Segmentation method and apparatus for magnetic resonance vascular architectural imaging
By combining magnetic resonance angiography and deep learning algorithms, the problems of insufficient tumor segmentation accuracy and neglect of vascular structures have been solved, enabling precise segmentation and quantitative analysis of tumor regions, and providing a scientific basis for tumor diagnosis and treatment.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- SHENZHEN INST OF ADVANCED TECH
- Filing Date
- 2025-01-06
- Publication Date
- 2026-07-02
AI Technical Summary
Existing tumor segmentation methods have low segmentation accuracy when tumor boundaries are blurred, tumors are small, or tumors are mixed with surrounding tissues. They also lack sufficient consideration and quantitative analysis of vascular structures, resulting in inaccurate segmentation.
A deep learning method based on magnetic resonance vascular structure imaging was adopted, combining convolutional neural networks (CNN) and U-Net architecture. By preprocessing VAI images, a deep learning model was trained, and the vascular structure information was used to automatically segment tumor regions and perform quantitative analysis.
It enables precise segmentation and quantitative assessment of tumor regions, and provides key indicators such as vascular density and microvascular density, supporting early diagnosis, staging, and monitoring of treatment effects of tumors.
Smart Images
Figure CN2025070815_02072026_PF_FP_ABST
Abstract
Description
Segmentation methods and devices for magnetic resonance vascular imaging Technical Field
[0001] This invention relates to the field of medical imaging technology, and more specifically, to a segmentation method and apparatus for magnetic resonance vascular structure imaging. Background Technology
[0002] Vascular imaging (VAI) is a magnetic resonance imaging (MRI) technique that uses various imaging sequences (such as gradient echo (GE) and spin echo (SE) to obtain quantitative information about vascular structures. VAI can provide information about the size of blood vessels and microvessel density in tumors and surrounding tissues, which is crucial for assessing tumor growth, stage, and treatment efficacy. Traditional VAI techniques typically rely on manual or semi-automated segmentation methods and lack precise automatic segmentation of tumor regions.
[0003] Currently available imaging methods for brain tumors include dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI), magnetic resonance angiography (MRA), and functional MRI (fMRI). These methods enhance image clarity by introducing contrast agents, and have certain advantages, particularly in assessing angiogenesis and tumor blood flow.
[0004] However, most existing tumor segmentation methods are based on traditional image processing techniques or deep learning models. Traditional methods, such as thresholding, region growing, and edge detection, can segment tumor regions in some cases, but their accuracy and robustness are poor. In recent years, the deep learning-based 3D U-Net model has been applied to automatic segmentation tasks in medical images, and it can better solve the segmentation problem of complex tumor morphologies. However, existing automatic segmentation methods often fail to fully utilize the vascular information in VAI images to optimize tumor region segmentation, and quantitative analysis of microvascular density around the tumor remains limited.
[0005] In summary, the shortcomings of existing technologies are:
[0006] 1. Insufficient segmentation accuracy: Although existing automatic segmentation methods can handle the basic segmentation of tumors, the segmentation accuracy is still low when the tumor boundaries are blurred, the tumor is small, or the tumor is mixed with surrounding tissues.
[0007] 2. Insufficient consideration of vascular structures: Most existing segmentation methods do not fully consider the impact of vascular structures on the tumor region, especially the indicative role of vascular features in VAI images on tumor boundaries and microvessel density.
[0008] 3. Lack of quantitative analysis: Most existing segmentation techniques focus on qualitative description and lack precise quantitative analysis methods, which cannot provide a scientific basis for tumor treatment. Summary of the Invention
[0009] This invention provides a segmentation method and apparatus for magnetic resonance vascular structure imaging, which at least solves the technical problem of low segmentation accuracy in existing tumor images.
[0010] According to an embodiment of the present invention, a segmentation method for magnetic resonance vascular structure imaging is provided, comprising the following steps:
[0011] S101: Perform data preprocessing on the VAI image set;
[0012] S102: Construct a deep learning model, train the deep learning model using the preprocessed VAI image set, and optimize the model parameters;
[0013] S103: Use the trained and optimized deep learning model to automatically segment the tumor region of the new VAI image and generate the segmentation result.
[0014] Furthermore, the method also includes:
[0015] S104: Perform quantitative analysis on the segmentation results.
[0016] Furthermore, step S104 specifically includes:
[0017] The segmentation results are used to calculate the vascular density and microvascular density indices of the tumor region, and then combined with other image data from the VAI image set for comprehensive evaluation.
[0018] Furthermore, step S101 specifically includes:
[0019] The VAI image set was standardized to remove noise and enhance vascular structure information.
[0020] Furthermore, vascular structural information includes vascular density, vascular size, and microvascular density.
[0021] Furthermore, step S102 specifically includes:
[0022] The model is trained using a 3D U-Net network architecture, combined with vascular information from VAI images.
[0023] Furthermore, high-quality labeled datasets are used during training, and the accuracy of tumor segmentation is improved by optimizing the loss function.
[0024] Further, in step S101, the VAI image set includes the patient's VAI images and other relevant imaging data, including T1-weighted imaging and SAGE imaging.
[0025] According to another embodiment of the present invention, a segmentation device for magnetic resonance vascular structure imaging is provided, comprising:
[0026] The preprocessing unit is used to preprocess the VAI image set.
[0027] The training unit is used to build deep learning models, train them using preprocessed VAI image sets, and optimize model parameters.
[0028] The segmentation unit is used to automatically segment tumor regions in new VAI images using a trained and optimized deep learning model, and generate segmentation results.
[0029] Furthermore, the device also includes:
[0030] The quantitative analysis unit is used to perform quantitative analysis on the segmentation results.
[0031] A storage medium storing a program file capable of implementing any of the above-described segmentation methods for magnetic resonance vascular structure imaging.
[0032] A processor for running a program, wherein the program executes the segmentation method for magnetic resonance vascular structure imaging of any of the above-mentioned methods during runtime.
[0033] The segmentation method and apparatus for magnetic resonance vascular structure imaging in this embodiment of the invention, by combining deep learning models such as convolutional neural networks (CNN), can automatically learn the features of vascular structures and accurately segment microvascular changes in tumor regions. Attached Figure Description
[0034] Figure 1 is a flowchart of the segmentation method for magnetic resonance vascular structure imaging of the present invention;
[0035] Figure 2 is a preferred flowchart of the segmentation method for magnetic resonance vascular structure imaging of the present invention;
[0036] Figure 3 is a framework diagram of the segmentation method for magnetic resonance vascular structure imaging of the present invention;
[0037] Figure 4 is a block diagram of the segmentation device for magnetic resonance vascular structure imaging of the present invention;
[0038] Figure 5 is a preferred module diagram of the segmentation device for magnetic resonance vascular structure imaging of the present invention. Detailed Implementation
[0039] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.
[0040] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0041] Example 1
[0042] According to an embodiment of the present invention, a segmentation method for magnetic resonance vascular structure imaging is provided, as shown in Figure 1, comprising the following steps:
[0043] S101: Perform data preprocessing on the VAI image set;
[0044] S102: Construct a deep learning model, train the deep learning model using the preprocessed VAI image set, and optimize the model parameters;
[0045] S103: Use the trained and optimized deep learning model to automatically segment the tumor region of the new VAI image and generate the segmentation result.
[0046] The segmentation method for magnetic resonance vascular structure imaging in this embodiment of the invention, by combining deep learning models such as convolutional neural networks (CNN), can automatically learn the features of vascular structures and accurately segment microvascular changes in tumor regions.
[0047] As shown in Figure 2, the method also includes:
[0048] S104: Perform quantitative analysis on the segmentation results.
[0049] Existing tumor segmentation methods generally rely on traditional image processing techniques or deep learning algorithms. However, in cases of blurred tumor boundaries or low image contrast, they may fail to accurately segment the tumor region. Furthermore, the significant variations in vascular density within the tumor region increase the segmentation difficulty, making it challenging for traditional segmentation algorithms to effectively distinguish tumor areas from normal tissue. The technical problem this invention aims to solve is how to effectively and accurately perform automatic tumor region segmentation on brain tumor imaging data based on vascular imaging (VAI), and to evaluate key indicators such as tumor vascular structure and microvascular density using the segmentation results, providing effective data support for early tumor diagnosis, staging, and treatment monitoring.
[0050] This invention addresses the limitations of traditional tumor segmentation methods by proposing a novel tumor segmentation method based on magnetic resonance vascular imaging (VAI) technology. By combining deep learning models such as convolutional neural networks (CNNs), it can automatically learn the features of vascular structures and accurately segment tumors based on microvascular changes.
[0051] The purpose of this invention is to provide an automated tumor segmentation method based on VAI (vascular structure imaging). By using deep learning algorithms and combining vascular structure information, the method can accurately segment tumor regions and perform quantitative analysis, such as vascular density and microvascular density, providing reliable support for early diagnosis, staging, and monitoring of treatment effects of tumors.
[0052] The basic content of the technical solution of this invention is as follows:
[0053] This invention employs a deep learning algorithm based on visual imaging of blood vessels (VAI) for automatic segmentation of brain tumors. By using a convolutional neural network (CNN) or U-Net architecture, combined with vascular structure information (such as vessel density, vessel size, and microvessel density) from VAI images, accurate segmentation and quantitative analysis of tumor regions are achieved. Specific methods include:
[0054] 1. Data preprocessing: Standardize the VAI images to remove noise and enhance vascular structure information.
[0055] 2. Deep learning model training: Train a deep learning model using VAI images and their labeled dataset, optimize model parameters, and improve segmentation accuracy.
[0056] 3. Automatic tumor segmentation: The trained model is used to automatically segment the tumor region in the new VAI image.
[0057] 4. Quantitative analysis: By calculating indicators such as vascular density and microvascular density in the tumor region based on the segmentation results, a basis for monitoring tumor growth and treatment in clinical practice is provided.
[0058] The technical solution of the present invention is described in detail below:
[0059] This invention was tested on over 100 brain tumor patients using 3T and 5T MRI machines, and animal experiments were conducted on glioma-bearing mice using a 9.4T MRI machine to verify the performance of magnetic resonance vascular structure imaging technology. Furthermore, U-Net was incorporated to segment the tumor region, verifying the accuracy and effectiveness of the technology.
[0060] Referring to Figure 3, the main technical contents of this invention are as follows:
[0061] 1. Data Acquisition and Preprocessing: First, acquire the patient's VAI images and other relevant imaging data (such as T1-weighted imaging, SAGE imaging, etc.). These data are then denoised, standardized, and registered to ensure alignment of different imaging data.
[0062] 2. Construction and Training of Deep Learning Model: A 3D U-Net network architecture was used, combined with vascular information from VAI images for model training. High-quality labeled datasets were used during training, and the accuracy of tumor segmentation was improved by optimizing the loss function.
[0063] 3. Automatic Segmentation and Quantitative Analysis: After model training, it is applied to new VAI image data to automatically segment tumor regions. Based on the segmentation results, further quantitative analysis is performed to calculate features such as vascular density and microvascular density in the tumor region, and a comprehensive evaluation is conducted in conjunction with other image data.
[0064] 4. Clinical Applications: Through precise segmentation and quantitative analysis of tumor regions, it helps doctors make early diagnoses, stage tumors, monitor treatment effects, and provide a scientific basis for the development of personalized treatment plans.
[0065] Through the above technical solutions, the present invention can provide a more scientific, non-invasive and efficient technical means for the early diagnosis, treatment effect monitoring and efficacy evaluation of brain tumors, thereby improving the accuracy of tumor treatment and reducing the risk to patients.
[0066] This invention features precise tumor segmentation: By combining VAI imaging with deep learning algorithms, this invention can extract detailed tumor and microvascular features from vascular structure imaging and perform precise tumor segmentation.
[0067] This invention features precise quantitative analysis: compared with existing DCE-MRI and MRA technologies, this invention can provide more accurate quantitative data, such as important indicators like microvascular density and vascular size (e.g., vascular radius) in tumor regions, overcoming the limitations of existing technologies that mainly focus on large vascular structures.
[0068] This invention provides a comprehensive assessment of microvascular status: it not only focuses on angiogenesis within the tumor but also quantitatively evaluates microvascular changes in the surrounding tumor region. This integrated analysis can provide a more comprehensive basis for tumor staging, treatment planning, and efficacy evaluation.
[0069] The key points and areas to be protected in this invention are:
[0070] 1. Application of VAI vascular structure information: This invention is the first to combine VAI vascular structure information with a deep learning model to achieve accurate segmentation of tumor regions.
[0071] 2. Automated tumor segmentation method: This invention provides an automated tumor segmentation method that avoids the tediousness and inaccuracy of manual segmentation.
[0072] 3. Quantitative analysis of vascular density and microvascular density: This invention provides support for the diagnosis and treatment of tumors by quantitatively analyzing key indicators such as vascular density and microvascular density in the tumor region.
[0073] Compared with the prior art, the advantages of the present invention are:
[0074] 1. Integrating Vascular Structure Information: Most existing technologies focus on image-based tumor segmentation, neglecting the influence of vascular structure on tumor boundaries and microvessel density. This invention, by integrating vascular information from VAI images, can more accurately identify tumor regions.
[0075] 2. High degree of automation: Compared with traditional manual or semi-automatic segmentation methods, the automated tumor segmentation method of this invention is more efficient and accurate, reducing human error.
[0076] 3. Strong quantitative analysis capability: Most existing technologies are limited to qualitative analysis, while this invention can provide accurate quantitative data, such as vascular density and microvascular density, to help doctors make more scientific treatment decisions.
[0077] This invention has been proven feasible through experiments, simulations, and applications. Its feasibility has been verified through simulation experiments. Through training and testing on multiple tumor datasets, experimental results show that this method can achieve high-precision segmentation in different types of tumors and successfully extract quantitative indicators such as vascular density and microvascular density in tumor regions.
[0078] The modified design (alternative) and other uses of this invention are as follows: The deep learning model of this invention can be adjusted according to different tumor types to adapt to segmentation tasks of other types of tumors (such as lung cancer, breast cancer, etc.). In addition to brain tumor segmentation, the technology of this invention can also be widely applied to automatic segmentation tasks of other medical images, such as microvascular assessment of cardiovascular diseases and diabetes, and has broad application prospects.
[0079] Example 2
[0080] According to another embodiment of the present invention, a segmentation device for magnetic resonance vascular structure imaging is provided, as shown in FIG4, comprising:
[0081] Preprocessing unit 201 is used to preprocess the VAI image set;
[0082] Training unit 202 is used to build a deep learning model, train the deep learning model using the preprocessed VAI image set, and optimize the model parameters.
[0083] The segmentation unit 203 is used to automatically segment the tumor region of a new VAI image using a trained and optimized deep learning model, and generate segmentation results.
[0084] The segmentation device for magnetic resonance vascular structure imaging in this embodiment of the invention can automatically learn the features of vascular structures and accurately segment microvascular changes in tumor areas by combining deep learning models such as convolutional neural networks (CNN).
[0085] Referring to Figure 5, the device also includes:
[0086] The quantitative analysis unit 204 is used to perform quantitative analysis on the segmentation results.
[0087] Existing tumor segmentation methods generally rely on traditional image processing techniques or deep learning algorithms. However, in cases of blurred tumor boundaries or low image contrast, they may fail to accurately segment the tumor region. Furthermore, the significant variations in vascular density within the tumor region increase the segmentation difficulty, making it challenging for traditional segmentation algorithms to effectively distinguish tumor areas from normal tissue. The technical problem this invention aims to solve is how to effectively and accurately perform automatic tumor region segmentation on brain tumor imaging data based on vascular imaging (VAI), and to evaluate key indicators such as tumor vascular structure and microvascular density using the segmentation results, providing effective data support for early tumor diagnosis, staging, and treatment monitoring.
[0088] This invention addresses the limitations of traditional tumor segmentation methods by proposing a novel tumor segmentation method based on magnetic resonance vascular imaging (VAI) technology. By combining deep learning models such as convolutional neural networks (CNNs), it can automatically learn the features of vascular structures and accurately segment tumors based on microvascular changes.
[0089] The purpose of this invention is to provide an automated tumor segmentation device based on VAI (vascular structure imaging). By using deep learning algorithms and combining vascular structure information, the device can accurately segment tumor regions and perform quantitative analysis, such as vascular density and microvascular density, providing reliable support for early diagnosis, staging, and monitoring of treatment effects of tumors.
[0090] The basic content of the technical solution of this invention is as follows:
[0091] This invention employs a deep learning algorithm based on visual imaging of blood vessels (VAI) for automatic segmentation of brain tumors. By using a convolutional neural network (CNN) or U-Net architecture, combined with vascular structure information (such as vessel density, vessel size, and microvessel density) from VAI images, accurate segmentation and quantitative analysis of tumor regions are achieved. Specific methods include:
[0092] 1. Data preprocessing: Standardize the VAI images to remove noise and enhance vascular structure information.
[0093] 2. Deep learning model training: Train a deep learning model using VAI images and their labeled dataset, optimize model parameters, and improve segmentation accuracy.
[0094] 3. Automatic tumor segmentation: The trained model is used to automatically segment the tumor region in the new VAI image.
[0095] 4. Quantitative analysis: By calculating indicators such as vascular density and microvascular density in the tumor region based on the segmentation results, a basis for monitoring tumor growth and treatment in clinical practice is provided.
[0096] The technical solution of the present invention is described in detail below:
[0097] This invention was tested on over 100 brain tumor patients using 3T and 5T MRI machines, and animal experiments were conducted on glioma-bearing mice using a 9.4T MRI machine to verify the performance of magnetic resonance vascular structure imaging technology. Furthermore, U-Net was incorporated to segment the tumor region, verifying the accuracy and effectiveness of the technology.
[0098] Referring to Figure 3, the main technical contents of this invention are as follows:
[0099] 1. Data Acquisition and Preprocessing: First, acquire the patient's VAI images and other relevant imaging data (such as T1-weighted imaging, SAGE imaging, etc.). These data are then denoised, standardized, and registered to ensure alignment of different imaging data.
[0100] 2. Construction and Training of Deep Learning Model: A 3D U-Net network architecture was used, combined with vascular information from VAI images for model training. High-quality labeled datasets were used during training, and the accuracy of tumor segmentation was improved by optimizing the loss function.
[0101] 3. Automatic Segmentation and Quantitative Analysis: After model training, it is applied to new VAI image data to automatically segment tumor regions. Based on the segmentation results, further quantitative analysis is performed to calculate features such as vascular density and microvascular density in the tumor region, and a comprehensive evaluation is conducted in conjunction with other image data.
[0102] 4. Clinical Applications: Through precise segmentation and quantitative analysis of tumor regions, it helps doctors make early diagnoses, stage tumors, monitor treatment effects, and provide a scientific basis for the development of personalized treatment plans.
[0103] Through the above technical solutions, the present invention can provide a more scientific, non-invasive and efficient technical means for the early diagnosis, treatment effect monitoring and efficacy evaluation of brain tumors, thereby improving the accuracy of tumor treatment and reducing the risk to patients.
[0104] This invention features precise tumor segmentation: By combining VAI imaging with deep learning algorithms, this invention can extract detailed tumor and microvascular features from vascular structure imaging and perform precise tumor segmentation.
[0105] This invention features precise quantitative analysis: compared with existing DCE-MRI and MRA technologies, this invention can provide more accurate quantitative data, such as important indicators like microvascular density and vascular size (e.g., vascular radius) in tumor regions, overcoming the limitations of existing technologies that mainly focus on large vascular structures.
[0106] This invention provides a comprehensive assessment of microvascular status: it not only focuses on angiogenesis within the tumor but also quantitatively evaluates microvascular changes in the surrounding tumor region. This integrated analysis can provide a more comprehensive basis for tumor staging, treatment planning, and efficacy evaluation.
[0107] The key points and areas to be protected in this invention are:
[0108] 1. Application of VAI vascular structure information: This invention is the first to combine VAI vascular structure information with a deep learning model to achieve accurate segmentation of tumor regions.
[0109] 2. Automated tumor segmentation method: This invention provides an automated tumor segmentation method that avoids the tediousness and inaccuracy of manual segmentation.
[0110] 3. Quantitative analysis of vascular density and microvascular density: This invention provides support for the diagnosis and treatment of tumors by quantitatively analyzing key indicators such as vascular density and microvascular density in the tumor region.
[0111] Compared with the prior art, the advantages of the present invention are:
[0112] 1. Integrating Vascular Structure Information: Most existing technologies focus on image-based tumor segmentation, neglecting the influence of vascular structure on tumor boundaries and microvessel density. This invention, by integrating vascular information from VAI images, can more accurately identify tumor regions.
[0113] 2. High degree of automation: Compared with traditional manual or semi-automatic segmentation methods, the automated tumor segmentation method of this invention is more efficient and accurate, reducing human error.
[0114] 3. Strong quantitative analysis capability: Most existing technologies are limited to qualitative analysis, while this invention can provide accurate quantitative data, such as vascular density and microvascular density, to help doctors make more scientific treatment decisions.
[0115] This invention has been proven feasible through experiments, simulations, and applications. Its feasibility has been verified through simulation experiments. Through training and testing on multiple tumor datasets, experimental results show that this method can achieve high-precision segmentation in different types of tumors and successfully extract quantitative indicators such as vascular density and microvascular density in tumor regions.
[0116] The modified design (alternative) and other uses of this invention are as follows: The deep learning model of this invention can be adjusted according to different tumor types to adapt to segmentation tasks of other types of tumors (such as lung cancer, breast cancer, etc.). In addition to brain tumor segmentation, the technology of this invention can also be widely applied to automatic segmentation tasks of other medical images, such as microvascular assessment of cardiovascular diseases and diabetes, and has broad application prospects.
[0117] Example 3
[0118] A storage medium storing a program file capable of implementing any of the above-described segmentation methods for magnetic resonance vascular structure imaging.
[0119] Example 4
[0120] A processor for running a program, wherein the program executes the segmentation method for magnetic resonance vascular structure imaging of any of the above-mentioned methods during runtime.
[0121] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.
[0122] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.
[0123] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The system embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of units or modules may be electrical or other forms.
[0124] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0125] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.
[0126] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.
[0127] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.
Claims
1. A segmentation method for magnetic resonance vascular structure imaging, characterized in that, Includes the following steps: S101: Perform data preprocessing on the VAI image set; S102: Construct a deep learning model, train the deep learning model using the preprocessed VAI image set, and optimize the model parameters; S103: Use the trained and optimized deep learning model to automatically segment the tumor region of the new VAI image and generate the segmentation result.
2. The segmentation method for magnetic resonance vascular structure imaging according to claim 1, characterized in that, The method further includes: S104: Perform quantitative analysis on the segmentation results.
3. The segmentation method for magnetic resonance vascular structure imaging according to claim 2, characterized in that, Step S104 specifically includes: The segmentation results are used to calculate the vascular density and microvascular density indices of the tumor region, and then combined with other image data from the VAI image set for comprehensive evaluation.
4. The segmentation method for magnetic resonance vascular structure imaging according to claim 1, characterized in that, Step S101 specifically includes: The VAI image set was standardized to remove noise and enhance vascular structure information.
5. The segmentation method for magnetic resonance vascular structure imaging according to claim 4, characterized in that, Vascular structural information includes vascular density, vascular size, and microvascular density.
6. The segmentation method for magnetic resonance vascular structure imaging according to claim 1, characterized in that, Step S102 specifically includes: The model is trained using a 3D U-Net network architecture, combined with vascular information from VAI images.
7. The segmentation method for magnetic resonance vascular structure imaging according to claim 6, characterized in that, High-quality labeled datasets are used during training, and the accuracy of tumor segmentation is improved by optimizing the loss function.
8. The segmentation method for magnetic resonance vascular structure imaging according to claim 1, characterized in that, In step S101, the VAI image set includes the patient's VAI images and other relevant imaging data, including T1-weighted imaging and SAGE imaging.
9. A segmentation device for magnetic resonance vascular structure imaging, characterized in that, include: The preprocessing unit is used to preprocess the VAI image set. The training unit is used to build deep learning models, train them using preprocessed VAI image sets, and optimize model parameters. The segmentation unit is used to automatically segment tumor regions in new VAI images using a trained and optimized deep learning model, and generate segmentation results.
10. The segmentation device for magnetic resonance vascular structure imaging according to claim 9, characterized in that, The device further includes: The quantitative analysis unit is used to perform quantitative analysis on the segmentation results.