Method and device for processing magnetic resonance image sequence, computer equipment and medium

Abstract

The invention relates to the technical field of medical image processing, and discloses a magnetic resonance image sequence processing method and device, computer equipment and a medium. The method comprises the following steps: acquiring a plurality of magnetic resonance image sequences corresponding to a target blood vessel region and pre-labeled aneurysm information; cutting the plurality of registered magnetic resonance image sequences according to the aneurysm information to obtain a target image; determining a blood vessel lumen segmentation mask corresponding to the target image; according to the aneurysm information, constructing an initial aneurysm wall inner layer surface of the aneurysm; reconstructing an evolved neoplasm wall inner layer surface according to the characteristics of the initial neoplasm wall inner layer surface; performing three-dimensional reconstruction of the aneurysm according to the evolution tumor wall inner layer surface and the blood vessel lumen segmentation mask to obtain a target tumor wall inner layer surface; and analyzing the signal intensity mapping of the inner layer surface of the target aneurysm wall, and determining a quantitative index of the aneurysm. By implementing the technical scheme of the invention, accurate segmentation, reconstruction and quantitative analysis of the aneurysm in the magnetic resonance image sequence can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of medical image processing, and in particular to a method, device, computer equipment and medium for processing a magnetic resonance image sequence. Background Art

[0002] Magnetic resonance imaging (MRI) is a non-invasive medical imaging technique that uses powerful magnetic fields and harmless radio waves to create detailed images of the inside of the human body. MRI is able to generate high-resolution cross-sectional images showing structures and organs inside the body, such as the brain, spinal cord, joints, and internal organs. However, in current MRI processing, there may be positional deviations or image quality differences in MRI image sequences acquired at different time points or in different imaging modalities, which can affect the accurate detection and localization of aneurysms; secondly, due to image noise, the complexity of vascular morphology, and changes in the structure around the aneurysm, traditional automatic segmentation methods may have difficulty accurately dividing vascular contours; thirdly, traditional manual quantitative analysis may be subjective and inconsistent, while automated quantitative analysis can improve the objectivity and accuracy of the analysis. Summary of the invention

[0003] In view of this, the present invention provides a method, apparatus, computer equipment and medium for processing a magnetic resonance image sequence to solve the problem of how to improve the image consistency, accurate blood vessel segmentation and objective quantitative analysis of magnetic resonance imaging.

[0004] In a first aspect, the present invention provides a method for processing a magnetic resonance image sequence, comprising: obtaining a plurality of magnetic resonance image sequences corresponding to a target vascular region, and obtaining aneurysm information pre-annotated in any magnetic resonance image sequence; wherein the plurality of magnetic resonance image sequences are a plurality of images obtained during a magnetic resonance imaging process, and the aneurysm information is used to characterize the aneurysm position in the target vascular region; aligning the plurality of magnetic resonance image sequences to the same sequence space, and cropping the aligned plurality of magnetic resonance image sequences according to the aneurysm information to obtain a target image corresponding to the aneurysm; wherein the same sequence space is a space corresponding to any sequence in the plurality of magnetic resonance image sequences, and the target image is used to characterize the aneurysm position in the target vascular region. aneurysm and the first vascular structure around it; processing the target image to determine the vascular lumen segmentation mask corresponding to the target image, the vascular lumen segmentation mask is used to characterize the contours of all vascular lumens in the target image; extracting the aneurysm information included in the target image, and constructing the initial aneurysm wall inner surface of the aneurysm according to the aneurysm information; extracting features of the initial aneurysm wall inner surface, reconstructing the initial aneurysm wall inner surface according to the extracted features, and obtaining the evolved aneurysm wall inner surface; and performing three-dimensional reconstruction of the aneurysm according to the evolved aneurysm wall inner surface and the vascular lumen segmentation mask to obtain the target aneurysm wall inner surface; analyzing the signal intensity mapping of the target aneurysm wall inner surface to determine the quantitative index of the aneurysm.

[0005] The method for processing magnetic resonance image sequences provided by the embodiment of the present invention obtains and utilizes multiple magnetic resonance image sequences, comprehensively displays the target vascular area and the aneurysm information therein, can provide more comprehensive and multi-angle information, and is helpful for more accurate analysis and diagnosis of aneurysms. By accurately registering multiple magnetic resonance image sequences and cropping them according to the pre-marked aneurysm information, a target image corresponding to the aneurysm is obtained. This ensures that the spatial position and anatomical structure are kept consistent between different images, providing a reliable basis for subsequent analysis and processing. The information of the aneurysm in the target image is extracted and analyzed, and the initial inner surface of the aneurysm wall is constructed based on this information. Through feature extraction and reconstruction technology, the inner surface of the evolving aneurysm wall can be presented more finely, which helps doctors to have a deeper understanding of the morphology and structure of the aneurysm. The three-dimensional reconstruction of the aneurysm is performed based on the inner surface of the evolving aneurysm wall and the segmentation mask of the vascular lumen, which not only provides the spatial information of the aneurysm, but also can quantitatively analyze the characteristics and signal intensity mapping of the aneurysm, thereby providing a scientific basis to support clinical decision-making.

[0006] In an optional embodiment, the multiple magnetic resonance image sequences include a first magnetic resonance image sequence, a second magnetic resonance image sequence and a third magnetic resonance image sequence; the first magnetic resonance image sequence is used to display a second vascular structure corresponding to a target vascular region, the second magnetic resonance image sequence is used to display the anatomical structure and signal intensity difference of tissue included in the target vascular region, and the third magnetic resonance image sequence is used to display the contrast enhancement effect of an aneurysm corresponding to the target vascular region after adding a contrast agent; the multiple magnetic resonance image sequences are aligned to the same sequence space, including: preprocessing the first magnetic resonance image sequence, the second magnetic resonance image sequence and the third magnetic resonance image sequence; and aligning the preprocessed first magnetic resonance image sequence, the second magnetic resonance image sequence and the third magnetic resonance image sequence according to a preset medical image registration library, so as to align the first magnetic resonance image sequence and the second magnetic resonance image sequence to the third magnetic resonance image sequence.

[0007] The method for processing a magnetic resonance image sequence provided in an embodiment of the present invention utilizes three different types of magnetic resonance image sequences. Through these different images, the target vascular area and the aneurysm information therein are comprehensively displayed, which helps doctors to fully understand the condition. Multiple magnetic resonance image sequences are accurately registered, which includes preprocessing each type of image and registering according to a preset medical image registration library, which can ensure that the spatial position and anatomical structure between different images are consistent, thereby providing a reliable basis for subsequent analysis and diagnosis. The first magnetic resonance image sequence, the second magnetic resonance image sequence and the aneurysm information are registered to the third magnetic resonance image sequence. The third magnetic resonance image sequence is usually used to display the distribution of tissue contrast agents and the enhancement effect of aneurysms. By applying the registered information to this sequence, the location and characteristics of the aneurysm can be observed more clearly.

[0008] In an optional embodiment, the registered multiple magnetic resonance image sequences are cropped according to the aneurysm information to obtain a target image corresponding to the aneurysm, including: cropping the registered first magnetic resonance image sequence, the second magnetic resonance image sequence, and the third magnetic resonance image sequence according to the aneurysm information to obtain a first target sequence corresponding to the first magnetic resonance image sequence, a second target sequence corresponding to the second magnetic resonance image sequence, and a third target sequence corresponding to the third magnetic resonance image sequence.

[0009] The method for processing a magnetic resonance image sequence provided in an embodiment of the present invention crops multiple registered magnetic resonance image sequences according to aneurysm information to obtain a target image corresponding to the aneurysm. This precise cropping operation can ensure that the extracted image area only contains the aneurysm portion of interest, excluding other irrelevant structures and backgrounds, which is helpful for subsequent analysis and diagnosis. The first magnetic resonance image sequence, the second magnetic resonance image sequence, and the third magnetic resonance image sequence are cropped respectively to obtain the corresponding target sequences, thereby retaining the multimodal information of each image sequence, such as structure, anatomical features, and enhancement effects, and being able to comprehensively display the various characteristics of the aneurysm. The target sequence obtained after cropping can be directly used for subsequent computer-aided diagnosis or other processing tasks. These image sequences have been located in the aneurysm region of interest and are aligned in spatial position, making further quantitative analysis or comparative analysis more convenient and accurate.

[0010] In an optional embodiment, the target image is processed to determine a blood vessel lumen segmentation mask corresponding to the target image, including: processing a first target sequence to determine a white blood mask, the white blood mask is used to characterize the position and shape of all blood vessels in the target image; performing Gaussian filter edge enhancement processing on the second target sequence and the third target sequence, and synthesizing the second target sequence and the third target sequence after the Gaussian filter edge enhancement processing to generate a synthetic image; obtaining a preset pixel threshold, the pixel threshold is used to distinguish the blood vessel lumen in the synthetic image; processing the synthetic image according to the pixel threshold to obtain a binary mask image; and extracting the blood vessel lumen segmentation mask from the mask image according to the white blood mask and the mask image.

[0011] The processing method of the magnetic resonance image sequence provided by the embodiment of the present invention processes the first target sequence to determine the white blood mask, which is used to accurately locate the position and shape of all blood vessels in the target image, and provide basic information for the subsequent segmentation of the blood vessel lumen. The edge enhancement processing of the second target sequence and the third target sequence by Gaussian filtering can enhance the edge features of the blood vessels, make the contours of the blood vessels clearer, and facilitate the subsequent segmentation and analysis of the blood vessels. The second target sequence and the third target sequence after the edge enhancement processing by Gaussian filtering are synthesized to generate a synthetic image. The synthetic image integrates multimodal information, can provide more comprehensive visual information, and helps to distinguish the blood vessel lumen more accurately. Obtain a preset pixel threshold for distinguishing the blood vessel lumen in the synthetic image. By setting the threshold, the blood vessel part in the image can be separated from the background to form a preliminary binary mask image. According to the white blood mask and the mask image, the blood vessel lumen segmentation mask is extracted from the mask image, which can accurately segment the position of the blood vessel and provide accurate data support for subsequent quantitative analysis and diagnosis.

[0012] In an optional embodiment, an initial inner surface of the aneurysm wall is constructed based on the aneurysm information, including: converting the aneurysm information into a three-dimensional convex hull; converting the three-dimensional convex hull into a first inner surface of the aneurysm wall; smoothing the first inner surface of the aneurysm wall to obtain a second inner surface of the aneurysm wall; isotropically resampling the second inner surface of the aneurysm wall to obtain a three-dimensional voxel data set; and constructing the initial inner surface of the aneurysm wall based on the three-dimensional voxel data set.

[0013] The processing method of the magnetic resonance image sequence provided by the embodiment of the present invention is an effective way to convert the aneurysm information into a three-dimensional convex hull, which can convert the input aneurysm data into a basic representation of a geometric shape. The convex hull representation method is simple and clear, and can better reflect the overall shape and contour of the aneurysm. The three-dimensional convex hull is converted into the inner surface of the first aneurysm wall, so that the geometric features of the convex hull can be retained and effectively converted into a more refined and realistic surface representation. The smoothing of the inner surface of the first aneurysm wall is to eliminate possible noise and irregularities, make the surface more continuous and natural, help improve the quality of the results of subsequent processing, and ensure that the constructed inner surface of the aneurysm wall has good geometric characteristics. Through isotropic voxel resampling, the smoothed inner surface of the second aneurysm wall can be converted into a three-dimensional voxel data set, which can provide an accurate description of the specific position in space and provide a basis for subsequent analysis and visualization. According to the three-dimensional voxel data set, the initial inner surface of the aneurysm wall is constructed, and the voxel data is converted into a geometric surface, which provides an accurate shape description for further analysis and simulation, and is helpful for medical imaging diagnosis and treatment planning.

[0014] In an optional embodiment, features are extracted from an initial inner surface of the tumor wall, and the initial inner surface of the tumor wall is reconstructed according to the extracted features to obtain an evolved inner surface of the tumor wall, including: extracting first position information and edge information of a first vertex in the initial inner surface of the tumor wall, and processing the first position information and edge information to obtain a first normal vector and an evolution direction of the first vertex; performing trilinear interpolation on a composite image to obtain an interpolated image function; and calculating a directional derivative in the evolution direction according to the interpolated image function and the first normal vector, and determining a maximum value of the directional derivative within a preset distance around the first vertex; obtaining a preset target displacement distance, and determining a value obtained by adding the target displacement distance to the maximum value as a target movement amount corresponding to the first vertex; updating the first position information according to the target movement amount to obtain the target position information corresponding to the first vertex; and reconstructing the initial inner surface of the tumor wall according to the target position information until the initial inner surface of the tumor wall converges to obtain the evolved inner surface of the tumor wall.

[0015] The processing method of the magnetic resonance image sequence provided by the embodiment of the present invention can extract the key feature information of the surface by extracting the features of the initial inner surface of the tumor wall. These feature extraction steps help to accurately capture the shape and structural features of the tumor wall. Based on the extracted feature information, the inner surface of the tumor wall can be reconstructed by calculating the evolution direction and the target movement amount, that is, not only the geometric shape of the surface is taken into account, but also the dynamic changes of the surface during the evolution process are taken into account, so as to more accurately simulate and reconstruct the real morphology of the tumor wall. In the process of synthesizing images, trilinear interpolation is applied to effectively infer new values ​​from known data points, thereby generating a smoother and continuous interpolation image function, thereby reducing the noise and discontinuity caused by discrete data and improving the accuracy of the interpolation result. The first position information is updated multiple times according to the target movement amount until the surface converges. This iterative update method ensures the stability and convergence of the inner surface of the evolving tumor wall in morphology and structure, and can generate a more realistic and reliable model. Therefore, a variety of technical means such as feature extraction, interpolation calculation, directional derivative analysis and iterative update are combined to make the reconstruction process of the inner surface of the evolving tumor wall more complete and controllable.

[0016] In an optional embodiment, a three-dimensional reconstruction of the aneurysm is performed based on the evolved aneurysm wall inner surface and the vascular lumen segmentation mask to obtain a target aneurysm wall inner surface, including: voxel sampling of a first volume enclosed by the evolved aneurysm wall inner surface to obtain voxel data; mapping the voxel data to a third target sequence to obtain an aneurysm cavity mask, the aneurysm cavity mask is used to characterize the cavity area inside the aneurysm; merging the aneurysm cavity mask and the vascular lumen segmentation mask to obtain an inner cavity mask, the inner cavity mask is a mask that includes the complete internal structure of the aneurysm; and a three-dimensional reconstruction of the aneurysm is performed based on the inner cavity mask to obtain a target aneurysm wall inner surface.

[0017] The processing method of the magnetic resonance image sequence provided by the embodiment of the present invention combines multiple information sources such as the inner surface of the evolving aneurysm wall, the vascular lumen segmentation mask and the aneurysm cavity mask, and these information are used together for the three-dimensional reconstruction of the aneurysm. By comprehensively utilizing different data and masks, the internal and external structures of the aneurysm can be captured and reconstructed more comprehensively, and the accuracy and completeness of the reconstruction results are improved. The voxel sampling technology is used to process the first volume surrounded by the inner surface of the evolving aneurysm wall, and then the voxel data is mapped to the third target sequence to obtain the aneurysm cavity mask. That is, the voxelization method is used to effectively represent and process complex structural information, providing basic data for subsequent steps. By merging the aneurysm cavity mask and the vascular lumen segmentation mask, an inner cavity mask is obtained. The inner cavity mask is used to characterize the cavity area inside the aneurysm. The generation of this mask can effectively distinguish the inner and outer parts of the aneurysm and accurately describe its internal structure, providing accurate data support for subsequent three-dimensional reconstruction. Finally, the aneurysm is reconstructed three-dimensionally based on the lumen mask to obtain the inner surface of the target aneurysm wall, which can ensure the accuracy and completeness of the reconstruction results and contribute to the in-depth understanding and precise quantification of the aneurysm structure in medical imaging analysis.

[0018] In an optional embodiment, the signal intensity mapping of the target aneurysm wall inner surface is analyzed to determine the quantitative index of the aneurysm, including: extracting the second position information of the second vertex in the evolving aneurysm wall inner surface; calculating the center of gravity of the second volume enclosed by the evolving aneurysm wall inner surface according to the second position information, and expanding the evolving aneurysm wall inner surface with the center of gravity as the center to obtain the expanded third aneurysm wall inner surface; determining the target area in the target aneurysm wall inner surface surrounded by the third aneurysm wall inner surface; and determining the third vertex in the target area, and determining each third vertex as an initial vertex set; repairing the initial vertex set according to the reconstruction closing operation to obtain a repaired target vertex set; and performing signal intensity mapping analysis on the target vertex set to obtain the corresponding quantitative index.

[0019] The processing method of the magnetic resonance image sequence provided by the embodiment of the present invention realizes the extraction of quantitative indicators of aneurysms by analyzing the signal intensity mapping of the inner surface of the target aneurysm wall. These indicators include the position information of the second vertex, the calculation of the center of gravity of the second volume, and the expansion and repair of the inner surface of the aneurysm wall based on the center of gravity, which helps to accurately quantify the morphological characteristics and internal structure of the aneurysm and provide objective data support for clinical evaluation. By determining the second and third vertex sets of the inner surface of the evolving aneurysm wall and performing closed operation repair, complex aneurysm wall structures and boundary conditions can be effectively processed. This multi-level vertex set screening and repair process helps to reduce noise and unnecessary structures in the data, and improves the accuracy and reliability of quantitative analysis. Mapping and analyzing the signal intensity of the target vertex set is a key step in evaluating the tissue characteristics of the inner surface of the aneurysm wall. Through this analysis, signal intensity information of different parts can be obtained, which further helps doctors understand the nature, size and degree of impact of the lesion on the surrounding tissues, and provides an important reference for clinical decision-making. Therefore, the integration of multiple algorithms and data processing techniques, including geometric information of the inner surface of the evolving aneurysm wall, volume calculation, dilation operation, vertex patching, and signal intensity mapping, helps to fully understand and analyze the complex structure of aneurysms and provide a comprehensive quantitative evaluation.

[0020] In an optional embodiment, a signal strength mapping analysis is performed on a set of target vertices to obtain a corresponding quantitative index, including: calculating a second normal vector of a third vertex; creating a sampling spoke of a preset length from the third vertex along the direction of the second normal vector; using the sampling spoke to perform sampling on the second target sequence and the third target sequence after trilinear interpolation to obtain a sampling result; determining the maximum value among each sampling result as the signal strength sampling value corresponding to the third vertex; and determining the quantitative index based on the signal strength sampling value.

[0021] The processing method of the magnetic resonance image sequence provided by the embodiment of the present invention can perform accurate sampling on the target image by calculating the second normal vector of the third vertex and creating a sampling spoke of a preset length along the normal vector direction. By adopting the trilinear interpolation technology, data interpolation can be effectively performed between the second and third target sequences, thereby obtaining more accurate and reliable sampling results. The signal intensity sampling values ​​corresponding to multiple third vertices are used to determine the final quantitative index. This multi-point sampling approach can fully reflect the signal intensity distribution of the target area, thereby more comprehensively evaluating the nature and extent of the lesion. By adopting the sampling spoke of a preset length and the trilinear interpolation technology, flexibility and adaptability can be maintained when processing lesion areas of different sizes and shapes. This feature makes it suitable for various aneurysm shapes and sizes, thereby improving the versatility and practicality of the method. Signal intensity analysis based on mathematical models and interpolation algorithms is scientific and repeatable, ensuring the accuracy and verifiability of the analysis results.

[0022] In a second aspect, the present invention provides a processing device for a magnetic resonance image sequence, comprising: an acquisition module, used to acquire multiple magnetic resonance image sequences corresponding to a target vascular region, and to acquire aneurysm information pre-annotated in any magnetic resonance image sequence; wherein the multiple magnetic resonance image sequences are multiple images acquired during the magnetic resonance imaging process, and the aneurysm information is used to characterize the aneurysm position in the target vascular region; a registration module, used to register the multiple magnetic resonance image sequences to the same sequence space, and to crop the registered multiple magnetic resonance image sequences according to the aneurysm information to obtain a target image corresponding to the aneurysm; wherein the same sequence space is a space corresponding to any sequence in the multiple magnetic resonance image sequences, and the target image is used to characterize the aneurysm and its surroundings. a vascular structure; a processing module for processing a target image and determining a vascular lumen segmentation mask corresponding to the target image, wherein the vascular lumen segmentation mask is used to characterize the contours of all vascular lumens in the target image; a construction module for extracting aneurysm information included in the target image and constructing an initial inner surface of the aneurysm wall according to the aneurysm information; a reconstruction module for extracting features of the initial inner surface of the aneurysm wall, reconstructing the initial inner surface of the aneurysm wall according to the extracted features, and obtaining an evolved inner surface of the aneurysm wall; and performing three-dimensional reconstruction of the aneurysm according to the evolved inner surface of the aneurysm wall and the vascular lumen segmentation mask to obtain a target inner surface of the aneurysm wall; and a quantification module for analyzing the signal intensity mapping of the target inner surface of the aneurysm wall and determining the quantitative index of the aneurysm.

[0023] In a third aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the method for processing a magnetic resonance image sequence of the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0024] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the method for processing a magnetic resonance image sequence of the first aspect or any corresponding embodiment thereof.

[0025] In a fifth aspect, the present invention provides a computer program product, comprising computer instructions for causing a computer to execute the method for processing a magnetic resonance image sequence according to the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0027] Figure 1 is a schematic flow chart of a method for processing a magnetic resonance image sequence according to an embodiment of the present invention;

[0028] Figure 2 is a flow chart of another method for processing a magnetic resonance image sequence according to an embodiment of the present invention;

[0029] Figure 3 is a schematic flow chart of another method for processing a magnetic resonance image sequence according to an embodiment of the present invention;

[0030] Figure 4 is a flowchart of a method for processing a magnetic resonance image sequence according to an embodiment of the present invention;

[0031] Figure 5 is a structural block diagram of a device for processing a magnetic resonance image sequence according to an embodiment of the present invention;

[0032] Figure 6 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0034] The current assessment of cerebral aneurysm risk mainly relies on quantitative aneurysm morphology indicators and enhanced black blood sequences after gadolinium angiography. The enhanced part observed in the enhanced sequence may indicate the heterogeneous risk of the region. Aneurysm wall enhancement (AWE) is considered to be a marker of potential risk of unruptured aneurysm (UIA). Studies have shown that AWE is closely related to aneurysm growth and rupture risk. However, the definition of AWE in different literature and clinical practice has not yet been unified.

[0035] Currently, clinical evaluation mainly relies on the physician's visual observation, which requires comparison of plain scan and enhanced sequence images to confirm the presence of AWE. Since the cross-sectional images are processed and analyzed in a two-dimensional view, they are limited by the manual sampling plane selection. In addition, black blood images are challenging to visually evaluate, and the physician's subjective judgment may affect the consistency and accuracy of the results.

[0036] A variety of quantitative indices based on AWE have been proposed, such as calculating the aneurysm wall risk index by the signal intensity characteristics of the aneurysm in high-resolution MRI images. However, manual calculation of these indices requires precise drawing of the aneurysm wall boundary, which is time-consuming and limited to planar views, limiting their widespread clinical application.

[0037] In recent years, some studies have been devoted to the realization of automated visualization methods for tumor wall enhancement. These methods usually rely on experienced doctors to manually perform fine annotations and map the signal intensity to the surface of the tumor wall through surface normal vector sampling. However, due to the imaging characteristics of the tumor wall in high-resolution magnetic resonance imaging (MRI), its thickness is usually only 2-3 voxels and its distinction from surrounding tissues is low, so separating the tumor wall still requires highly experienced doctors to ensure accuracy.

[0038] In view of this, in order to accurately realize the automatic quantification of aneurysm wall enhancement indicators, the technical solution of the present invention preprocesses and jointly registers various magnetic resonance image sequences of aneurysm patients, and extracts the vascular lumen segmentation mask from the second target sequence and the third target sequence using the segmentation algorithm. The surface of the aneurysm wall is accurately extracted using an adaptive algorithm and integrated into the vascular lumen segmentation mask to reconstruct the three-dimensional aneurysm. The quantitative indicators of the aneurysm are obtained by mapping the signal intensity of the aneurysm wall.

[0039] According to an embodiment of the present invention, an embodiment of a method for processing a magnetic resonance image sequence is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0040] In this embodiment, a method for processing a magnetic resonance image sequence is provided, which can be used in a computer device, such as a notebook computer, a desktop computer, etc. Figure 1 is a flow chart of a method for processing a magnetic resonance image sequence according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0041] Step S101, obtaining multiple magnetic resonance image sequences corresponding to the target blood vessel region, and obtaining aneurysm information pre-annotated in any magnetic resonance image sequence; wherein the multiple magnetic resonance image sequences are multiple images obtained during the magnetic resonance imaging process, and the aneurysm information is used to characterize the aneurysm position in the target blood vessel region.

[0042] The multiple magnetic resonance image sequences are multiple magnetic resonance imaging (MRI) sequences for observing and analyzing specific blood vessels or vascular systems. Specifically, different magnetic resonance image sequences have different imaging characteristics, such as a Time-of-Flight Magnetic Resonance Angiography (TOF-MRA) sequence in a DICOM file format for displaying vascular structures, and a T1-weighted imaging sequence or a T2-weighted imaging sequence for anatomical structure and lesion assessment.

[0043] In any magnetic resonance image sequence, doctors need to manually mark aneurysms. Specifically, this can include outlining the aneurysm or painting the aneurysm area. These manually marked aneurysm annotations will undergo a morphological opening operation to filter tiny connected domains and calculate the three-dimensional convex hull to form the actual annotation mask, i.e., the aneurysm information. This ensures the accuracy and usability of the annotation, so that it can be used as input data for subsequent quantitative analysis.

[0044] Step S102, aligning the multiple magnetic resonance image sequences to the same sequence space, and cropping the multiple aligned magnetic resonance image sequences according to the aneurysm information to obtain a target image corresponding to the aneurysm; wherein the same sequence space is a space corresponding to any sequence in the multiple magnetic resonance image sequences, and the target image is used to characterize the aneurysm and the first vascular structure surrounding it.

[0045] The first vascular structure refers to the main vascular structure closest to the aneurysm. Since multiple magnetic resonance image sequences have deviations in spatial position and direction, they need to be registered, that is, they are aligned to a unified coordinate system. For example, the TOF-MRA sequence and the T1-weighted imaging sequence can be registered to the T1-weighted enhanced imaging sequence space. The purpose of registration is to ensure that the anatomical structures between different magnetic resonance image sequences are completely spatially corresponding, so that the aneurysm and its surrounding structures can be accurately analyzed. Specifically, after completing the image registration, the multiple registered magnetic resonance image sequences are cropped according to the aneurysm information that is pre-labeled or automatically detected. The goal of cropping is to extract the region of interest (Volume of Interest, VOI) including the aneurysm and the first vascular structure around it, so as to ensure that in the subsequent analysis process, only the aneurysm area and its corresponding vascular structure are focused on, avoiding the interference of irrelevant structures on the analysis results. After cropping, the image obtained is the target image. These target images are specifically used for subsequent quantitative analysis, such as measuring the size, shape, position of the aneurysm and the characteristics of the surrounding blood vessels.

[0046] Step S103 : Process the target image to determine a blood vessel lumen segmentation mask corresponding to the target image. The blood vessel lumen segmentation mask is used to characterize the contours of all blood vessel lumens in the target image.

[0047] The target image is processed, blood vessel-related features are extracted from the target image, and the vascular lumen area in the image is separated from the surrounding tissue. The segmented vascular area forms a mask. The mask is a binary image in which the vascular lumen area is marked as white (or a value of 1) and the background is black (or a value of 0). This mask is called a vascular lumen segmentation mask.

[0048] Step S104: extracting aneurysm information included in the target image, and constructing an initial inner surface of the aneurysm wall according to the aneurysm information.

[0049] Image processing and analysis techniques are used to extract annotation masks from the target image, which may include the location, shape, size, signal intensity, signal characteristics of the T1-weighted imaging sequence, and the contrast between the aneurysm and the surrounding tissue. Specifically, the aneurysm is separated from the surrounding tissue by image segmentation technology based on the annotation mask, and the three-dimensional surface of the aneurysm wall is reconstructed by a reconstruction algorithm, that is, the initial inner surface of the aneurysm wall.

[0050] Step S105, extract features from the initial inner surface of the aneurysm wall, reconstruct the initial inner surface of the aneurysm wall according to the extracted features to obtain the evolved inner surface of the aneurysm wall; and perform three-dimensional reconstruction of the aneurysm according to the evolved inner surface of the aneurysm wall and the vascular lumen segmentation mask to obtain the target inner surface of the aneurysm wall.

[0051] Computer vision and image processing techniques are used to analyze image data or other medical imaging data to identify and quantify the features of the initial inner surface of the tumor wall. These features may include the surface geometry, curvature, convexity and concavity, and the boundary with the surrounding tissue. Specifically, through mathematical modeling and optimization techniques, the initial inner surface of the tumor wall is adjusted and reconstructed according to the extracted features to make it closer to the actual inner structure of the tumor wall, and the evolved inner surface of the tumor wall is obtained. That is, the goal of reconstruction is to improve the accuracy and fidelity of the surface so that subsequent analysis and processing can more accurately reflect the actual situation. The evolved inner surface of the tumor wall refers to the final surface state obtained after the reconstruction process. This surface can be the result of gradual improvement through an iterative optimization process to ensure that it is as consistent as possible with the actual observed inner layer of the tumor wall.

[0052] After obtaining the evolved inner surface of the aneurysm wall, the 3D reconstruction of the aneurysm is performed in combination with the vascular lumen segmentation mask. Specifically, the vascular lumen segmentation mask is a binary representation of the internal structure of the blood vessel. Combining the evolved inner surface of the aneurysm wall with the vascular lumen segmentation mask can generate a complete 3D aneurysm model, wherein the surface of the 3D aneurysm model is the accurately reconstructed inner layer of the aneurysm wall, i.e., the target inner surface of the aneurysm wall.

[0053] Step S106, analyzing the signal intensity mapping of the inner surface of the target aneurysm wall to determine the quantitative index of the aneurysm.

[0054] Quantitative indicators are used to quantitatively describe the characteristics or parameters of aneurysms in magnetic resonance image sequences in order to more accurately evaluate the disease state corresponding to the aneurysm, predict the treatment effect, or monitor the disease changes. Specifically, after reconstructing the three-dimensional aneurysm model, the vertex set on the inner surface of the target aneurysm wall is extracted, and the signal intensity at each vertex is mapped and analyzed to obtain the signal intensity mapping result. The signal intensity mapping results are presented through visualization technology, such as making a heat map or surface map, so that doctors and researchers can intuitively understand the signal characteristics of the aneurysm wall inside the aneurysm and extract useful medical information from it.

[0055] The method for processing magnetic resonance image sequences provided by the embodiment of the present invention obtains and utilizes multiple magnetic resonance image sequences, comprehensively displays the target vascular area and the aneurysm information therein, can provide more comprehensive and multi-angle information, and is helpful for more accurate analysis and diagnosis of aneurysms. By accurately registering multiple magnetic resonance image sequences and cropping them according to the pre-marked aneurysm information, a target image corresponding to the aneurysm is obtained. This ensures that the spatial position and anatomical structure are kept consistent between different images, providing a reliable basis for subsequent analysis and processing. The information of the aneurysm in the target image is extracted and analyzed, and the initial inner surface of the aneurysm wall is constructed based on this information. Through feature extraction and reconstruction technology, the inner surface of the evolving aneurysm wall can be presented more finely, which helps doctors to have a deeper understanding of the morphology and structure of the aneurysm. The three-dimensional reconstruction of the aneurysm is performed based on the inner surface of the evolving aneurysm wall and the segmentation mask of the vascular lumen, which not only provides the spatial information of the aneurysm, but also can quantitatively analyze the characteristics and signal intensity mapping of the aneurysm, thereby providing a scientific basis to support clinical decision-making.

[0056] In this embodiment, a method for processing a magnetic resonance image sequence is provided, which can be used in a computer device, such as a notebook computer, a desktop computer, etc. Figure 2 is a flow chart of a method for processing a magnetic resonance image sequence according to an embodiment of the present invention. Figure 2 As shown, the process includes the following steps:

[0057] Step S201, obtaining multiple magnetic resonance image sequences corresponding to the target vascular region, and obtaining aneurysm information pre-annotated in any magnetic resonance image sequence; wherein the multiple magnetic resonance image sequences are multiple images obtained during the magnetic resonance imaging process, and the aneurysm information is used to characterize the aneurysm position in the target vascular region. For details, please refer to Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.

[0058] Step S202, aligning the multiple magnetic resonance image sequences to the same sequence space, and cropping the multiple aligned magnetic resonance image sequences according to the aneurysm information to obtain a target image corresponding to the aneurysm; wherein the same sequence space is a space corresponding to any sequence in the multiple magnetic resonance image sequences, and the target image is used to characterize the aneurysm and the first vascular structure surrounding it.

[0059] Among them, the multiple magnetic resonance image sequences include a first magnetic resonance image sequence, a second magnetic resonance image sequence and a third magnetic resonance image sequence; the first magnetic resonance image sequence is used to display a second vascular structure corresponding to a target vascular region, the second magnetic resonance image sequence is used to display the anatomical structure and signal intensity difference of tissue included in the target vascular region, and the third magnetic resonance image sequence is used to display the contrast enhancement effect of an aneurysm corresponding to the target vascular region after adding a contrast agent.

[0060] The first magnetic resonance image sequence is mainly used to display the second vascular structure corresponding to the target vascular region, that is, it mainly displays the structure of the vascular network, including the distribution of major arteries and veins, and their positional relationship in the entire image. For example, the first magnetic resonance image sequence can be a time-of-flight magnetic resonance angiography sequence.

[0061] The second magnetic resonance image sequence is used to display the anatomical structure and signal intensity difference of the tissue included in the target vascular region, that is, to provide more detailed anatomical information, including the tissue types around the aneurysm (such as brain tissue, vascular wall, other surrounding tissues) and the signal intensity difference of these tissues on the magnetic resonance image sequence. For example, the second magnetic resonance image sequence can be a T1-weighted imaging sequence.

[0062] The third magnetic resonance image sequence is an image after adding a contrast agent, which is used to show the contrast enhancement effect of the aneurysm corresponding to the target vascular area. Specifically, a contrast agent is a substance that increases the contrast between blood vessels and tissues, and is used to highlight abnormal vascular structures or lesions (such as aneurysms). This image can show the enhancement of the aneurysm itself and the surrounding vascular network under the influence of the contrast agent, which helps to more accurately identify and evaluate the characteristics of the aneurysm. For example, the third magnetic resonance image sequence can be a T1-weighted enhanced imaging sequence.

[0063] Specifically, the above step S202 includes:

[0064] Step S2021 , preprocessing the first magnetic resonance image sequence, the second magnetic resonance image sequence, and the third magnetic resonance image sequence.

[0065] Preprocessing is performed on the first, second and third magnetic resonance image sequences. Specifically, N4 field strength calibration and Min / Max Curvature Flow Denoising steps are used to optimize image quality and remove noise.

[0066] Step S2022: registering the preprocessed first magnetic resonance image sequence, the second magnetic resonance image sequence, and the third magnetic resonance image sequence according to a preset medical image registration library, so as to register the first magnetic resonance image sequence and the second magnetic resonance image sequence to the third magnetic resonance image sequence.

[0067] The preprocessed first magnetic resonance image sequence, the second magnetic resonance image sequence, and the third magnetic resonance image sequence are jointly registered using the medical image registration library ANTs (Advanced Normalization Tools). Specifically, the purpose of the registration is to align the spatial positions and anatomical structures of these images to ensure that subsequent analysis can be performed in the same anatomical space, that is, the third magnetic resonance image sequence space.

[0068] Step S2023, cropping the registered first magnetic resonance image sequence, the second magnetic resonance image sequence, and the third magnetic resonance image sequence respectively according to the aneurysm information to obtain a first target sequence corresponding to the first magnetic resonance image sequence, a second target sequence corresponding to the second magnetic resonance image sequence, and a third target sequence corresponding to the third magnetic resonance image sequence.

[0069] According to the aneurysm information, the first magnetic resonance image sequence, the second magnetic resonance image sequence and the third magnetic resonance image sequence after registration are cropped. Specifically, the purpose of cropping is to extract the region of interest around the aneurysm for subsequent quantitative analysis and comparison. After cropping, each magnetic resonance image sequence is obtained to obtain the corresponding first target sequence (the sequence after the first magnetic resonance image sequence is cropped), the second target sequence (the sequence after the second magnetic resonance image sequence is cropped) and the third target sequence (the sequence after the third magnetic resonance image sequence is cropped).

[0070] Furthermore, the signal density of the second target sequence and the third target sequence can be normalized according to the definition of the quantitative index [OMODAKA 2016]. This is done by normalizing the average signal density of the white matter around the aneurysm to eliminate the difference in overall signal intensity between different sequences, so as to make subsequent analysis and comparison more accurate.

[0071] The processing method of the magnetic resonance image sequence provided by the embodiment of the present invention utilizes three different types of magnetic resonance image sequences. Through these different images, the target blood vessel area and the aneurysm information therein are comprehensively displayed, which helps doctors to fully understand the condition. The multiple magnetic resonance image sequences are accurately registered, which includes preprocessing each type of image and registering according to a preset medical image registration library, which can ensure that the spatial position and anatomical structure between different images are consistent, thereby providing a reliable basis for subsequent analysis and diagnosis. The first magnetic resonance image sequence, the second magnetic resonance image sequence and the aneurysm information are registered to the third magnetic resonance image sequence. The third magnetic resonance image sequence is usually used to display the distribution of tissue contrast agents and the enhancement effect of aneurysms. By applying the registered information to this sequence, the location and characteristics of the aneurysm can be observed more clearly. According to the aneurysm information, the multiple registered magnetic resonance image sequences are cropped to obtain the target image corresponding to the aneurysm. This precise cropping operation can ensure that the extracted image area only contains the aneurysm part of interest, excluding other irrelevant structures and backgrounds, which is helpful for subsequent analysis and diagnosis. The first magnetic resonance image sequence, the second magnetic resonance image sequence, and the third magnetic resonance image sequence are cropped respectively to obtain the corresponding target sequences, thereby retaining the multimodal information of each image sequence, such as structure, anatomical features, and enhancement effects, and being able to comprehensively display the multifaceted characteristics of aneurysms. The target sequences obtained after cropping can be directly used for subsequent computer-aided diagnosis or other processing tasks. These image sequences have been located in the aneurysm area of ​​interest and are aligned in spatial position, making further quantitative analysis or comparative analysis more convenient and accurate.

[0072] Step S203 : Process the target image to determine a blood vessel lumen segmentation mask corresponding to the target image. The blood vessel lumen segmentation mask is used to characterize the contours of all blood vessel lumens in the target image.

[0073] Specifically, the above step S203 includes:

[0074] Step S2031: Process the first target sequence to determine a white blood mask, where the white blood mask is used to characterize the position and shape of all blood vessels in the target image.

[0075] The first target sequence is processed, including skull removal, high signal intensity (SI) voxel threshold segmentation, and connected domain filtering operations, to obtain a white blood mask, which is used to represent the location and shape of blood vessels in the target image.

[0076] Step S2032: perform Gaussian filter edge enhancement processing on the second target sequence and the third target sequence, and synthesize the second target sequence and the third target sequence after the Gaussian filter edge enhancement processing to generate a synthesized image.

[0077] The Gaussian filter edge enhancement method is applied to the second target sequence and the third target sequence. The formulas for Gaussian filter edge enhancement are shown in equations (1) and (2), and the edge features of the blood vessel wall are enhanced by convolution calculation. The second target sequence and the third target sequence that have undergone edge enhancement processing are synthesized according to equation (3) to obtain a new synthetic image, so as to comprehensively utilize the mixed features of the second target sequence and the third target sequence.

[0078]

[0079] F(I)=I+max(0,IG*I) (2)

[0080] I Mixed =F(I T1 )+F(I T1+C ) (3)

[0081] Among them, I is the three-dimensional image matrix, and * represents the convolution symbol.

[0082] Among them, formula (1) is the continuous function of three-dimensional Gaussian filtering, formula (2) is the edge enhancement function, and formula (3) is the synthesis function.

[0083] For example, when the second target sequence is a sequence cropped from a T1-weighted imaging sequence and the third target sequence is a sequence cropped from a T1-weighted enhanced imaging sequence, the second target sequence I T1 and the third target sequence I T1+C Substituting into formula (3), we get the composite image I Mixed .

[0084] Step S2033, obtaining a preset pixel threshold, where the pixel threshold is used to distinguish the blood vessel lumen in the synthesized image.

[0085] The pixel threshold is determined by doctors or researchers based on specific image processing requirements and image features. Specifically, the pixel threshold can be set based on the grayscale level or color intensity of the magnetic resonance image sequence to extract a specific object of interest, namely, the vascular lumen, from the composite image.

[0086] Step S2034, processing the synthesized image according to the pixel threshold to obtain a binary mask image.

[0087] The composite image is processed using a preset pixel threshold to obtain a binary mask image. Specifically, the composite image is converted into a grayscale image, and the grayscale image is segmented using the pixel threshold. For example, pixels with grayscale values ​​higher than the threshold are set to one value (usually 255, indicating white), and pixels with grayscale values ​​lower than the threshold are set to another value (usually 0, indicating black), thereby generating a binary mask image.

[0088] Based on the result of threshold segmentation, a binary mask image is generated. In this mask image, white pixels represent areas that meet the conditions (vascular lumen), and black pixels represent other areas (such as background or non-interested areas).

[0089] Step S2035 , extracting a blood vessel lumen segmentation mask from the mask image according to the white blood mask and the mask image.

[0090] Combine the white blood mask and the binary mask image, and perform corresponding image processing operations, such as vessel lumen segmentation operations, to obtain the final black blood vessel lumen mask I VL , i.e., the vascular lumen segmentation mask.

[0091] Step S204: extract the aneurysm information included in the target image, and construct the initial aneurysm wall inner surface according to the aneurysm information. Figure 1 Step S104 of the illustrated embodiment will not be described in detail here.

[0092] Step S205: extract features from the initial inner surface of the aneurysm wall, reconstruct the initial inner surface of the aneurysm wall based on the extracted features to obtain the evolved inner surface of the aneurysm wall; and perform 3D reconstruction of the aneurysm based on the evolved inner surface of the aneurysm wall and the vascular lumen segmentation mask to obtain the target inner surface of the aneurysm wall. Figure 1 Step S105 of the illustrated embodiment will not be described in detail here.

[0093] Step S206: Analyze the signal intensity mapping of the inner surface of the target aneurysm wall to determine the quantitative index of the aneurysm. Figure 1 Step S106 of the illustrated embodiment will not be described in detail here.

[0094] The processing method of the magnetic resonance image sequence provided by the embodiment of the present invention processes the first target sequence to determine the white blood mask, which is used to accurately locate the position and shape of all blood vessels in the target image, and provide basic information for the subsequent segmentation of the blood vessel lumen. The edge enhancement processing of the second target sequence and the third target sequence by Gaussian filtering can enhance the edge features of the blood vessels, make the contours of the blood vessels clearer, and facilitate the subsequent segmentation and analysis of the blood vessels. The second target sequence and the third target sequence after the edge enhancement processing by Gaussian filtering are synthesized to generate a synthetic image. The synthetic image integrates multimodal information, can provide more comprehensive visual information, and helps to distinguish the blood vessel lumen more accurately. Obtain a preset pixel threshold for distinguishing the blood vessel lumen in the synthetic image. By setting the threshold, the blood vessel part in the image can be separated from the background to form a preliminary binary mask image. According to the white blood mask and the mask image, the blood vessel lumen segmentation mask is extracted from the mask image, which can accurately segment the position of the blood vessel and provide accurate data support for subsequent quantitative analysis and diagnosis.

[0095] In this embodiment, a method for processing a magnetic resonance image sequence is provided, which can be used in a computer device, such as a notebook computer, a desktop computer, etc. Figure 3 is a flow chart of a method for processing a magnetic resonance image sequence according to an embodiment of the present invention. Figure 3 As shown, the process includes the following steps:

[0096] Step S301, obtaining multiple magnetic resonance image sequences corresponding to the target vascular region, and obtaining aneurysm information pre-annotated in any magnetic resonance image sequence; wherein the multiple magnetic resonance image sequences are multiple images obtained during the magnetic resonance imaging process, and the aneurysm information is used to characterize the aneurysm position in the target vascular region. For details, please refer to Figure 2 Step S201 of the illustrated embodiment will not be described in detail here.

[0097] Step S302, align multiple magnetic resonance image sequences to the same sequence space, and crop the aligned multiple magnetic resonance image sequences according to the aneurysm information to obtain a target image corresponding to the aneurysm; wherein the same sequence space is a space corresponding to any sequence in the multiple magnetic resonance image sequences, and the target image is used to characterize the aneurysm and the first vascular structure around it. For details, please refer to Figure 2 Step S202 of the illustrated embodiment will not be described in detail here.

[0098] Step S303, process the target image to determine the vascular lumen segmentation mask corresponding to the target image, and the vascular lumen segmentation mask is used to characterize the contours of all vascular lumens in the target image. Figure 2 Step S203 of the illustrated embodiment will not be described in detail here.

[0099] Step S304: extract the aneurysm information included in the target image, and construct the initial aneurysm wall inner surface of the aneurysm according to the aneurysm information.

[0100] Specifically, the above step S304 includes:

[0101] Step S3041: extract aneurysm information included in the target image, and convert the aneurysm information into a three-dimensional convex hull.

[0102] The extracted aneurysm information is converted into a three-dimensional convex hull. Specifically, the three-dimensional convex hull is the smallest convex polygon or convex polyhedron surrounding the aneurysm area, which can effectively describe the overall shape and structure of the aneurysm.

[0103] Step S3042, converting the three-dimensional convex hull into the inner surface of the first tumor wall.

[0104] The first aneurysm wall inner surface is a surface of a three-dimensional convex hull converted based on aneurysm information, that is, a surface initially obtained from the three-dimensional convex hull. Specifically, the three-dimensional convex hull can be converted into a smooth surface mesh using a marching cube algorithm. The marching cube algorithm converts discrete three-dimensional data (voxel data represented by a three-dimensional convex hull) into a continuous surface model for structural surface reconstruction.

[0105] Step S3043, smoothing the first tumor wall inner layer surface to obtain a second tumor wall inner layer surface.

[0106] The first aneurysm wall inner surface is smoothed to reduce surface irregularities and noise. For example, the Taunin-constrained Laplace algorithm or other smoothing algorithms are used to ensure the continuity of the surface and the naturalness of the morphology. After smoothing, the resulting surface is called the second aneurysm wall inner surface to further optimize the surface morphology of the aneurysm to make it more consistent with the actual situation and prepare more accurate data for subsequent processing.

[0107] Step S3044: performing isotropic voxel resampling on the inner surface of the second tumor wall to obtain a three-dimensional voxel data set.

[0108] For the inner surface of the second tumor wall, isotropic voxel resampling is performed to obtain a three-dimensional voxel data set. Specifically, considering the irregularity of voxel spacing and low resolution in the original image sequence, isotropic voxel resampling can improve the spatial resolution and consistency of the data, making subsequent analysis and visualization more accurate.

[0109] Step S3045: constructing the initial inner surface of the aneurysm wall according to the three-dimensional voxel data set.

[0110] The inner surface of the aneurysm wall, i.e., the initial inner surface of the aneurysm wall, was reconstructed using the isotropically resampled 3D voxel dataset, which ensured accurate aneurysm shape and structural information was obtained from high-resolution data, providing a good basis for further analysis and model building.

[0111] The processing method of the magnetic resonance image sequence provided by the embodiment of the present invention is an effective way to convert the aneurysm information into a three-dimensional convex hull, which can convert the input aneurysm data into a basic representation of a geometric shape. The convex hull representation method is simple and clear, and can better reflect the overall shape and contour of the aneurysm. The three-dimensional convex hull is converted into the inner surface of the first aneurysm wall, so that the geometric features of the convex hull can be retained and effectively converted into a more refined and realistic surface representation. The smoothing of the inner surface of the first aneurysm wall is to eliminate possible noise and irregularities, make the surface more continuous and natural, help improve the quality of the results of subsequent processing, and ensure that the constructed inner surface of the aneurysm wall has good geometric characteristics. Through isotropic voxel resampling, the smoothed inner surface of the second aneurysm wall can be converted into a three-dimensional voxel data set, which can provide an accurate description of the specific position in space and provide a basis for subsequent analysis and visualization. According to the three-dimensional voxel data set, the initial inner surface of the aneurysm wall is constructed, and the voxel data is converted into a geometric surface, which provides an accurate shape description for further analysis and simulation, and is helpful for medical imaging diagnosis and treatment planning.

[0112] Step S305, extract features from the initial inner surface of the aneurysm wall, reconstruct the initial inner surface of the aneurysm wall based on the extracted features to obtain the evolved inner surface of the aneurysm wall; and perform three-dimensional reconstruction of the aneurysm based on the evolved inner surface of the aneurysm wall and the vascular lumen segmentation mask to obtain the target inner surface of the aneurysm wall.

[0113] Specifically, the above step S305 includes:

[0114] Step S3051: extract the first position information and edge information of the first vertex in the initial tumor wall inner surface, and process the first position information and edge information to obtain the first normal vector and evolution direction of the first vertex.

[0115] The first vertex is a specific point in the initial inner surface of the tumor wall, and its specific position information and related attributes are used for feature extraction and subsequent processing. The initial inner surface of the tumor wall is a mesh model (such as a triangular mesh) composed of vertices (first vertices) and edges connecting these vertices. The first position information is the specific position coordinates of each vertex in three-dimensional space, and the edge information is the information of the edges connecting the vertices, including which vertices are connected and the geometric relationship between them, such as the length and angle of the edge. Specifically, the reconstructed initial inner surface of the tumor wall is recorded as The three-dimensional grid undirected graph in is There are N first vertices in total. For the picture The vertex set of The edge set of Any first vertex in The set of vertices whose minimum number of edges required to pass from the first vertex v is less than or equal to k is called the k-neighborhood vertex set. 1-Neighborhood Vertex Set and the k-neighborhood vertex set (k=2,3,4,L) can be obtained recursively by the following equations (4) and (5). Equation (6) is the k-neighborhood average matrix Q k , where m(·) represents the vertex weight, and the Voronoi area of ​​the vertex is usually used as the weight value.

[0116]

[0117]

[0118]

[0119] Furthermore, ideally, the initial inner surface of the tumor wall is at the boundary between high and low signal intensity (SI), and the variation of SI on the inner surface is more obvious. Therefore, the maximum value of the derivative can be used as the ideal dividing point to perform iterative optimization of the surface.

[0120] Calculate the unit normal vector of each first vertex, and the coordinate matrix of all vertices is Calculate the unit normal vector matrix of all vertices on the surface Specifically, the first normal vector of each first vertex is obtained according to the direction of the facet corresponding to each first vertex after the initial tumor wall inner layer surface is reconstructed, and is a reconstruction result.

[0121] Calculate the evolution direction of each first vertex. The evolution direction is obtained by mixing the normal vector averaged over the neighborhood with the direction of the centroid ray. The unit evolution direction matrix of all first vertices is calculated using equations (7) and (8): in Represents the centroid of the volume enclosed by the three-dimensional surface. The parameter λ adjusts the mixing ratio of the first normal vector and the direction of the central ray, for example, λ=1.

[0122]

[0123]

[0124] Step S3052, performing trilinear interpolation on the synthesized image to obtain an interpolated image function; and calculating the directional derivative in the evolution direction according to the interpolated image function and the first normal vector, and determining the maximum value of the directional derivative within a preset distance around the first vertex.

[0125] Perform trilinear interpolation on the edge-enhanced composite image to obtain the interpolation image function According to the starting point of the first vertex and its corresponding evolution direction, the directional derivative in the evolution direction of each first vertex is calculated, and the maximum value of the derivative is calculated within a certain distance T near the vertex.

[0126] Step S3053, obtaining a preset target displacement distance, and determining a value obtained by adding the target displacement distance to the maximum value as the target movement amount corresponding to the first vertex.

[0127] The preset target displacement distance is determined by doctors or researchers based on the specific aneurysm medical imaging analysis task and the individual conditions of the patient. Specifically, the maximum value of the derivative plus the target displacement distance offset is used as the target displacement of each first vertex i after evolution by formula (9): Calculate the target movement of all first vertices

[0128]

[0129] Step S3054: update the first position information according to the target movement amount to obtain the target position information corresponding to the first vertex.

[0130] For each first vertex, the corresponding target position information is obtained according to the calculated target movement amount. The target position information is the position that the first vertex is expected to reach in the next iteration. Specifically, the neighborhood smoothness constraint is applied, and the position of the first vertex is updated according to formula (10) to make the surface adaptively deformed. The parameter μ∈(0,1] is the step size, which is used to adjust the amplitude of iterative evolution, for example, μ=0.25.

[0131] X←X+μN * Q k t * (10)

[0132] Step S3055: reconstruct the initial inner surface of the tumor wall according to the target position information until the initial inner surface of the tumor wall converges to obtain the evolved inner surface of the tumor wall.

[0133] The initial inner surface of the tumor wall is recalculated or adjusted using the updated target position information. The reconstruction process is iterative, and each iteration adjusts the surface according to the updated target position information so that it gradually approaches or converges to the expected shape, obtaining the evolved inner surface of the tumor wall. This iterative process can be controlled by setting the number of iterations or defining convergence conditions. That is, by optimizing and adjusting the initial inner surface of the tumor wall, multiple iterations are performed according to the target position information and the optimization algorithm until the inner surface matches or converges with the actual tumor boundary or the edge of the medical imaging data. The final surface is the evolved inner surface of the tumor wall.

[0134] However, since the aneurysm is assumed to be a continuous smooth three-dimensional volume, the neck of the inner surface of the evolving aneurysm wall is calculated to be "fictitious". In order to more accurately measure the surface area of ​​the aneurysm wall and the proportion of the enhanced area, it is necessary to perform a separation process of the aneurysm wall and the vessel wall.

[0135] Step S3056: voxel sampling is performed on the first volume enclosed by the inner surface of the evolving tumor wall to obtain voxel data.

[0136] The first volume is a specific volume formed around the inner layer of the evolving tumor wall. Specifically, the first volume is decomposed into discrete voxels (three-dimensional pixels) to obtain voxel data.

[0137] Step S3057, mapping the voxel data to the third target sequence to obtain a cavity mask, where the cavity mask is used to characterize the cavity area inside the aneurysm.

[0138] The voxel data of the inner surface of the evolving tumor wall are obtained by voxel sampling, and the voxel data are mapped to the third target sequence, and the voxel data are converted into a tumor cavity mask I according to a preset mapping rule. AL Specifically, the tumor cavity mask I AL is a binary mask describing the internal cavity area of ​​the aneurysm, indicating which voxels belong to the internal structure of the aneurysm.

[0139] However, due to the limitation of the segmentation algorithm and the difference in the accuracy of the surface iteration algorithm, the mask and reconstructed surface of the aneurysm part in the vascular lumen segmentation mask will be different from the inner surface of the evolving aneurysm wall calculated by the surface iteration algorithm. Therefore, it is necessary to restore the vertices of the aneurysm wall (each first vertex) to the previously calculated inner layer position of the aneurysm wall (the inner layer surface of the evolved aneurysm wall). ), thereby ensuring accurate reconstruction of the internal structure of the aneurysm and making the surface of the aneurysm wall more consistent and matching with the mask of the aneurysm cavity and the vascular cavity.

[0140] Step S3058: merge the aneurysm cavity mask and the blood vessel lumen segmentation mask to obtain an inner cavity mask, which is a mask containing the complete internal structure of the aneurysm.

[0141] Mask the tumor cavity I AL and vessel lumen segmentation mask I VL Merge, element-by-element logical OR operation to obtain cavity mask I Lumen .

[0142] Step S3059, performing three-dimensional reconstruction of the aneurysm according to the lumen mask to obtain the inner surface of the target aneurysm wall.

[0143] The convex hull inside the aneurysm is calculated through the lumen mask and aligned with the third target sequence. The marching cubes algorithm can generate an accurate three-dimensional surface of the aneurysm from the lumen mask. The generated accurate three-dimensional surface is smoothed using the Taubin constrained Laplace algorithm. The smoothed annotated three-dimensional surface image is resampled to improve the resolution consistency in all directions. Using the isotropic high-resolution three-dimensional mask image, a smooth surface reconstruction is performed again to produce a more detailed and accurate inner surface of the aneurysm wall, that is, the inner surface of the target aneurysm wall. In order to carry out more detailed analysis and subsequent processing.

[0144] Step S306: Analyze the signal intensity mapping of the inner surface of the target aneurysm wall to determine the quantitative index of the aneurysm. Figure 2 Step S206 of the illustrated embodiment will not be described in detail here.

[0145] The processing method of the magnetic resonance image sequence provided by the embodiment of the present invention can extract the key feature information of the surface by extracting the features of the initial inner surface of the tumor wall. These feature extraction steps help to accurately capture the shape and structural features of the tumor wall. Based on the extracted feature information, the inner surface of the tumor wall can be reconstructed by calculating the evolution direction and the target movement amount, that is, not only the geometric shape of the surface is taken into account, but also the dynamic changes of the surface during the evolution process are taken into account, so as to more accurately simulate and reconstruct the real morphology of the tumor wall. In the process of synthesizing images, trilinear interpolation is applied to effectively infer new values ​​from known data points, thereby generating a smoother and continuous interpolation image function, thereby reducing the noise and discontinuity caused by discrete data and improving the accuracy of the interpolation result. The first position information is updated multiple times according to the target movement amount until the surface converges. This iterative update method ensures the stability and convergence of the inner surface of the evolving tumor wall in morphology and structure, and can generate a more realistic and reliable model. Therefore, a variety of technical means such as feature extraction, interpolation calculation, directional derivative analysis and iterative update are combined to make the reconstruction process of the inner surface of the evolving tumor wall more complete and controllable. The paper combines multiple information sources such as the inner surface of the evolving aneurysm wall, the segmentation mask of the vascular lumen, and the aneurysm cavity mask, which are used together for the three-dimensional reconstruction of the aneurysm. By comprehensively utilizing different data and masks, the internal and external structures of the aneurysm can be captured and reconstructed more comprehensively, improving the accuracy and completeness of the reconstruction results. The voxel sampling technology is used to process the first volume enclosed by the inner surface of the evolving aneurysm wall, and then the voxel data is mapped to the second target sequence to obtain the aneurysm cavity mask. That is, the voxelization method is used to effectively represent and process complex structural information, providing basic data for subsequent steps. The lumen mask is obtained by merging the aneurysm cavity mask and the vascular lumen segmentation mask. The lumen mask is used to characterize the cavity area inside the aneurysm. The generation of this mask can effectively distinguish the inner and outer parts of the aneurysm and accurately describe its internal structure, providing accurate data support for subsequent three-dimensional reconstruction. Finally, the three-dimensional reconstruction of the aneurysm is performed according to the lumen mask to obtain the inner surface of the target aneurysm wall, which can ensure the accuracy and completeness of the reconstruction results and help to deeply understand and accurately quantify the aneurysm structure in medical image analysis.

[0146] In this embodiment, a method for processing a magnetic resonance image sequence is provided, which can be used in a computer device, such as a notebook computer, a desktop computer, etc. Figure 4 is a flow chart of a method for processing a magnetic resonance image sequence according to an embodiment of the present invention. Figure 4 As shown, the process includes the following steps:

[0147] Step S401, obtaining multiple magnetic resonance image sequences corresponding to the target blood vessel region, and obtaining aneurysm information pre-annotated in any magnetic resonance image sequence; wherein the multiple magnetic resonance image sequences are multiple images obtained during the magnetic resonance imaging process, and the aneurysm information is used to characterize the aneurysm position in the target blood vessel region. For details, please refer to Figure 3 Step S301 of the illustrated embodiment will not be described in detail here.

[0148] Step S402, aligning multiple magnetic resonance image sequences to the same sequence space, and cropping the aligned multiple magnetic resonance image sequences according to the aneurysm information to obtain a target image corresponding to the aneurysm; wherein the same sequence space is a space corresponding to any sequence in the multiple magnetic resonance image sequences, and the target image is used to characterize the aneurysm and the first vascular structure around it. For details, please refer to Figure 3 Step S302 of the illustrated embodiment will not be described in detail here.

[0149] Step S403, the target image is processed to determine the vascular lumen segmentation mask corresponding to the target image, and the vascular lumen segmentation mask is used to characterize the contours of all vascular lumens in the target image. Figure 3 Step S303 of the illustrated embodiment will not be described in detail here.

[0150] Step S404: extract the aneurysm information included in the target image, and construct the initial aneurysm wall inner surface according to the aneurysm information. Figure 3 Step S304 of the illustrated embodiment will not be described in detail here.

[0151] Step S405: extract features from the initial inner surface of the aneurysm wall, reconstruct the initial inner surface of the aneurysm wall based on the extracted features to obtain the evolved inner surface of the aneurysm wall; and perform 3D reconstruction of the aneurysm based on the evolved inner surface of the aneurysm wall and the vascular lumen segmentation mask to obtain the target inner surface of the aneurysm wall. Figure 3 Step S305 of the illustrated embodiment will not be described in detail here.

[0152] Step S406: Analyze the signal intensity mapping of the inner surface of the target aneurysm wall to determine the quantitative index of the aneurysm.

[0153] Specifically, the above step S406 includes:

[0154] Step S4061, extracting the second position information of the second vertex in the inner surface of the evolving tumor wall.

[0155] Each second vertex is extracted from the inner surface of the evolving aneurysm wall, and the second vertex represents a key feature point of the inner surface of the evolving aneurysm wall, such as a concave-convex point, a corner point, etc. In the process of extracting the second vertex, additional position information may be added, such as a normal, a curvature, etc. The second position information of the second vertex is a detailed position description associated with each vertex, such as its local curvature, the direction of the local normal, etc.

[0156] Step S4062: Calculate the center of gravity of the second volume enclosed by the evolved tumor wall inner surface according to the second position information, and expand the evolved tumor wall inner surface around the center of gravity to obtain an expanded third tumor wall inner surface.

[0157] The center of gravity (i.e., the mass center) of the volume enclosed by the evolving inner surface of the tumor wall is calculated using the second position information. Then, the inner surface of the tumor wall is expanded with the center of gravity as the center to more accurately define the boundary of the tumor wall, thereby obtaining a new expanded surface, i.e., the third inner surface of the tumor wall. Specifically, the evolving inner surface of the tumor wall is calculated. Center of gravity of the enclosed volume And let the surface by The third aneurysm wall inner surface is obtained by expanding a certain distance as the center.

[0158] Step S4063, determining a target region in the target tumor wall inner layer surface that is surrounded by the third tumor wall inner layer surface; and determining third vertices in the target region, and determining each third vertex as an initial vertex set.

[0159] Obtain the inner surface of the target tumor wall The inner surface of the third aneurysm wall The enclosed part is the target area. Further, the inner surface of the target tumor wall is determined The third vertices in the target area on the three-dimensional mesh model are determined as the initial vertex set. As a result of initial tumor wall separation.

[0160] Step S4064, repairing the initial vertex set according to the reconstruction closing operation to obtain a repaired target vertex set.

[0161] Using the reconstruction closing operation of graph morphology, the initial vertex set Repair holes and use corrosion to The redundant selected vertices generated by the expansion are removed to obtain the repaired target vertex set Specifically, formula (11) is the k-neighborhood dilation in graph morphology, formula (12) is the k-neighborhood erosion, and formula (13) is the formula for the k-neighborhood closing operation, where is the vertex set to be operated on.

[0162]

[0163] Further, get the target vertex set Then, these vertices are Send out rays from the center to each third vertex, and record the inner surface of the evolving tumor wall hit by the corresponding rays closest to the third vertex itself The coordinates on the image are used to update the position of each third vertex to the specific position hit by the emitted ray. Therefore, it can help to accurately update the position of each third vertex to its actual position on the inner surface of the tumor wall in the image, thereby providing an accurate data basis for subsequent analysis and processing.

[0164] Step S4065: Perform signal strength mapping analysis on the target vertex set to obtain corresponding quantitative indicators.

[0165] The signal intensity mapping analysis is performed on the target vertex set. Specifically, a certain attribute or feature of the tumor wall is quantitatively analyzed, such as density, thickness, signal intensity, etc. Through this analysis, quantitative indicators of tumor wall characteristics can be obtained, which is helpful for further medical diagnosis or research.

[0166] In some optional implementations, the above step S4065 includes:

[0167] Step a1, calculating the second normal vector of the third vertex.

[0168] The surface normal vector, ie, the second normal vector, is calculated for each third vertex. Specifically, a weighted average normal vector may be used to obtain the normal vector of each third vertex by performing a weighted average based on the normal vectors of the connected facets.

[0169] Step a2: creating a sampling spoke of a preset length from the third vertex along the direction of the second normal vector.

[0170] For each third vertex, a set of sampling spokes of preset length is created along the direction of its normal vector. Specifically, the sampling spokes are a tool for sampling the surrounding area to obtain an estimated value of the surface signal strength. For example, the spoke sampling range can be [0,1.2] mm outward of the vertex normal vector.

[0171] Step a3: Use sampling spokes to perform sampling on the second target sequence and the third target sequence after trilinear interpolation to obtain sampling results.

[0172] The trilinear interpolation technique is used to interpolate the sampling spokes on the second target sequence and the third target sequence, so that the signal strength of the area around the third vertex can be estimated more accurately. Specifically, each sampling spoke is interpolated on the second target sequence after trilinear interpolation. and the third target sequence Sampled separately.

[0173] Step a4: determine the maximum value among the sampling results as the signal strength sampling value corresponding to the third vertex.

[0174] For each third vertex, the maximum value is found from the interpolation sampling results obtained above as the signal intensity (SI) sampling value of the third vertex, which helps to eliminate the numerical deviation caused by imaging problems, because selecting the maximum value can improve the stability and accuracy of sampling.

[0175] Step a5, determining a quantitative index according to the signal strength sampling value.

[0176] Using the obtained signal intensity sampling values, some quantitative indicators, such as the enhancement degree of the aneurysm wall, can be calculated and determined. Specifically, the quantitative indicators can help analyze the characteristics and imaging manifestations of the aneurysm wall.

[0177] Furthermore, after obtaining the quantitative results, the manually selected WEI reference points can be compared and correlated with the reference points obtained by the automatic algorithm to verify the effectiveness of the method. Therefore, when there is good consistency with the indicators obtained by manual annotation and the three-dimensional visualization results can reflect the degree and range of enhancement of the aneurysm wall, it can be used to assist manual aneurysm diagnosis in clinical practice.

[0178] The processing method of the magnetic resonance image sequence provided by the embodiment of the present invention realizes the extraction of quantitative indicators of aneurysms by analyzing the signal intensity mapping of the inner surface of the target aneurysm wall. These indicators include the position information of the second vertex, the calculation of the center of gravity of the second volume, and the expansion and repair of the inner surface of the aneurysm wall based on the center of gravity, which helps to accurately quantify the morphological characteristics and internal structure of the aneurysm and provide objective data support for clinical evaluation. By determining the second and third vertex sets of the inner surface of the evolving aneurysm wall and performing closed operation repair, complex aneurysm wall structures and boundary conditions can be effectively processed. This multi-level vertex set screening and repair process helps to reduce noise and unnecessary structures in the data, and improves the accuracy and reliability of quantitative analysis. Mapping and analyzing the signal intensity of the target vertex set is a key step in evaluating the tissue characteristics of the inner surface of the aneurysm wall. Through this analysis, signal intensity information of different parts can be obtained, which further helps doctors understand the nature, size and degree of impact of the lesion on the surrounding tissues, and provides an important reference for clinical decision-making. Therefore, the integration of multiple algorithms and data processing techniques, including geometric information of the inner surface of the evolving aneurysm wall, volume calculation, dilation operation, vertex patching, and signal intensity mapping, helps to fully understand and analyze the complex structure of aneurysms and provide a comprehensive quantitative evaluation.

[0179] In this embodiment, a processing device for a magnetic resonance image sequence is also provided, and the device is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made are not repeated here. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0180] This embodiment provides a processing device for a magnetic resonance image sequence, such as Figure 5 As shown, including:

[0181] The acquisition module 501 is used to acquire multiple magnetic resonance image sequences corresponding to the target blood vessel region, and to acquire aneurysm information pre-marked in any magnetic resonance image sequence; wherein the multiple magnetic resonance image sequences are multiple images acquired during the magnetic resonance imaging process, and the aneurysm information is used to characterize the aneurysm position in the target blood vessel region.

[0182] The registration module 502 is used to register the multiple magnetic resonance image sequences to the same sequence space, and to crop the multiple registered magnetic resonance image sequences according to the aneurysm information to obtain a target image corresponding to the aneurysm; wherein the same sequence space is a space corresponding to any sequence in the multiple magnetic resonance image sequences, and the target image is used to characterize the aneurysm and the first vascular structure surrounding it.

[0183] The processing module 503 is used to process the target image and determine the blood vessel lumen segmentation mask corresponding to the target image. The blood vessel lumen segmentation mask is used to characterize the contours of all blood vessel lumens in the target image.

[0184] The construction module 504 is used to extract the aneurysm information included in the target image and construct the initial aneurysm wall inner surface of the aneurysm according to the aneurysm information.

[0185] The reconstruction module 505 is used to extract features from the initial inner surface of the aneurysm wall, reconstruct the initial inner surface of the aneurysm wall according to the extracted features, and obtain the evolved inner surface of the aneurysm wall; and perform three-dimensional reconstruction of the aneurysm according to the evolved inner surface of the aneurysm wall and the vascular lumen segmentation mask to obtain the target inner surface of the aneurysm wall.

[0186] The quantification module 506 is used to analyze the signal intensity mapping of the inner surface of the target aneurysm wall to determine the quantitative index of the aneurysm.

[0187] In some optional embodiments, the plurality of magnetic resonance image sequences include a first magnetic resonance image sequence, a second magnetic resonance image sequence, and a third magnetic resonance image sequence; the first magnetic resonance image sequence is used to display a second vascular structure corresponding to the target vascular region, the second magnetic resonance image sequence is used to display the anatomical structure and signal intensity difference of the tissue included in the target vascular region, and the third magnetic resonance image sequence is used to display the contrast enhancement effect of the aneurysm corresponding to the target vascular region after adding a contrast agent; the registration module 502 includes:

[0188] The preprocessing submodule is used to preprocess the first magnetic resonance image sequence, the second magnetic resonance image sequence and the third magnetic resonance image sequence.

[0189] The registration submodule is used to register the preprocessed first magnetic resonance image sequence, the second magnetic resonance image sequence and the third magnetic resonance image sequence according to a preset medical image registration library, so as to register the first magnetic resonance image sequence and the second magnetic resonance image sequence to the third magnetic resonance image sequence.

[0190] In some optional implementations, the registration module 502 further includes:

[0191] The cropping submodule is used to crop the registered first magnetic resonance image sequence, the second magnetic resonance image sequence and the third magnetic resonance image sequence according to the aneurysm information to obtain a first target sequence corresponding to the first magnetic resonance image sequence, a second target sequence corresponding to the second magnetic resonance image sequence and a third target sequence corresponding to the third magnetic resonance image sequence.

[0192] In some optional implementations, the processing module 503 includes:

[0193] The first processing submodule is used to process the first target sequence to determine a white blood mask, where the white blood mask is used to characterize the position and shape of all blood vessels in the target image.

[0194] The second processing submodule is used to perform Gaussian filter edge enhancement processing on the second target sequence and the third target sequence, and synthesize the second target sequence and the third target sequence after the Gaussian filter edge enhancement processing to generate a synthesized image.

[0195] The first acquisition submodule is used to acquire a preset pixel threshold, where the pixel threshold is used to distinguish the blood vessel lumen in the synthesized image.

[0196] The third processing submodule is used to process the synthesized image according to the pixel threshold to obtain a binary mask image.

[0197] The first extraction submodule is used to extract a blood vessel lumen segmentation mask from the mask image according to the white blood mask and the mask image.

[0198] In some optional implementations, the construction module 504 includes:

[0199] The first conversion submodule is used to convert aneurysm information into a three-dimensional convex hull.

[0200] The second conversion submodule is used to convert the three-dimensional convex hull into the inner surface of the first tumor wall.

[0201] The fourth processing submodule is used to perform smoothing processing on the first tumor wall inner layer surface to obtain the second tumor wall inner layer surface.

[0202] The resampling submodule is used to perform isotropic voxel resampling on the inner surface of the second tumor wall to obtain a three-dimensional voxel data set.

[0203] The construction submodule is used to construct the initial inner surface of the aneurysm wall according to the three-dimensional voxel data set.

[0204] In some optional implementations, the reconstruction module 505 includes:

[0205] The second extraction submodule is used to extract the first position information and edge information of the first vertex in the initial tumor wall inner surface, and process the first position information and edge information to obtain the first normal vector and evolution direction of the first vertex.

[0206] The first calculation submodule is used to perform trilinear interpolation on the composite image to obtain an interpolation image function; and calculate the directional derivative in the evolution direction according to the interpolation image function and the first normal vector, and determine the maximum value of the directional derivative within a preset distance around the first vertex.

[0207] The second acquisition submodule is used to acquire a preset target displacement distance, and determine a value obtained by adding the target displacement distance to the maximum value as a target movement amount corresponding to the first vertex.

[0208] The updating submodule is used to update the first position information according to the target movement amount to obtain the target position information corresponding to the first vertex.

[0209] The first reconstruction submodule is used to reconstruct the initial inner surface of the tumor wall according to the target position information until the initial inner surface of the tumor wall converges to obtain the evolved inner surface of the tumor wall.

[0210] In some optional implementations, the reconstruction module 505 further includes:

[0211] The sampling submodule is used to perform voxel sampling on the first volume enclosed by the inner surface of the evolving tumor wall to obtain voxel data.

[0212] The mapping submodule is used to map the voxel data to the third target sequence to obtain a tumor cavity mask, and the tumor cavity mask is used to characterize the cavity area inside the aneurysm.

[0213] The merging submodule is used to merge the aneurysm cavity mask and the blood vessel lumen segmentation mask to obtain an inner cavity mask, which is a mask containing the complete internal structure of the aneurysm.

[0214] The second reconstruction submodule is used to perform three-dimensional reconstruction of the aneurysm according to the lumen mask to obtain the inner surface of the target aneurysm wall.

[0215] In some optional implementations, the quantization module 506 includes:

[0216] The third extraction submodule is used to extract the second position information of the second vertex in the inner surface of the evolving tumor wall.

[0217] The second calculation submodule is used to calculate the center of gravity of the second volume enclosed by the evolving tumor wall inner surface according to the second position information, and expand the evolving tumor wall inner surface around the center of gravity to obtain an expanded third tumor wall inner surface.

[0218] The determination submodule is used to determine a target area in the target tumor wall inner layer surface surrounded by the third tumor wall inner layer surface; and determine the third vertices in the target area, and determine each third vertex as an initial vertex set.

[0219] The patching submodule is used to patch the initial vertex set according to the reconstruction closing operation to obtain a patched target vertex set.

[0220] The quantization submodule is used to perform a mapping analysis of the signal strength of the target vertex set to obtain corresponding quantitative indicators.

[0221] In some optional implementations, the quantization submodule includes:

[0222] A calculation unit, used for calculating a second normal vector of a third vertex.

[0223] The creation unit is used to create a sampling spoke of a preset length from the third vertex along the second normal vector direction.

[0224] The sampling unit is used to perform sampling on the second target sequence and the third target sequence after trilinear interpolation by using the sampling spokes to obtain a sampling result.

[0225] The first determining unit is used to determine the maximum value among the sampling results as the signal strength sampling value corresponding to the third vertex.

[0226] The second determining unit is used to determine the quantitative index according to the signal strength sampling value.

[0227] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0228] The processing device for the magnetic resonance image sequence in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0229] The processing device of the magnetic resonance image sequence provided by the embodiment of the present invention obtains and utilizes multiple magnetic resonance image sequences, comprehensively displays the target vascular area and the aneurysm information therein, and can provide more comprehensive and multi-angle information, which is helpful for more accurate analysis and diagnosis of aneurysms. By accurately aligning multiple magnetic resonance image sequences and cropping according to the pre-marked aneurysm information, the target image corresponding to the aneurysm is obtained. This ensures that the spatial position and anatomical structure are maintained consistently between different images, providing a reliable basis for subsequent analysis and processing. The information of the aneurysm in the target image is extracted and analyzed, and the initial inner surface of the aneurysm wall is constructed based on this information. Through feature extraction and reconstruction technology, the inner surface of the evolving aneurysm wall can be presented more finely, which helps doctors to have a deeper understanding of the morphology and structure of the aneurysm. The three-dimensional reconstruction of the aneurysm is performed based on the inner surface of the evolving aneurysm wall and the segmentation mask of the vascular lumen, which not only provides the spatial information of the aneurysm, but also can quantitatively analyze the characteristics and signal intensity mapping of the aneurysm, thereby providing a scientific basis to support clinical decision-making.

[0230] The embodiment of the present invention also provides a computer device having the above Figure 5 A device for processing a magnetic resonance image sequence is shown.

[0231] See also Figure 6 , Figure 6 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 6 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 6 A processor 10 is taken as an example.

[0232] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0233] The memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.

[0234] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0235] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0236] The computer device also includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 6 The example of connecting through bus is taken in the following.

[0237] The input device 30 can receive input digital or character information, and generate key signal input related to the user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator bar, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0238] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0239] A part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of the computer program instruction in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc., and accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.

[0240] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for processing a magnetic resonance image sequence, characterized in that: The method comprises: Acquire multiple magnetic resonance image sequences corresponding to the target blood vessel region, and acquire aneurysm information pre-annotated in any of the magnetic resonance image sequences; wherein the multiple magnetic resonance image sequences are multiple images acquired during magnetic resonance imaging, and the aneurysm information is used to characterize the aneurysm position in the target blood vessel region; The multiple magnetic resonance image sequences are registered to the same sequence space, and the registered multiple magnetic resonance image sequences are cropped according to the aneurysm information to obtain a target image corresponding to the aneurysm; wherein the same sequence space is a space corresponding to any sequence in the multiple magnetic resonance image sequences, and the target image is used to characterize the aneurysm and a first vascular structure surrounding the aneurysm; Processing the target image to determine a blood vessel lumen segmentation mask corresponding to the target image, wherein the blood vessel lumen segmentation mask is used to characterize the contours of all blood vessel lumens in the target image; Extracting the aneurysm information included in the target image, and constructing an initial aneurysm wall inner surface of the aneurysm according to the aneurysm information; Extracting features from the initial inner surface of the aneurysm wall, reconstructing the initial inner surface of the aneurysm wall according to the extracted features to obtain an evolved inner surface of the aneurysm wall; and performing three-dimensional reconstruction of the aneurysm according to the evolved inner surface of the aneurysm wall and the vascular lumen segmentation mask to obtain a target inner surface of the aneurysm wall; The signal intensity mapping of the inner surface of the target aneurysm wall is analyzed to determine the quantitative index of the aneurysm.

2. The method according to claim 1, characterized in that The multiple magnetic resonance image sequences include a first magnetic resonance image sequence, a second magnetic resonance image sequence, and a third magnetic resonance image sequence; the first magnetic resonance image sequence is used to display a second blood vessel structure corresponding to the target blood vessel region, the second magnetic resonance image sequence is used to display an anatomical structure and a signal intensity difference of a tissue included in the target blood vessel region, and the third magnetic resonance image sequence is used to display a contrast enhancement effect of the aneurysm corresponding to the target blood vessel region after adding a contrast agent; The registering the multiple magnetic resonance image sequences to the same sequence space comprises: preprocessing the first magnetic resonance image sequence, the second magnetic resonance image sequence, and the third magnetic resonance image sequence; The preprocessed first magnetic resonance image sequence, the second magnetic resonance image sequence and the third magnetic resonance image sequence are registered according to a preset medical image registration library to register the first magnetic resonance image sequence and the second magnetic resonance image sequence to the third magnetic resonance image sequence.

3. The method according to claim 2, characterized in that The multiple registered magnetic resonance image sequences are cropped according to the aneurysm information to obtain a target image corresponding to the aneurysm, including: The first magnetic resonance image sequence, the second magnetic resonance image sequence, and the third magnetic resonance image sequence after registration are cropped according to the aneurysm information to obtain a first target sequence corresponding to the first magnetic resonance image sequence, a second target sequence corresponding to the second magnetic resonance image sequence, and a third target sequence corresponding to the third magnetic resonance image sequence.

4. The method according to claim 3, characterized in that Processing the target image to determine a blood vessel lumen segmentation mask corresponding to the target image includes: Processing the first target sequence to determine a white blood mask, wherein the white blood mask is used to characterize the position and shape of all blood vessels in the target image; performing Gaussian filter edge enhancement processing on the second target sequence and the third target sequence, and synthesizing the second target sequence and the third target sequence after the Gaussian filter edge enhancement processing to generate a synthesized image; Acquiring a preset pixel threshold, where the pixel threshold is used to distinguish a blood vessel lumen in the composite image; Processing the synthesized image according to the pixel threshold to obtain a binary mask image; The blood vessel lumen segmentation mask is extracted from the mask image according to the white blood mask and the mask image.

5. The method according to claim 1, characterized in that The step of constructing the initial inner surface of the aneurysm wall according to the aneurysm information comprises: Converting the aneurysm information into a three-dimensional convex hull; Converting the three-dimensional convex hull into the inner surface of the first tumor wall; Smoothing the first tumor wall inner layer surface to obtain a second tumor wall inner layer surface; performing isotropic voxel resampling on the inner surface of the second tumor wall to obtain a three-dimensional voxel data set; An initial inner surface of the aneurysm wall is constructed according to the three-dimensional voxel data set.

6. The method according to claim 5, characterized in that Extracting features from the initial inner surface of the tumor wall, and reconstructing the initial inner surface of the tumor wall according to the extracted features to obtain an evolved inner surface of the tumor wall, including: Extracting first position information and edge information of a first vertex in the inner surface of the initial tumor wall, and processing the first position information and the edge information to obtain a first normal vector and an evolution direction of the first vertex; Perform trilinear interpolation on the synthesized image to obtain an interpolation image function; and calculate a directional derivative in the evolution direction according to the interpolation image function and the first normal vector, and determine a maximum value of the directional derivative within a preset distance around the first vertex; Acquire a preset target displacement distance, and determine a value obtained by adding the target displacement distance to the maximum value as a target movement amount corresponding to the first vertex; The first position information is updated according to the target movement amount to obtain the target position information corresponding to the first vertex; The initial tumor wall inner layer surface is reconstructed according to the target position information until the initial tumor wall inner layer surface converges to obtain the evolved tumor wall inner layer surface.

7. The method according to claim 6, characterized in that Performing three-dimensional reconstruction of the aneurysm according to the evolved aneurysm wall inner surface and the blood vessel lumen segmentation mask to obtain a target aneurysm wall inner surface includes: Performing voxel sampling on a first volume enclosed by the inner surface of the evolving tumor wall to obtain voxel data; Mapping the voxel data to a third target sequence to obtain a cavity mask, wherein the cavity mask is used to characterize a cavity region inside the aneurysm; The aneurysm cavity mask and the blood vessel lumen segmentation mask are combined to obtain an inner cavity mask, wherein the inner cavity mask is a mask including the complete inner structure of the aneurysm; The aneurysm is reconstructed three-dimensionally according to the lumen mask to obtain the inner surface of the target aneurysm wall.

8. The method according to claim 7, characterized in that Analyzing the signal intensity mapping of the inner surface of the target aneurysm wall to determine the quantitative index of the aneurysm includes: Extracting second position information of a second vertex in the inner surface of the evolving tumor wall; Calculating the center of gravity of a second volume enclosed by the inner surface of the evolved tumor wall according to the second position information, and expanding the inner surface of the evolved tumor wall with the center of gravity as the center to obtain a third expanded inner surface of the tumor wall; Determine a target region in the target tumor wall inner layer surface that is surrounded by the third tumor wall inner layer surface; and determine third vertices in the target region, and determine each of the third vertices as an initial vertex set; Repairing the initial vertex set according to the reconstruction closing operation to obtain a repaired target vertex set; A signal strength mapping analysis is performed on the target vertex set to obtain the corresponding quantitative index.

9. The method according to claim 8, characterized in that The mapping analysis of the signal strength of the target vertex set to obtain the corresponding quantitative index includes: Calculate a second normal vector of the third vertex; Creating a sampling spoke of a preset length from the third vertex along the second normal vector direction; Using the sampling spokes to perform sampling on the second target sequence and the third target sequence after trilinear interpolation to obtain a sampling result; Determine the maximum value among the sampling results as the signal strength sampling value corresponding to the third vertex; The quantitative index is determined according to the signal strength sampling value.

10. A device for processing a magnetic resonance image sequence, characterized in that: The device comprises: An acquisition module, used to acquire a plurality of magnetic resonance image sequences corresponding to a target blood vessel region, and to acquire aneurysm information pre-annotated in any of the magnetic resonance image sequences; wherein the plurality of magnetic resonance image sequences are a plurality of images acquired during a magnetic resonance imaging process, and the aneurysm information is used to characterize the location of the aneurysm in the target blood vessel region; A registration module, used for registering the multiple magnetic resonance image sequences to the same sequence space, and cropping the multiple registered magnetic resonance image sequences according to the aneurysm information to obtain a target image corresponding to the aneurysm; wherein the same sequence space is a space corresponding to any sequence in the multiple magnetic resonance image sequences, and the target image is used to characterize the aneurysm and a first vascular structure around it; A processing module, used to process the target image to determine a blood vessel lumen segmentation mask corresponding to the target image, wherein the blood vessel lumen segmentation mask is used to characterize the contours of all blood vessel lumens in the target image; A construction module, used for extracting the aneurysm information included in the target image, and constructing an initial aneurysm wall inner surface of the aneurysm according to the aneurysm information; A reconstruction module is used to extract features from the initial inner surface of the aneurysm wall, reconstruct the initial inner surface of the aneurysm wall according to the extracted features, and obtain an evolved inner surface of the aneurysm wall; and perform three-dimensional reconstruction of the aneurysm according to the evolved inner surface of the aneurysm wall and the vascular lumen segmentation mask to obtain a target inner surface of the aneurysm wall; The quantification module is used to analyze the signal intensity mapping of the inner surface of the target aneurysm wall to determine the quantitative index of the aneurysm.

11. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method for processing a magnetic resonance image sequence according to any one of claims 1 to 9 by executing the computer instructions.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the method for processing a magnetic resonance image sequence according to any one of claims 1 to 9.

13. A computer program product, characterized in that The method comprises computer instructions for causing a computer to execute the method for processing a magnetic resonance image sequence according to any one of claims 1 to 9.

Images