis and treatment of cerebrovascular diseases rely on the accurate interpretation of cerebrovascular images. However, current methods for evaluating cerebrovascular images mainly focus on some basic image quality parameters, such as commonly used image evaluation parameters like signal-to-noise ratio. These parameters often cannot comprehensively reflect the complex structure and functional status of blood vessels. To improve the accuracy of diagnosis and treatment effects, it is particularly important to develop a quantitative evaluation scheme applicable to cerebrovascular images. Currently, the commonly used cerebrovascular image scanning sequences in clinical practice include the T1-weighted (T1W) brain structure image and the time-of-flight magnetic resonance angiography (TOF) cerebrovascular image.

[0022] T1W images are used to visualize brain anatomical structures and pathologies, such as gray and white matter, tumors, and hemorrhages, and play a crucial role in the diagnosis of brain diseases. Therefore, it is the basic sequence for clinical examinations.

[0023] TOF is a commonly used non-invasive cerebrovascular imaging technique that obtains blood flow information through the inflow effect of blood. The high-signal regions in the image are the blood vessel lumens. This technique is usually used for detecting cerebral aneurysms, arteriovenous malformations, and other cerebrovascular abnormalities, and is also an important tool for the initial screening and monitoring of cerebrovascular diseases. For large-scale brain magnetic resonance image databases, it is very necessary to extract physiological parameters, which provides a reliable basis for subsequent scientific research and clinical applications.

[0024] Regarding how to accurately evaluate cerebrovascular images, current researchers pay more attention to the morphology of blood vessels, data atlases, and related parameters of several major arterial branches. In terms of morphology, researchers are more concerned about blood vessel radius, blood vessel length, blood vessel volume, and blood vessel surface area; in terms of data atlases, by statistically analyzing cerebrovascular data of a large population, average cerebrovascular image atlases, average blood vessel occurrence probability atlases, and average blood vessel radius atlases of a large population can be constructed to quantitatively analyze the subtle cerebrovascular changes of individual subjects.

[0025] Therefore, the embodiments of the present invention propose a vascular analysis method based on multi-contrast magnetic resonance images. This method automatically, accurately, and reliably uses the information provided by multi-contrast magnetic resonance images to comprehensively evaluate vascular-related parameters.

[0026] In the embodiments of the present invention, the multi-contrast brain magnetic resonance scanning sequences (including T1W and TOF) can provide various contrast information. T1W and TOF solve the problem of manual preprocessing of the skull part when extracting vascular parameters using only a single TOF image in the past, and can additionally provide the transformation results in the standard brain space. In addition, compared with the currently used manual and semi-automatic measurement methods, the embodiments of the present invention can realize a fully automatic parameter extraction process, which saves more manpower and time, has no artificial factor intervention, and the results are more stable. In addition, the embodiments of the present invention can provide various vascular parameters of multiple vascular groups, and can quantify the degree of vascular variation according to the transformation results in the standard brain space.

[0027] Specifically, the embodiments of the present invention develop a vascular analysis method based on multi-contrast magnetic resonance images. Figure 1 The step flowchart of the method is shown. Figure 2 The detailed step flowchart of the method is shown, as Figure 1 shown, the method includes the following steps: S101, obtain multi-contrast magnetic resonance images in the first format (DICOM format), perform format conversion and standardization processing on the multi-contrast magnetic resonance images in the first format to obtain an image file to be processed, and the image file to be processed includes: brain structure image (T1W image) data and cerebrovascular image (TOF image) data.

[0028] In the embodiments of the present invention, the contrast magnetic resonance images include: T1W and TOF, specifically as Figure 3 shown, which shows a sample image of the multi-contrast brain magnetic resonance scanning sequence, where part (a) is the brain structure image T1W and part (b) is the cerebrovascular image TOF.

[0029] In the embodiments of the present invention, the DICOM (Digital Imaging and Communications in Medicine) format is mainly used for storing and transmitting medical images. It has the characteristics of rich metadata, including patient information, examination parameters, device information, etc., which is convenient for clinical diagnosis and image processing; hierarchical structure: adopts Information Object Definition (IOD) and modular structure, supports multiple image types and compression methods; multi-file slices: usually each slice generates a file, which is convenient for displaying the structures of different layers and is a series of two-dimensional layers.

[0030] In practical applications, the original first format file can be converted into a more common second format (Nifti format).

[0031] The Nifti format is also used to store medical imaging data and is suitable for scientific research and analysis. It has a simple structure, and the original data is stored as a three-dimensional image, which is easy to process and analyze.

[0032] In practical applications, it is only necessary to collect and export the DICOM format images from the machine, and then the subsequent parameter extraction process can be completed based on the technical solution provided by the embodiments of the present invention.

[0033] In an alternative embodiment, the step S101 includes the following sub-steps: S1011, Use the dcm2niix conversion tool to convert the original DICOM format file into the Nifti format.

[0034] S1012, Use the median value of the file to normalize the obtained Nifti file.

[0035] As Figure 2 shown, the multi-contrast vascular imaging automatic parameter extraction algorithm provided by the embodiments of the present invention mainly includes four stages: file management, registration preprocessing, structured processing flow, and physiological parameter generation. In the file management part, the dcm2niix conversion tool can be used for file format conversion, converting the original Dicom format file into a more general Nifti format, and synchronously generating a Json file containing rich information. After completing the file format conversion and naming standardization, data standardization processing is performed on the obtained Nifti file. The specific operation is: perform normalization processing using the median value of the file to effectively prevent the possible offset phenomenon of the data during the processing. This step is of great significance for ensuring the accuracy and reliability of the data analysis results.

[0036] S102, Use a general brain imaging processing tool to perform registration preprocessing operations on the to-be-processed image file, obtain a brain mask in the TOF original space, perform TOF skull stripping operation, and standard brain space registration operation.

[0037] In the embodiments of the present invention, the general brain imaging processing tool can be the FSL tool.

[0038] In the embodiments of the present invention, through the registration preprocessing operation, based on images with two contrasts (TOF images and T1W images), the TOF images can be transformed into the T1W space, and then configured into the standard brain space. Then, in this standard space, the TOF skull stripping operation can be uniformly performed. In the TOF images, the presence of the skull may interfere with the segmentation of blood vessels. Therefore, the skull stripping operation needs to be performed first. In other brain blood vessel image evaluation algorithms in the related art, the skull stripping operation can only be achieved based on deep learning algorithms or manual parameter adjustment. However, in the embodiments of the present invention, by uniformly configuring the TOF images into the standard brain space and then obtaining the transformation matrix, the skull can be automatically removed. In addition, in the embodiments of the present invention, after performing the skull stripping operation on the TOF images, the optimal segmentation point can also be directly calculated based on threshold segmentation, so as to automatically extract the blood vessel segmentation images, and no manual intervention is required during the subsequent automatic annotation and topological structure generation processes.

[0039] In an alternative embodiment, the registration preprocessing operation includes: obtaining a non-linear transformation file from the T1W image in the original space to the standard brain template, obtaining a linear transformation file for linearly registering the TOF image in the original space to the T1W space in the original space, and using the non-linear transformation file in combination with the linear transformation file to transform the TOF post-processing parameter file.

[0040] After completing the file management, this patent uses the FSL tool to perform the registration preprocessing operation on all Nifti files. In short, in the embodiments of the present invention, the registration preprocessing operation includes: Using tools in the FSL tool to perform operations such as region of interest cropping, skull stripping, linear registration, and non-linear registration on the T1W image in the original space, and obtaining the transformation field from the original scan space to the 0.5mm standard brain template.

[0041] Perform the registration operation on the TOF images. First, linearly register the TOF image in the original space to the T1W image in the original space. Using the previously obtained non-linear transformation file from the T1W original space to the standard brain space, in combination with the linear transformation file from the TOF original space to the T1W space, transform the standard brain mask to obtain the brain mask in the TOF original space, and perform the TOF skull stripping operation.

[0042] Adopt the same transformation method (TOF→T1W→standard brain), as Figure 2 shown in the registration preprocessing part, to transform the blood vessel segmentation images, blood vessel radius distribution images, etc. generated in the subsequent steps into the standard brain space.

[0043] The standard brain template is a standard brain template widely used in neuroimaging research. It is based on the brain MRI scan data of multiple normal individuals and undergoes spatial normalization processing through non-linear deformation technology to create an average and standardized brain structure model. The advantage of the standard brain template is that it provides a unified coordinate system, enabling effective comparison and analysis between different studies. In the embodiments of the present invention, a standard brain template with a resolution of 0.5 mm is adopted because this template has high spatial accuracy and can better capture the details of the brain structure. By registering individual images to the standard brain template, the embodiments of the present invention can not only achieve the spatial normalization of image data but also promote the integration and analysis of brain image data across studies and populations, which is of great significance for revealing the general laws of brain structure and function.

[0044] Specifically, the TOF image (TOF_orig) is linearly registered to the T1W structure (T1W_orig) and then non-linearly registered and mapped to the standard brain standard space. The process is divided into three parts: Linear registration from TOF_orig to T1W_orig; non-linear registration from T1W_orig to the standard brain space; final result verification and masking.

[0045] Specifically, the linear registration (flirt) process from TOF_orig (TOF image) to T1W_orig (T1W image) includes: Use the flirt tool in FSL for transformation to align the TOF image to the T1W structure (TOF_orig_to_T1W_orig). Generate a transformation matrix to record the spatial mapping relationship. Combine multiple transformation matrices through convert_xfm for multi-step registration or cross-modal alignment. Mask generation and application (mask): Perform brain extraction on the T1W image (T1W_roi_brain) to generate a brain mask (T1W_brain_mask) to exclude non-brain tissues such as the skull. Apply the mask to the TOF image (TOF_orig_brain) to ensure that subsequent analysis only focuses on the brain region.

[0046] The non-linear registration process from T1W_orig (T1W image) to the standard brain space includes: Use maskFOV for rough cropping of the region of interest, use bet to remove the skull region, use standard_space_roi to further reduce the region of interest to focus on a specific region in the standard space, use linear registration (flirt) to obtain a transformation file for preliminary registration to a specific region in the standard space, and use non-linear registration (fnirt) to register the T1W image to the standard brain template.

[0047] The convertwrap combines the linear transformation field from TOF_orig to T1W_orig and the non-linear transformation field from T1W_orig to the standard brain standard space, and maps the TOF image to the standard brain space.

[0048] S103, perform structured processing on the TOF image data in the original space to obtain a vascular segmentation image; convert the vascular segmentation image into a three-dimensional skeleton, convert the three-dimensional skeleton into a graph structure, generate a topological structure file containing nodes and edges, and the topological structure file characterizes the connectivity and original structure of the skeleton. Determine the blood vessel length, curvature, and surface area based on the topological structure file; obtain the segmentation results of the bilateral internal carotid arteries, bilateral middle cerebral arteries, bilateral anterior cerebral arteries, bilateral posterior cerebral arteries, and basilar artery based on the vascular segmentation image and a pre-trained classification network.

[0049] In the embodiment of the present invention, structured processing is performed on the TOF brain image data in the original space to obtain a file for subsequent parameter extraction.

[0050] In an alternative embodiment, performing structured processing on the TOF image data in the original space to obtain a vascular segmentation image includes: S1, analyze the gray histogram of the TOF image data to determine the maximum peak. After this maximum peak, the frequency value rapidly decreases as the brightness increases and finally approaches the x-axis.

[0051] S2, select two points at the maximum peak and the maximum brightness frequency and draw a straight line, and use the point farthest from this straight line in the gray curve as the segmentation threshold.

[0052] S3, calculate the volume of all connected regions based on the segmentation threshold, and retain the connected regions with a volume greater than 95% of the total volume to obtain a vascular segmentation image.

[0053] In the embodiment of the present invention, by analyzing the gray histogram of the image, some peaks can be observed, and one of the peaks is the most significant. After this maximum peak, the frequency value rapidly decreases as the brightness increases and finally approaches the x-axis. Select two points at the maximum peak and the maximum brightness frequency and draw a straight line, and the point farthest from this straight line in the gray curve is the segmentation threshold. Subsequently, calculate the volume of all connected regions, and retain the connected regions with a volume greater than 95% of the total volume, which is the vascular segmentation result, to obtain a vascular segmentation image.

[0054] Subsequently, in the embodiments of the present invention, the Skeleton3D function in MATLAB can be used to convert the segmented image into a skeleton, which is used as the centerline of the blood vessel. Then, the Skel2Graph3D function is used to convert the three-dimensional skeleton into a graph structure, generating a topological structure file containing nodes and edges, representing the connectivity of the skeleton and reflecting its original structure for subsequent analysis. The algorithm will traverse all edges, obtain the blood vessel segmentation blocks around each node on the edge, and calculate the radius of the largest inscribed sphere in this area as the blood vessel radius corresponding to this skeleton node. In the embodiments of the present invention, the blood vessel volume is defined as the total volume of the blood vessel segmentation results within the region of interest (which can be determined according to the segmentation results of the bilateral internal carotid arteries, bilateral middle cerebral arteries, bilateral anterior cerebral arteries, bilateral posterior cerebral arteries, and basilar artery). For the calculation of blood vessel length, curvature, and surface area, first obtain the blood vessel skeleton to be calculated and its topological connection structure, and calculate the blood vessel length, curvature, and surface area point by point according to the position information and radius information of the nodes on each edge.

[0055] In an alternative embodiment, determining the blood vessel length, curvature, and surface area based on the topological structure file includes: S11, traverse all edges, obtain the blood vessel segmentation blocks around each skeleton node on the edge, and calculate the radius of the largest inscribed sphere around this skeleton node as the blood vessel radius corresponding to this skeleton node.

[0056] S12, obtain the blood vessel skeleton to be calculated and its topological connection structure, and determine the blood vessel length, curvature, and surface area point by point according to the position information and radius information of the skeleton nodes on each edge.

[0057] In an alternative embodiment, the following formulas are used to calculate the blood vessel length, curvature, and surface area.

[0058] ; ; ; Where, ( x i ，y i ，z i ) and ( x i+! ，y i+1 ，z i+1 ) respectively represent the position information of two adjacent points on the edge to be calculated; length represents the blood vessel length, surface represents the blood vessel surface area, radius represents the blood vessel radius, tortuosity represents the curvature,norm Represents the straight-line distance between the starting point and the ending point of a blood vessel.

[0059] In practical applications, after determining the segmentation results of the bilateral internal carotid arteries, bilateral middle cerebral arteries, bilateral anterior cerebral arteries, bilateral posterior cerebral arteries, and basilar artery, the blood vessel radius, length, curvature, surface area, and volume of different branch arteries can be determined respectively according to the topological structure and the above formula.

[0060] In the embodiments of the present invention, the arterial automatic labeling is completed by a pre-trained nnU-net network, and the pre-trained nnU-net network can output the segmentation results of the bilateral internal carotid arteries, bilateral middle cerebral arteries, bilateral anterior cerebral arteries, bilateral posterior cerebral arteries, and basilar artery according to the input blood vessel segmentation file.

[0061] In the embodiments of the present invention, the segmentation neural network (nnU-net) adopted is a type of deep learning network, which is a publicly available and mature segmentation algorithm with good segmentation effects on multiple datasets. This model can segment 2D images and can also segment 3D images.

[0062] As Figure 2 shown, in the embodiments of the present invention, in the structured processing flow, the blood vessel segmentation, centerline extraction, blood vessel automatic annotation, and topological structure generation processes are completed.

[0063] S104, determine the blood vessel radius, length, curvature, surface area, and volume of different branch arteries based on the segmentation results of the bilateral internal carotid arteries, bilateral middle cerebral arteries, bilateral anterior cerebral arteries, bilateral posterior cerebral arteries, and basilar artery; for the statistical atlas in the standard brain space, perform radius result allocation on the blood vessel segmentation image, search for the centerline closest to a point in the mask, obtain the radius value of this centerline point, and obtain the blood vessel radius file; based on the registration preprocessing operation, transform the blood vessel radius file and the blood vessel segmentation image into the standard brain template to generate a statistical atlas of blood vessel radius distribution.

[0064] As Figure 2 shown, in the physiological parameter generation stage, the blood vessel radius, length, curvature, surface area, and volume of different branch arteries will be calculated according to the arterial automatic labeling results.

[0065] For the statistical atlas in the standard brain space, first perform radius result allocation on the blood vessel segmentation mask, search for the centerline closest to a point in the mask, and obtain the radius value of this centerline point. Finally, transform this blood vessel radius file and the blood vessel segmentation file into the standard brain space, and they can be quantitatively compared with the average atlas in the database.

[0066] In the embodiments of the present invention, in the process of calculating the statistical atlas, it is first necessary to assign a corresponding radius value to each part of the blood vessel. For the calculation of this radius value, the embodiments of the present invention propose that for the blood vessel segmentation mask, the radius value of the centerline point closest to it is obtained. By adopting this strategy, the entire blood vessel area can be fully filled with radii, so that each point on the blood vessel has a radius assignment. Then, the blood vessel segmentation file with the assigned blood vessel radius is transformed into this standard brain space, and the average value is obtained, which can be quantitatively compared with the average atlas in the database. Specifically, the radius possessed by each position of the blood vessel image can be quantitatively compared with the radius possessed by each position of the blood vessel image in the average atlas.

[0067] The blood vessel analysis method based on multi-contrast magnetic resonance images proposed in the embodiments of the present invention can realize the automatic extraction of parameters of brain blood vessel images, has more stable performance than traditional algorithms, and can batch process large-scale data. Moreover, the embodiments of the present invention can provide various parameters including the whole cerebral blood vessels and branch arteries based on multi-contrast images, and provide the quantitative parameter distribution under the standard brain to realize the comprehensive evaluation of blood vessel images.

[0068] Based on the same inventive concept, the present invention also provides a blood vessel analysis system based on multi-contrast magnetic resonance images, as Figure 4 shown. The blood vessel analysis system 400 based on multi-contrast magnetic resonance images includes: A file management module 401, configured to obtain multi-contrast magnetic resonance images in DICOM format, perform format conversion and standardization processing on the multi-contrast magnetic resonance images in DICOM format to obtain an image file to be processed, and the image file to be processed includes: T1W image data and TOF image data; A registration preprocessing module 402, configured to perform registration preprocessing operations on the image file to be processed using a general brain image processing tool, obtain a brain mask in the TOF original space, perform TOF skull stripping operation, and standard brain space registration operation; A structuring processing module 403, configured to perform structuring processing on the TOF image data in the original space to obtain a blood vessel segmentation image; convert the blood vessel segmentation image into a three-dimensional skeleton, convert the three-dimensional skeleton into a graph structure, generate a topological structure file including nodes and edges, and the topological structure file characterizes the connectivity and original structure of the skeleton. Determine the blood vessel length, curvature, and surface area based on the topological structure file; obtain the segmentation results of bilateral internal carotid arteries, bilateral middle cerebral arteries, bilateral anterior cerebral arteries, bilateral posterior cerebral arteries, and basilar artery based on the blood vessel segmentation image and a pre-trained classification network; The physiological parameter generation module 404 is configured to determine the blood vessel radius, length, curvature, surface area, and volume of different branch arteries based on the segmentation results of the bilateral internal carotid arteries, bilateral middle cerebral arteries, bilateral anterior cerebral arteries, bilateral posterior cerebral arteries, and basilar artery; for the statistical atlas in the standard brain space, allocate the radius results to the blood vessel segmentation image, search for the centerline closest to a point in the distance mask, obtain the radius value of the centerline point, and obtain the blood vessel radius file; based on the registration preprocessing operation, transform the blood vessel radius file and the blood vessel segmentation image to the standard brain template to generate a statistical atlas of blood vessel radius distribution.

[0069] Optionally, the registration preprocessing module 402 is configured to: Obtain the non-linear transformation file from the T1W image in the original space to the standard brain template, obtain the linear transformation file for linearly registering the TOF image in the original space to the T1W space in the original space, and use the non-linear transformation file in combination with the linear transformation file to transform the TOF post-processing parameter file.

[0070] Optionally, the structuring processing module 403 is configured to: Analyze the gray histogram of the TOF image data to determine the maximum peak value. After this maximum peak value, the frequency value rapidly decreases as the brightness increases and finally approaches the x-axis; Select two points at the maximum peak value and the maximum brightness frequency and draw a straight line, and use the point farthest from this straight line in the gray curve as the segmentation threshold; Calculate the volume of all connected regions based on the segmentation threshold, and retain the connected regions with a volume greater than 95% of the total volume to obtain the blood vessel segmentation image.

[0071] Optionally, the structuring processing module 403 is further configured to: Traverse all edges, obtain the blood vessel segmentation chunks around each skeleton node on the edge, and calculate the radius of the largest inscribed sphere around the skeleton node as the blood vessel radius corresponding to the skeleton node; Obtain the blood vessel skeleton to be calculated and its topological connection structure, and determine the blood vessel length, curvature, and surface area point by point according to the position information and radius information of the skeleton nodes on each edge.

[0072] Optionally, determining the blood vessel length, curvature, and surface area point by point according to the position information and radius information of the skeleton nodes on each edge includes: ; ; ; Wherein, ( x i ，y i，z i ), and ( x i+1 ，y i+1 ，z i+1 ) respectively represent the position information of two adjacent points on the side to be calculated; length represents the blood vessel length, surface represents the blood vessel surface area, radius represents the blood vessel radius, tortuosity represents the curvature, norm represents the straight-line distance between the starting point and the ending point of the blood vessel.

[0073] Optionally, the file management module 501 is used for: Using the dcm2niix conversion tool to convert the original Dicom format file into Nifti format; Normalizing the obtained Nifti file by using the median value of the file.

[0074] Based on the same inventive concept, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes, it implements the steps in the method for analyzing blood vessels based on multi-contrast magnetic resonance images as described in any of the above embodiments.

[0075] Based on the same inventive concept, the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the steps in the method for analyzing blood vessels based on multi-contrast magnetic resonance images as described in any of the above embodiments.

[0076] Based on the same inventive concept, the present invention provides a computer program product, including a computer program / instructions. When the computer program / instructions are executed by a processor, it implements the steps in the method for analyzing blood vessels based on multi-contrast magnetic resonance images as described in any of the above embodiments.

[0077] Each embodiment in this specification is described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The same or similar parts among the embodiments can be referred to each other.

[0078] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a device, or a computer program product. Therefore, the present invention can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes.

[0079] The present invention is described with reference to the flowcharts and / or block diagrams of methods, terminal devices (apparatuses), and computer program products according to the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and combinations of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable terminal devices to produce a machine, such that the instructions executed by the processor of the computer or other programmable terminal devices generate means for implementing the functions specified in one Figure 1 flow or multiple flows and / or blocks Figure 1 block or multiple blocks.

[0080] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable terminal devices to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including instruction means that implement the functions specified in one Figure 1 flow or multiple flows and / or blocks Figure 1 block or multiple blocks.

[0081] These computer program instructions can also be loaded onto a computer or other programmable terminal devices, such that a series of operation steps are executed on the computer or other programmable terminal devices to produce a computer-implemented process, and thus the instructions executed on the computer or other programmable terminal devices provide steps for implementing the functions specified in one Figure 1 flow or multiple flows and / or blocks Figure 1 block or multiple blocks.

[0082] Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications to these embodiments once they learn the basic creative concepts. Therefore, the appended claims are intended to be construed as including the preferred embodiments and all changes and modifications falling within the scope of the present invention.

[0083] Finally, it should also be noted that in the present invention, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or terminal device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or terminal device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or terminal device comprising the element.

[0084] The above has introduced in detail a vascular analysis method based on multi-contrast magnetic resonance images provided by the present invention. Specific examples are used in the present invention to elaborate on the principle and implementation manner of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A blood vessel analysis method based on multi-contrast magnetic resonance imaging, characterized in that: The method comprises: Acquire a multi-contrast magnetic resonance image in DICOM format, perform format conversion and standardization processing on the multi-contrast magnetic resonance image in DICOM format, and obtain an image file to be processed, wherein the image file to be processed includes: T1W image data and TOF image data; Use a general brain image processing tool to perform a registration preprocessing operation on the image file to be processed, obtain a brain mask in the TOF original space, perform a TOF skull removal operation, and a standard brain space registration operation; Performing structured processing on the TOF image data in the original space to obtain a blood vessel segmentation image; converting the blood vessel segmentation image into a three-dimensional skeleton, converting the three-dimensional skeleton into a graph structure, generating a topological structure file containing nodes and edges, wherein the topological structure file characterizes the connectivity and original structure of the skeleton, and determining the blood vessel length, curvature, and surface area based on the topological structure file; obtaining the segmentation results of the bilateral internal carotid arteries, bilateral middle cerebral arteries, bilateral anterior cerebral arteries, bilateral posterior cerebral arteries, and basilar arteries based on the blood vessel segmentation image and a pre-trained classification network; Based on the segmentation results of the bilateral internal carotid arteries, bilateral middle cerebral arteries, bilateral anterior cerebral arteries, bilateral posterior cerebral arteries, and basilar arteries, the vascular radius, length, curvature, surface area, and volume of different branch arteries are determined; with respect to the statistical atlas under the standard brain space, the radius results are assigned to the vascular segmentation image, the center line closest to a point in the mask is searched, the radius value of the center line point is obtained, and a vascular radius file is obtained; based on the registration preprocessing operation, the vascular radius file and the vascular segmentation image are transformed into a standard brain template to generate a statistical atlas of vascular radius distribution.

2. The blood vessel analysis method based on multi-contrast magnetic resonance imaging according to claim 1, characterized in that: The registration preprocessing operation includes: A nonlinear transformation file of the original space T1W image to the standard brain template is obtained, and a linear transformation file of the original space TOF image linearly aligned to the original space T1W space is obtained. The TOF post-processing parameter file is transformed by combining the nonlinear transformation file with the linear transformation file.

3. The blood vessel analysis method based on multi-contrast magnetic resonance imaging according to claim 1, characterized in that: The TOF image data in the original space is structured to obtain a vascular segmentation image, including: The grayscale histogram of the TOF image data is analyzed to determine the maximum peak value. After this maximum peak value, the frequency value decreases rapidly as the brightness increases and eventually approaches the x-axis. Select the two points of the maximum peak and the maximum brightness frequency as a straight line, and use the point farthest from the straight line in the grayscale curve as the segmentation threshold; The volumes of all connected domains are calculated based on the segmentation threshold, and connected domains larger than 95% of the total volume are retained to obtain a blood vessel segmentation image.

4. The blood vessel analysis method based on multi-contrast magnetic resonance imaging according to claim 1, characterized in that: Determining the length, curvature, and surface area of ​​the blood vessel based on the topological structure file includes: Traverse all edges, obtain the blood vessel segmentation blocks around each skeleton node on the edge, calculate the radius of the maximum inscribed sphere around the skeleton node, and use it as the corresponding blood vessel radius on the skeleton node; The blood vessel skeleton and its topological connection structure to be calculated are obtained, and the blood vessel length, curvature, and surface area are determined point by point based on the position information and radius information of the skeleton nodes on each edge.

5. The blood vessel analysis method based on multi-contrast magnetic resonance imaging according to claim 4, characterized in that: The length, curvature, and surface area of ​​the blood vessel are determined point by point based on the position information and radius information of the skeleton nodes on each edge, including: ； ； ； in,( x i ，y i ，z i )and( x i+1 ，y i+1 ，z i+1 ) represent the position information of two adjacent points on the edge that need to be calculated; length represents the length of the blood vessel, surface represents the vascular surface area, radius represents the vessel radius, tortuosity represents the curvature, norm Represents the straight-line distance between the start and end points of a vessel.

6. The blood vessel analysis method based on multi-contrast magnetic resonance imaging according to claim 1, characterized in that: The multi-contrast magnetic resonance image in the DICOM format is format converted and standardized to obtain an image file to be processed, including: Use the dcm2niix conversion tool to convert the original Dicom format file to Nifti format; The obtained Nifti files were normalized using the median value of the files.

7. A blood vessel analysis system based on multi-contrast magnetic resonance imaging, characterized in that: The blood vessel analysis system based on multi-contrast magnetic resonance imaging includes: A file management module is used to obtain multi-contrast magnetic resonance images in DICOM format, perform format conversion and standardization processing on the multi-contrast magnetic resonance images in DICOM format, and obtain image files to be processed, wherein the image files to be processed include: T1W image data and TOF image data; A registration preprocessing module, used to perform registration preprocessing operations on the image file to be processed using a general brain image processing tool, obtain a brain mask in the TOF original space, perform a TOF skull removal operation, and a standard brain space registration operation; A structured processing module is used to perform structured processing on the TOF image data in the original space to obtain a blood vessel segmentation image; convert the blood vessel segmentation image into a three-dimensional skeleton, convert the three-dimensional skeleton into a graph structure, generate a topological structure file containing nodes and edges, the topological structure file characterizes the connectivity and original structure of the skeleton, and determines the blood vessel length, curvature, and surface area based on the topological structure file; obtain the segmentation results of the bilateral internal carotid arteries, bilateral middle cerebral arteries, bilateral anterior cerebral arteries, bilateral posterior cerebral arteries, and basilar arteries based on the blood vessel segmentation image and a pre-trained classification network; The physiological parameter generation module is used to determine the vascular radius, length, curvature, surface area and volume of different branch arteries based on the segmentation results of the bilateral internal carotid arteries, bilateral middle cerebral arteries, bilateral anterior cerebral arteries, bilateral posterior cerebral arteries and basilar arteries; assign radius results to the vascular segmentation image according to the statistical atlas under the standard brain space, search for the center line closest to a point in the mask, obtain the radius value of the center line point, and obtain a vascular radius file; based on the registration preprocessing operation, transform the vascular radius file and the vascular segmentation image to the standard brain space template to generate a statistical atlas of vascular radius distribution.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the blood vessel analysis method based on multi-contrast magnetic resonance imaging according to any one of claims 1 to 6 is implemented.

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 blood vessel analysis method based on multi-contrast magnetic resonance imaging described in any one of claims 1 to 6 is implemented.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instruction is executed by a processor, the steps in the blood vessel analysis method based on multi-contrast magnetic resonance imaging described in any one of claims 1 to 6 are implemented.

Images