Human blood vessel mapping guide plate automatic generation method based on three-dimensional modeling

Abstract

The invention relates to the technical field of three-dimensional modeling, in particular to a human blood vessel mapping guide plate automatic generation method based on three-dimensional modeling. The method comprises the following steps: acquiring a patient focus tissue CT image and a patient focus tissue MRI image; fuzzy blood vessel area radiography optimization is carried out on the patient focus tissue CT image, and a standard patient focus tissue optimization CT image is generated; soft tissue identification is carried out on the patient focus tissue MRI image, and an MRI image soft tissue area image is generated; performing image merging on the standard patient focus tissue optimization CT image and the patient focus tissue MRI image to generate a patient focus tissue feature image; and performing preoperative plan image simulation based on the lesion tissue feature image of the patient to generate a preoperative plan image. Through multi-modal image fusion, AI automatic segmentation and function testing, the automatic generation precision and reliability of the three-dimensional blood vessel mapping guide plate are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of three-dimensional modeling, and in particular to a method for automatically generating a human blood vessel mapping guide plate based on three-dimensional modeling. Background Art

[0002] Initially, human vascular mapping guides mainly relied on two-dimensional medical imaging technologies, such as X-rays and CT scans, and vascular models were manually drawn through these images. However, this method is not only time-consuming and labor-intensive, but also easily affected by human factors, resulting in insufficient accuracy. With the emergence of three-dimensional imaging technology, especially the application of high-resolution CT and MRI technology, the stereoscopic information of medical images has become richer. This advancement enables medical image processing and analysis to be performed in three-dimensional space, thereby improving the ability to accurately model vascular structures. With the generation of these three-dimensional data, medical image processing technology has also begun to develop a series of automated tools to improve the efficiency and accuracy of model generation. Entering the 21st century, the rise of deep learning and artificial intelligence has provided new solutions for three-dimensional modeling. In particular, the application of deep learning models such as convolutional neural networks (CNN) and U-Net has significantly improved the automation level of vascular segmentation and mapping. AI technology can automatically identify and segment vascular structures and perform accurate three-dimensional reconstruction to generate high-quality vascular guides. However, the current traditional placement of radioactive particles still relies on manual operations, and the placement accuracy is low. At the same time, it is easy to damage the complex subcutaneous vascular environment, which leads to low safety. Summary of the invention

[0003] Based on this, it is necessary to provide a method for automatically generating a human blood vessel mapping guide based on three-dimensional modeling to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a method for automatically generating a human blood vessel mapping guide based on three-dimensional modeling is provided, the method comprising the following steps: Step S1: obtaining a CT image of the patient's lesion tissue and an MRI image of the patient's lesion tissue; performing fuzzy vascular area angiography optimization on the CT image of the patient's lesion tissue to generate a standard patient's lesion tissue optimized CT image; performing soft tissue identification on the MRI image of the patient's lesion tissue to generate an MRI image of the soft tissue area; merging the standard patient's lesion tissue optimized CT image and the patient's lesion tissue MRI image to generate a patient's lesion tissue characteristic image; Step S2: performing preoperative planning image simulation based on the patient's lesion tissue characteristic image to generate a preoperative planning image; performing image matching results on the preoperative planning image and the patient's lesion tissue characteristic image, and when the image matching result between the preoperative planning image and the patient's lesion tissue characteristic image is true, performing AI tissue segmentation on the preoperative planning image based on the patient's lesion tissue characteristic image to generate AI tissue segmentation data; when the image matching result between the preoperative planning image and the patient's lesion tissue characteristic image is false, re-planning the preoperative planning image; Step S3: Screen the puncture needle track and the marked dose range of the preoperative planning image through AI tissue segmentation data to obtain the puncture needle track data and the marked dose range data; perform entity conversion on the puncture needle track data and the marked dose range data to generate the puncture needle track entity data and the marked dose range entity data; perform vascular projection entity conversion on the puncture needle track entity data through the marked dose range entity data to generate vascular mapping entity data; perform three-dimensional reconstruction on the puncture needle track entity data, the local skin entity conversion data and the vascular mapping entity data to generate an initial three-dimensional planning guide; Step S4: cutting and printing the initial three-dimensional planned guide based on the preset needle channel cutting distance to generate a three-dimensional planned printed guide; performing a vascular function matching test on the three-dimensional planned printed guide to generate guide vascular function matching test data; optimizing the three-dimensional planned printed guide using the guide vascular function matching test data to generate a three-dimensional planned optimized guide.

[0005] The present invention can fuse blood vessel and soft tissue information and generate more comprehensive lesion tissue feature images by acquiring and optimizing CT and MRI images of patients. This multimodal image fusion improves the details and accuracy of the image and provides a high-quality data basis for subsequent processing. Fuzzy optimization of the blood vessel area in the CT image can highlight the blood vessel structure, reduce noise interference, and improve the clarity and accuracy of blood vessel identification. Identifying soft tissue in MRI images can clearly mark the soft tissue area and provide necessary information for the accurate reconstruction of blood vessels and tissue structures. By generating preoperative planning images based on feature images, it is helpful for doctors to carry out detailed surgical planning, judge the matching results of preoperative planning images and feature images, and ensure that the preoperative plan is consistent with the actual image data. When the matching result is true, AI is used to perform tissue segmentation to reduce human errors; when the matching result is false, re-planning is performed to ensure the reliability of the preoperative plan. Automatic tissue segmentation improves the speed and accuracy of segmentation, reduces the workload and errors of manual segmentation, and improves the effectiveness of preoperative planning. Using AI segmentation data to screen the puncture needle track and mark the dose range ensures the accuracy of the needle track design and dose distribution, and reduces the errors caused by manual design. The screened data is converted into entities to generate accurate entity data, supporting high-precision three-dimensional reconstruction. The vascular mapping entity data is generated by the vascular projection entity conversion, which provides the three-dimensional spatial distribution information of the blood vessels, which is helpful for the accuracy and functionality of the guide design. The puncture needle track, skin and vascular mapping data are three-dimensionally reconstructed to generate the initial three-dimensional planning guide, providing a detailed model for actual use. The initial guide is cut according to the preset needle track cutting distance, and the three-dimensional planning guide is printed. This step ensures that the size and design of the guide meet the actual surgical requirements and improves the practicality of the guide. The functional matching test of the three-dimensional planned printed guide can verify the adaptability and effect of the guide in the actual operation, ensuring that it meets the vascular function requirements. The guide is optimized according to the functional matching test data, which improves the accuracy and performance of the guide and ensures its effectiveness and accuracy in actual operation. Therefore, the present invention improves the automatic generation accuracy and reliability of the three-dimensional vascular mapping guide through multimodal image fusion, AI automatic segmentation and functional testing.

[0006] Preferably, step S1 comprises the following steps: Step S11: obtaining a CT image of the patient's lesion tissue and an MRI image of the patient's lesion tissue by computer tomography technology; Step S12: performing image preprocessing on the patient's lesion tissue CT image and the patient's lesion tissue MRI image to generate a standard patient's lesion tissue CT image and a standard patient's lesion tissue MRI image, wherein the image preprocessing includes image filtering, image brightness enhancement and image gray value conversion; Step S13: extracting high-density structural tissue from the CT image of the lesion tissue of the standard patient to obtain a high-density structural tissue region image of the CT image; identifying the vascular fuzzy region of the high-density structural tissue region image of the CT image to generate vascular fuzzy region data of the CT image; Step S14: performing angiography enhancement on the CT image blood vessel blurred area data to generate CT image blood vessel enhanced area data; performing blood vessel blurred area image replacement on the CT image high-density structure tissue area according to the CT image blood vessel enhanced area data to generate a standard patient lesion tissue optimized CT image; Step S15: Perform soft tissue identification on the MRI image of the patient's lesion tissue to generate an MRI image soft tissue region image; merge the standard patient lesion tissue optimized CT image and the MRI image soft tissue region image to generate a patient lesion tissue characteristic image.

[0007] The present invention can provide rich lesion information by acquiring image data of different imaging technologies (CT and MRI). CT images provide high-resolution bone and high-density tissue information, while MRI images provide detailed information of soft tissue. The combination of these two images can provide a more comprehensive understanding of the specific situation of the patient's lesions. Image preprocessing can remove noise, improve image contrast and details, make the image clearer, facilitate subsequent analysis and processing, and improve image quality through image filtering, image brightness enhancement, and image gray value transformation. The extraction of high-density structural tissue can help identify key lesion areas in CT images, such as tumors or calcified areas. By identifying vascular fuzzy areas, the accuracy of CT images can be further improved, providing a basis for subsequent vascular enhancement and image replacement. Angiography enhancement can make the blood vessels in CT images clearer and easier to identify and analyze. Image replacement based on the enhanced vascular image can generate a more accurate optimized CT image, making the CT image more accurate in displaying lesions and vascular structures. Soft tissue identification through MRI images can obtain detailed soft tissue information, which is very important for the diagnosis and evaluation of certain lesions (such as tumors). By merging the optimized CT images with the MRI images, a comprehensive image containing high-density structures and soft tissue features can be generated, providing clinicians with more comprehensive and accurate images of lesion characteristics.

[0008] Preferably, performing angiography enhancement on the CT image blood vessel blurred area data includes: Perform regional resolution detection on the CT image blood vessel blur region data to generate blood vessel blur region resolution detection data; perform image resolution improvement on the CT image blood vessel blur region data through the blood vessel blur region resolution detection data to generate blood vessel blur optimization region data; The vascular morphological features are extracted from the vascular fuzzy optimized area data to generate vascular morphological feature data; the vascular position is confirmed on the vascular fuzzy optimized area data based on the vascular morphological feature data to generate fuzzy area vascular position information data; the vascular fuzzy optimized area data is regionally enhanced by angiography based on the fuzzy area vascular position information data to generate CT image angiography enhanced area data.

[0009] The present invention can identify the resolution difference of the fuzzy area of ​​the blood vessel through regional resolution detection, and the generated data is used to guide the subsequent image processing steps. By understanding the resolution of these areas, the image quality can be improved more specifically to ensure that the processed image is clearer. Improving the image resolution of the fuzzy area of ​​the blood vessel can improve the detail display of the image and make the vascular structure clearer. After the resolution is improved, the image quality is significantly improved, which is helpful for the subsequent extraction and position confirmation of the vascular morphology features. Extracting the vascular morphology features can identify and record the morphological information of the blood vessel, such as the shape, size, and branches of the blood vessel. These feature data provide the necessary information support for accurately positioning and enhancing the blood vessel, making the vascular analysis more accurate. According to the vascular morphology feature data, the position of the blood vessel in the fuzzy area can be accurately located. In this way, inaccurate positioning caused by blur can be avoided and the accuracy of image processing can be improved. According to the confirmed vascular position information, the data of the fuzzy optimization area is enhanced by angiography, which can significantly improve the contrast and visibility of the blood vessel, making the blood vessel clearer and easier to identify in the CT image. This step improves the diagnostic value of CT images, allowing clinicians to more accurately evaluate the vascular state and related lesions.

[0010] Preferably, step S2 comprises the following steps: Step S21: Perform preoperative planning based on the patient's lesion tissue characteristic image to obtain preoperative planning data; Step S22: performing text semantic analysis on the preoperative planning data to generate preoperative planning text semantic data; performing keyword search on the preoperative planning data according to the preoperative planning text semantic data to generate preoperative planning keywords; Step S23: performing preoperative planning image simulation on the preoperative planning data according to the preoperative planning keywords to generate a preoperative planning image; performing image matching results on the preoperative planning image and the patient's lesion tissue characteristic image to determine if the image matching result between the preoperative planning image and the patient's lesion tissue characteristic image is true, performing AI tissue segmentation on the preoperative planning image based on the patient's lesion tissue characteristic image to generate AI tissue segmentation data; Step S24: When the image matching result between the preoperative planning image and the patient's lesion tissue feature image is false, the preoperative planning image is replanned.

[0011] The present invention can provide doctors with detailed preoperative preparation information by performing preoperative planning based on the patient's lesion tissue characteristic images. This planning based on specific image data can help doctors formulate the best surgical plan. Text semantic parsing of preoperative planning data can extract key information and generate semantic data for easy understanding and processing. Keyword retrieval based on these semantic data can quickly locate key preoperative planning content, improve data processing efficiency and accuracy, and ensure that important information is not missed. By performing image simulation of preoperative planning data through keywords, expected surgical images can be generated, providing intuitive visual reference for preoperative simulation. The preoperative planning image is matched and judged with the patient's lesion tissue characteristic image to ensure a high degree of consistency between the simulated image and the actual image. When the matching result is true, AI tissue segmentation can be further performed to accurately identify the surgical area and improve the accuracy and effect of surgical preparation. When the matching result of the preoperative planning image and the patient's lesion tissue characteristic image is false, re-planning processing can ensure the accuracy and applicability of the preoperative plan. Through repeated corrections and optimizations, a preoperative plan that best suits the actual situation can eventually be generated, avoiding surgical risks caused by inaccurate initial planning.

[0012] Preferably, when the image matching result between the preoperative planning image and the patient's lesion tissue characteristic image is true, performing AI tissue segmentation on the preoperative planning image based on the patient's lesion tissue characteristic image includes: When the image matching result between the preoperative planning image and the patient's lesion tissue characteristic image is true, the patient's lesion tissue characteristic is subjected to tissue depth analysis to generate tissue depth analysis data; the patient's lesion tissue characteristic image is subjected to tissue depth level division through the tissue depth analysis data to generate lesion tissue depth level division data; Based on the lesion tissue depth level division data, the patient's lesion tissue characteristic images are separated by tissue level to generate lesion tissue level separation data; machine learning methods are used to perform tissue segmentation and annotation on preoperative planning images through lesion tissue level separation data to generate AI tissue segmentation data.

[0013] The present invention generates tissue depth analysis data by performing a deep analysis on the patient's lesion tissue characteristics, which helps to accurately understand the internal structure and characteristics of the lesion tissue. This step provides detailed tissue information and lays the foundation for subsequent segmentation and annotation. The patient's lesion tissue feature image is divided into deep levels by the tissue depth analysis data, and the generated lesion tissue deep level division data can display the tissue structure in layers, helping doctors to understand the lesion tissues at different levels more intuitively. Tissue level separation based on the deep level division data can effectively separate the tissues at different levels, and the generated lesion tissue level separation data makes the tissue structures at different levels clearer. This helps doctors better understand the complex structure of the lesion before surgery and formulate a more accurate surgical plan. Using machine learning methods to perform tissue segmentation and annotation on preoperative planning images through lesion tissue level separation data, different tissue types can be automatically and accurately identified and annotated. The generated AI tissue segmentation data not only improves the efficiency of segmentation, but also enhances the accuracy and consistency of the segmentation results, and reduces the errors of manual annotation. The application of machine learning methods can continuously optimize the segmentation algorithm. With the increase of data and the improvement of the algorithm, the segmentation effect will gradually improve, providing more and more accurate preoperative planning support. Through deep analysis, hierarchical division and separation, and machine learning methods for tissue segmentation, it is possible to systematically and comprehensively process and analyze the characteristic images of lesion tissues. The final AI tissue segmentation data provides accurate segmentation information for preoperative planning.

[0014] Preferably, using a machine learning method to perform tissue segmentation and annotation on preoperative planning images through lesion tissue level separation data includes: The lesion tissue level separation data and the preset medical image resource library are accurately annotated to generate tissue annotation data; the tissue annotation data is enhanced to generate tissue annotation enhanced data; Divide the tissue annotation enhanced data into data sets to generate model training sets and model test sets; train the model training set through the U-Net deep learning model architecture to generate a tissue segmentation training model; use the model test set to evaluate and optimize the tissue segmentation training model to generate a tissue segmentation intelligent model; Import the preoperative planning images into the tissue segmentation intelligent model for intelligent tissue annotation to generate preoperative planning annotation images, and perform tissue segmentation on the preoperative planning annotation images to generate AI tissue segmentation data.

[0015] The present invention can utilize the existing rich annotation data to improve the accuracy and consistency of annotation by organizing the lesion tissue hierarchical separation data and the preset medical image resource library for accurate organization annotation. The generated tissue annotation data provides high-quality training samples for subsequent model training. Data enhancement of tissue annotation data can expand the diversity and scale of the data set, and the generated tissue annotation enhancement data can improve the generalization ability of the model, so that the model performs better on different types of image data and reduces overfitting. The tissue annotation enhancement data is divided into a model training set and a model test set to ensure the scientificity and rationality of model training and evaluation. By dividing the data set, different data can be used in the training and testing stages to avoid overfitting of the model on the training data and improve the practical application effect of the model. The model training set is trained using the U-Net deep learning model architecture to generate a tissue segmentation training model. The U-Net model performs well in medical image segmentation tasks, and its encoder-decoder structure can effectively capture the details and features in the image and improve the accuracy of segmentation. The tissue segmentation training model is evaluated and optimized using the model test set, the actual performance of the model can be detected, and the segmentation effect of the model can be improved through the optimization process. The generated intelligent tissue segmentation model has higher accuracy and robustness, and can provide more reliable segmentation results in practical applications. The preoperative planning images are imported into the intelligent tissue segmentation model for intelligent labeling. The generated preoperative planning labeling images can automatically and accurately label various tissues in the surgical area, improving the accuracy and efficiency of preoperative planning. The preoperative planning labeling images are subjected to tissue segmentation, and the generated AI tissue segmentation data provides detailed tissue distribution information for surgery.

[0016] Preferably, step S3 comprises the following steps: Step S31: screening the puncture needle track and the marked dose range of the preoperative planning image by using AI tissue segmentation data to obtain puncture needle track data and marked dose range data; Step S32: using a 3D compiler to perform entity conversion on the puncture needle track data and the marked dose range data to generate puncture needle track entity data and marked dose range entity data; Step S33: selecting local skin tissue from the puncture needle track entity data by marking the dose range entity data to obtain local skin tissue data; Step S34: Perform local skin entity conversion on the local skin tissue data to generate local skin entity conversion data; perform vascular projection entity conversion on the local skin entity conversion data to generate vascular mapping entity data; perform three-dimensional reconstruction on the puncture needle track entity data, local skin entity conversion data and vascular mapping entity data to generate an initial three-dimensional planning guide.

[0017] The present invention uses AI tissue segmentation data to screen the puncture needle track and the marked dose range of the preoperative planning image, and can automatically and efficiently determine the puncture path and dose distribution. This precise screening can improve the accuracy and safety of treatment and reduce damage to surrounding healthy tissues. The 3D compiler is used to perform entity conversion on the puncture needle track data and the marked dose range data, and intuitive three-dimensional entity data can be generated. These entity data provide doctors with detailed three-dimensional structural references, enabling them to better plan and perform operations. By selecting local skin tissue from the puncture needle track entity data with the marked dose range entity data, the skin area that needs to be treated during the operation can be accurately determined. This step ensures the accuracy of surgical planning, improves the accuracy of intraoperative operations, and reduces unnecessary damage. The local skin tissue data is converted to a local skin entity, and the generated local skin entity conversion data provides a detailed three-dimensional model of the surgical area, providing an important reference for doctors. The local skin entity conversion data is converted to a vascular projection entity, and the generated vascular mapping entity data can accurately display the vascular distribution, helping doctors avoid accidental injury to important blood vessels during surgery and reduce intraoperative bleeding and complications. The puncture needle track entity data, local skin entity conversion data and blood vessel mapping entity data are 3D reconstructed to generate an initial 3D planning guide that provides a detailed 3D navigation reference for the surgery. This guide can help doctors accurately guide the puncture path and dose distribution during surgery.

[0018] Preferably, step S34 includes the following steps: Step S341: discretizing the skin tissue surface of the local skin tissue data to generate discrete points on the skin tissue surface; performing point cloud conversion on the discrete points on the skin tissue surface to generate point cloud data on the skin tissue surface; Step S342: reconstructing the skin tissue surface point cloud data to generate skin tissue surface data; meshing the skin tissue surface data to generate skin tissue surface meshed data; performing normal extension and offset on the skin tissue surface meshed data to generate skin tissue surface offset data; Step S343: combining the skin tissue surface data and the skin tissue surface offset data to generate local skin entity transformation data; performing normal direction blood vessel projection on the local skin entity transformation data to generate blood vessel projection entity data; Step S344: Perform Boolean operation on the blood vessel transillumination entity data to generate blood vessel mapping entity data; perform three-dimensional reconstruction on the puncture needle track entity data, local skin entity conversion data and blood vessel mapping entity data to generate an initial three-dimensional planning guide.

[0019] The present invention can convert a continuous skin surface into a series of discrete points by discretizing local skin tissue data, and the generated discrete points of the skin tissue surface are convenient for further processing and analysis. The discrete points on the skin tissue surface are converted into point clouds, and the generated point cloud data of the skin tissue surface provides detailed three-dimensional point information, which is helpful for subsequent surface reconstruction and other three-dimensional processing steps. The point cloud data on the skin tissue surface is reconstructed, and the point cloud data can be converted into a continuous surface model. The generated skin tissue surface data more intuitively displays the skin surface morphology. The skin tissue surface data is meshed, and the generated skin tissue surface meshed data can make the surface more regular and easy to calculate, which is helpful for subsequent processing and treatment. The surface meshed data is normal extended and offset, and the generated skin tissue surface offset data can create an offset surface based on the original surface, providing more structural information for further processing. The skin tissue surface data and the skin tissue surface offset data are combined, and the generated local skin entity conversion data can provide a complete skin structure model, which enhances the integrity and accuracy of the model. The vascular projection entity data generated by performing normal vascular projection on the local skin entity transformation data can accurately display the vascular distribution under the skin, providing an important reference for subsequent surgical planning and execution. The vascular mapping entity data generated by performing Boolean operations on the vascular projection entity data can integrate vascular information and skin tissue structure, providing a more complete and detailed reference for three-dimensional reconstruction. The puncture needle track entity data, local skin entity transformation data, and vascular mapping entity data are reconstructed in three dimensions, and the generated initial three-dimensional planning guide can provide comprehensive three-dimensional navigation and guidance for surgery.

[0020] Preferably, step S4 comprises the following steps: Step S41: cutting the initial three-dimensional planned guide plate based on the preset needle track cutting distance to generate a three-dimensional planned cutting guide plate; Step S42: 3D printing the three-dimensional planned cutting guide to generate a three-dimensional planned printing guide; Step S43: performing a vascular function matching test on the three-dimensional plan printing guide to generate guide vascular function matching test data; optimizing the three-dimensional plan printing guide based on the guide vascular function matching test data to generate a three-dimensional plan optimized guide.

[0021] The present invention cuts the initial three-dimensional plan guide plate based on a preset needle channel cutting distance to ensure that the size and shape of the guide plate are suitable for actual operation requirements. In this way, unnecessary parts of the guide plate in use can be reduced, the flexibility and accuracy of its use can be improved, and it is convenient for doctors to operate during surgery. The cut three-dimensional plan guide plate is materialized by 3D printing technology to generate a three-dimensional plan printed guide plate. This materialized guide plate provides an intuitive and operable tool, enabling doctors to better plan and operate surgery before and during surgery. The three-dimensional plan printed guide plate is tested for vascular function matching to ensure that the guide plate matches the patient's vascular structure in actual use. This test step can verify the accuracy and practicality of the guide plate design and avoid operational errors and risks caused by guide plate mismatch during surgery. The three-dimensional plan printed guide plate is optimized by vascular function matching test data to generate a three-dimensional plan optimized guide plate. The optimized guide plate further improves its adaptability and functionality in actual use. Through guide plate cutting, 3D printing, vascular function matching test and guide plate optimization, the entire process ensures the accuracy and practicality of the guide plate design and actual use. The generated 3D plan optimization guide provides accurate navigation and operation guidance for doctors before and during surgery. The optimized guide can effectively reduce intraoperative risks, ensure the precise implementation of the puncture path and marked dose range, and avoid damage to surrounding healthy tissues.

[0022] Preferably, step S43 includes the following steps: Step S431: performing vascular fitting virtual matching on the three-dimensional planned printed guide plate to generate guide plate vascular matching data; Step S432: Performing a functional evaluation on the three-dimensional planned printed guide plate through the guide plate blood vessel matching data to generate guide plate functional evaluation data, wherein the functional evaluation includes blood vessel patency, hemodynamic changes, and the degree of fit between the guide plate and the blood vessel wall; Step S433: Score the site adaptability of the guide functional evaluation data to generate guide vascular function matching test data; identify guide problems on the guide vascular function matching test data to generate guide design problem data; optimize the three-dimensional planned printing guide using the guide design problem data to generate a three-dimensional planned optimized guide.

[0023] The present invention performs a vascular fit test on a three-dimensional plan printed guide plate through a virtual matching technology, and the generated guide plate vascular matching data can accurately evaluate the compatibility of the guide plate with the patient's vascular structure. This step ensures the fit of the guide plate in actual surgery and improves the accuracy and effectiveness of the surgery. The vascular patency of the three-dimensional plan printed guide plate is evaluated through the guide plate vascular matching data, which can ensure that the guide plate will not hinder the normal function of the blood vessel during use and ensure smooth blood flow during surgery. The influence of the guide plate on hemodynamics is evaluated to ensure that the guide plate will not cause significant changes in hemodynamics in actual application. The guide plate functional evaluation data is scored for site adaptability to generate guide plate vascular function matching test data. This scoring system can quantify the adaptability of the guide plate and provide a specific reference for subsequent optimization. By analyzing the guide plate vascular function matching test data, problems in the guide plate design are identified and guide plate design problem data are generated. This step can timely discover and correct deficiencies in the design and improve the design quality of the guide plate. According to the identified design problems, the three-dimensional plan printed guide plate is optimized to generate a three-dimensional plan optimized guide plate. The optimized guide can better adapt to surgical needs and improve its functionality and adaptability.

[0024] The beneficial effect of the present invention is that by acquiring CT and MRI images of the patient's lesion tissue, fuzzy vascular area angiography optimization and soft tissue identification are performed, and the generated patient lesion tissue characteristic images can fully display the structural characteristics and soft tissue conditions of the lesion. This comprehensive image data merger provides a detailed and accurate reference for subsequent preoperative planning, ensuring the accuracy of the surgical plan. The preoperative planning images generated based on the patient's lesion tissue characteristic images can help doctors plan surgical paths and strategies in advance. This step can reduce intraoperative uncertainty, and the image matching results are judged. When the matching result is true, AI is used for tissue segmentation. The generated AI tissue segmentation data can accurately mark various types of tissues in the surgical area and provide detailed tissue information for the operation; when the matching result is false, re-planning is performed to ensure the accuracy and applicability of the surgical plan. The preoperative planning images are screened through AI tissue segmentation data, and the obtained puncture needle track data and marked dose range data can accurately determine the surgical path and dose range. The puncture needle track data and marked dose range data are physically converted, and the generated three-dimensional planning guide provides an intuitive three-dimensional structural model. The initial three-dimensional planning guide is trimmed based on the preset needle track cutting distance and 3D printed. The generated three-dimensional planning printed guide has precise size and shape, which is suitable for actual surgical needs and convenient for doctors to use. The vascular function matching test of the three-dimensional planning printed guide can verify the adaptability of the guide to the patient's vascular structure and ensure the effectiveness and safety of the guide in actual use. The three-dimensional planning printed guide is optimized through test data, and the generated three-dimensional planning optimized guide can further improve the functionality and applicability of the guide. Therefore, the present invention improves the automatic generation accuracy and reliability of the three-dimensional vascular mapping guide through multimodal image fusion, AI automatic segmentation and functional testing. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A schematic diagram of the steps of a method for automatically generating a human blood vessel mapping guide based on three-dimensional modeling; Figure 2 for Figure 1 Detailed implementation steps of step S2 in the flowchart; Figure 3 for Figure 1 Detailed implementation steps of step S3 in FIG. Figure 4 for Figure 1 Detailed implementation steps of step S4 in FIG. The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0026] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. 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 technicians in this field without creative work are within the scope of protection of the present invention.

[0027] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0028] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0029] To achieve this, please refer to Figures 1 to 4 , a method for automatically generating a human blood vessel mapping guide based on three-dimensional modeling, the method comprising the following steps: Step S1: obtaining a CT image of the patient's lesion tissue and an MRI image of the patient's lesion tissue; performing fuzzy vascular area angiography optimization on the CT image of the patient's lesion tissue to generate a standard patient's lesion tissue optimized CT image; performing soft tissue identification on the MRI image of the patient's lesion tissue to generate an MRI image of the soft tissue area; merging the standard patient's lesion tissue optimized CT image and the patient's lesion tissue MRI image to generate a patient's lesion tissue characteristic image; Step S2: performing preoperative planning image simulation based on the patient's lesion tissue characteristic image to generate a preoperative planning image; performing image matching results on the preoperative planning image and the patient's lesion tissue characteristic image, and when the image matching result between the preoperative planning image and the patient's lesion tissue characteristic image is true, performing AI tissue segmentation on the preoperative planning image based on the patient's lesion tissue characteristic image to generate AI tissue segmentation data; when the image matching result between the preoperative planning image and the patient's lesion tissue characteristic image is false, re-planning the preoperative planning image; Step S3: Screen the puncture needle track and the marked dose range of the preoperative planning image through AI tissue segmentation data to obtain the puncture needle track data and the marked dose range data; perform entity conversion on the puncture needle track data and the marked dose range data to generate the puncture needle track entity data and the marked dose range entity data; perform vascular projection entity conversion on the puncture needle track entity data through the marked dose range entity data to generate vascular mapping entity data; perform three-dimensional reconstruction on the puncture needle track entity data, the local skin entity conversion data and the vascular mapping entity data to generate an initial three-dimensional planning guide; Step S4: cutting and printing the initial three-dimensional planned guide based on the preset needle channel cutting distance to generate a three-dimensional planned printed guide; performing a vascular function matching test on the three-dimensional planned printed guide to generate guide vascular function matching test data; optimizing the three-dimensional planned printed guide using the guide vascular function matching test data to generate a three-dimensional planned optimized guide.

[0030] The present invention can fuse blood vessel and soft tissue information and generate more comprehensive lesion tissue feature images by acquiring and optimizing CT and MRI images of patients. This multimodal image fusion improves the details and accuracy of the image and provides a high-quality data basis for subsequent processing. Fuzzy optimization of the blood vessel area in the CT image can highlight the blood vessel structure, reduce noise interference, and improve the clarity and accuracy of blood vessel identification. Identifying soft tissue in MRI images can clearly mark the soft tissue area and provide necessary information for the accurate reconstruction of blood vessels and tissue structures. By generating preoperative planning images based on feature images, it is helpful for doctors to carry out detailed surgical planning, judge the matching results of preoperative planning images and feature images, and ensure that the preoperative plan is consistent with the actual image data. When the matching result is true, AI is used to perform tissue segmentation to reduce human errors; when the matching result is false, re-planning is performed to ensure the reliability of the preoperative plan. Automatic tissue segmentation improves the speed and accuracy of segmentation, reduces the workload and errors of manual segmentation, and improves the effectiveness of preoperative planning. Using AI segmentation data to screen the puncture needle track and mark the dose range ensures the accuracy of the needle track design and dose distribution, and reduces the errors caused by manual design. The screened data is converted into entities to generate accurate entity data, supporting high-precision three-dimensional reconstruction. The vascular mapping entity data is generated by the vascular projection entity conversion, which provides the three-dimensional spatial distribution information of the blood vessels, which is helpful for the accuracy and functionality of the guide design. The puncture needle track, skin and vascular mapping data are three-dimensionally reconstructed to generate the initial three-dimensional planning guide, providing a detailed model for actual use. The initial guide is cut according to the preset needle track cutting distance, and the three-dimensional planning guide is printed. This step ensures that the size and design of the guide meet the actual surgical requirements and improves the practicality of the guide. The functional matching test of the three-dimensional planned printed guide can verify the adaptability and effect of the guide in the actual operation, ensuring that it meets the vascular function requirements. The guide is optimized according to the functional matching test data, which improves the accuracy and performance of the guide and ensures its effectiveness and accuracy in actual operation. Therefore, the present invention improves the automatic generation accuracy and reliability of the three-dimensional vascular mapping guide through multimodal image fusion, AI automatic segmentation and functional testing.

[0031] In the embodiment of the present invention, reference Figure 1 The above is a schematic flow chart of the steps of a method for automatically generating a human blood vessel mapping guide based on three-dimensional modeling of the present invention. In this example, the method for automatically generating a human blood vessel mapping guide based on three-dimensional modeling includes the following steps: Step S1: obtaining a CT image of the patient's lesion tissue and an MRI image of the patient's lesion tissue; performing fuzzy vascular area angiography optimization on the CT image of the patient's lesion tissue to generate a standard patient's lesion tissue optimized CT image; performing soft tissue identification on the MRI image of the patient's lesion tissue to generate an MRI image of the soft tissue area; merging the standard patient's lesion tissue optimized CT image and the patient's lesion tissue MRI image to generate a patient's lesion tissue characteristic image; In an embodiment of the present invention, a CT image of the patient's lesion tissue is obtained from a CT scanning system of a hospital or medical institution. Ensure that the resolution and slice thickness of the image meet the diagnostic requirements. Obtain an MRI image of the patient's lesion tissue from an MRI scanning system. Select appropriate sequences and contrast agents (such as T1 and T2 weighted imaging) to obtain high-quality soft tissue images. Perform noise removal, image smoothing and enhancement on the CT image. Use a filter (such as Gaussian filtering) to remove noise. Apply angiography techniques (such as maximum intensity projection, vascular enhancement algorithm) to optimize the fuzzy vascular area. Enhance the visualization of vascular structure so that the morphology and distribution of the blood vessels can be seen more clearly. Compare the optimized vascular area with the standard CT image, adjust parameters such as brightness and contrast, and generate a standard patient lesion tissue optimized CT image. Perform image preprocessing on the MRI image, including denoising, standardization, and contrast adjustment. Use an image segmentation algorithm (such as threshold segmentation, region growing, and deep learning method) to identify and mark the soft tissue area. The algorithm may include a model based on a convolutional neural network (CNN) to perform high-precision soft tissue segmentation, output the segmentation results as MRI image soft tissue area images, and display the soft tissue part of the lesion tissue. Select an appropriate registration algorithm (such as feature point-based registration, rigid registration, and non-rigid registration) to align the CT image with the MRI image. During the alignment process, the spatial resolution, calibration information, and patient position of the image should be considered, and the CT and MRI images should be registered so that the data of the two images can correspond in the same space. The registered standard CT image and the soft tissue area image are fused, and the data of the two images are merged using image fusion technology (such as weighted averaging, image overlay, and color coding), and the merged patient lesion tissue feature image is output, which comprehensively displays the vascular information of the CT image and the soft tissue information of the MRI image.

[0032] Step S2: performing preoperative planning image simulation based on the patient's lesion tissue characteristic image to generate a preoperative planning image; performing image matching results on the preoperative planning image and the patient's lesion tissue characteristic image, and when the image matching result between the preoperative planning image and the patient's lesion tissue characteristic image is true, performing AI tissue segmentation on the preoperative planning image based on the patient's lesion tissue characteristic image to generate AI tissue segmentation data; when the image matching result between the preoperative planning image and the patient's lesion tissue characteristic image is false, re-planning the preoperative planning image; In an embodiment of the present invention, a preoperative planning simulation is prepared by importing a patient's lesion tissue feature image into a preoperative planning software or platform. A preoperative planning model is established based on information such as the location, size, and morphology of the lesion. Use surgical planning tools (such as simulated surgery, virtual cutting, etc.) to design preoperative surgical paths and operation steps. Perform preoperative planning simulation to generate preoperative planning images. The image should show the expected surgical steps, instrument paths, target areas, etc. in the plan. Optimize the image according to the simulation results to ensure that it accurately reflects the surgical plan. Use image registration technology to align the preoperative planning image with the patient's lesion tissue feature image. Select a suitable registration algorithm (such as mutual information, phase correlation, deep learning registration, etc.) to perform spatial registration on the image. Evaluate the matching degree between the preoperative planning image and the patient's lesion tissue feature image. Calculate the overlap and consistency of the image based on the registration results. Set the judgment criteria for image matching (such as overlap threshold, matching threshold, etc.) to determine whether the preoperative planning image matches the patient's lesion tissue feature image. Input the preoperative planning image and the patient's lesion tissue feature image into the AI ​​tissue segmentation system. Use pre-trained AI models (such as U-Net, DeepLab, etc.) to perform tissue segmentation on preoperative planning images. The model performs segmentation and annotation based on the patient's lesion tissue feature images. Output AI tissue segmentation data, including annotated lesion areas, vascular areas, soft tissue areas, etc. Analyze the reasons for image matching failure, including image quality issues, registration errors, and inconsistency between the preoperative plan and the actual situation. Redesign the preoperative plan based on the actual situation. The surgical path, instrument position, target area, etc. can be adjusted to improve consistency with the actual lesion tissue feature images. Simulate the re-planned preoperative plan, generate a new preoperative planning image, and repeat the matching and segmentation steps.

[0033] Step S3: Screen the puncture needle track and the marked dose range of the preoperative planning image through AI tissue segmentation data to obtain the puncture needle track data and the marked dose range data; perform entity conversion on the puncture needle track data and the marked dose range data to generate the puncture needle track entity data and the marked dose range entity data; perform vascular projection entity conversion on the puncture needle track entity data through the marked dose range entity data to generate vascular mapping entity data; perform three-dimensional reconstruction on the puncture needle track entity data, the local skin entity conversion data and the vascular mapping entity data to generate an initial three-dimensional planning guide; In an embodiment of the present invention, AI tissue segmentation data (including segmented lesion areas, vascular areas, soft tissue areas, etc.) is imported into a processing system. Screening parameters are set, such as path requirements for the puncture needle track, standards for the marked dose range, etc. The path of the puncture needle track is planned based on the AI ​​tissue segmentation data. The optimal puncture path is selected taking into account the location, size and structure of the lesion. Data related to the puncture needle track is extracted from the tissue segmentation data to generate puncture needle track data. Based on the segmentation data and the surgical plan, the marked dose range is determined. This includes determining the dose distribution area of ​​radiation or drugs. Data related to the marked dose range is extracted from the segmentation data to generate marked dose range data. Entity conversion is performed using a three-dimensional modeling tool (such as CAD software, 3D modeling platform). Ensure that the puncture needle track data and the marked dose range data are input into the modeling tool in a format suitable for conversion (such as point cloud, mesh data). Convert the puncture needle track data from a 2D / 3D image format to a 3D solid model. Generate puncture needle track entity data, which includes the specific path and structure of the needle track. Convert the marked dose range data from an image format to a 3D solid model. Generate marked dose range entity data to display the distribution of dose range in space. Import marked dose range entity data and puncture needle track entity data into the vascular projection processing tool. Use the vascular projection algorithm to project the marked dose range entity data onto the puncture needle track entity data. Generate vascular mapping entity data to represent the relationship between the puncture needle track and the vascular structure. Ensure that the local skin entity transformation data is ready, including the 3D model of the skin and related information. Import the puncture needle track entity data, local skin entity transformation data, and vascular mapping entity data into the 3D modeling software. Fuse these data for 3D modeling and reconstruction. Ensure that each data type (needle track, skin, vascular mapping) is correctly aligned in space. After completing the 3D reconstruction, generate the initial 3D planning guide. The guide should comprehensively display the puncture needle track, marked dose range, local skin structure, and vascular mapping information.

[0034] Step S4: cutting and printing the initial three-dimensional planned guide based on the preset needle channel cutting distance to generate a three-dimensional planned printed guide; performing a vascular function matching test on the three-dimensional planned printed guide to generate guide vascular function matching test data; optimizing the three-dimensional planned printed guide using the guide vascular function matching test data to generate a three-dimensional planned optimized guide.

[0035] In an embodiment of the present invention, a preset value of the needle channel cutting distance is set. This is based on the specific parameters of the surgical requirements and the guide plate design. The initial three-dimensional planned guide plate is imported into the cutting processing software. According to the preset needle channel cutting distance, the guide plate cutting operation is performed. CAD software or a three-dimensional modeling tool can be used to cut a specific part of the guide plate to ensure that the guide plate matches the needle channel, and the three-dimensional planned cutting guide plate after cutting is output. The model file of the three-dimensional planned cutting guide plate is converted into the format required by the 3D printer (such as STL format), and the printer parameters are set, including material type, printing resolution and filling density. The converted file is imported into the 3D printer for printing. During the printing process, the progress is monitored to ensure the printing quality. After the printing is completed, the printed guide plate is taken out to ensure that it meets the design requirements. The three-dimensional planned printed guide plate is virtually matched with the blood vessel model. Computer-aided design (CAD) tools or simulation software are used to evaluate the degree of adaptation of the guide plate to the actual blood vessel structure. The vascular patency of the guide plate in actual use is evaluated, and whether the puncture path can pass through the blood vessel smoothly is checked. Simulate the hemodynamic changes after puncture and analyze the influence of the guide plate on the blood flow of the blood vessel. Check the fit between the guide and the vessel wall to ensure its stability and accuracy in actual surgery. Record all data from the vascular function matching test, including patency, hemodynamic changes, and fit. Analyze the guide vascular function matching test data to identify problems in the guide design, such as insufficient adaptability and poor fit. Identify problems in the guide design through test data and determine areas that need to be optimized. Adjust the guide design based on the test results, including adjusting the puncture path, improving the fit design, optimizing vascular patency, etc. Reprint the optimized design to generate a 3D planned optimized guide. Test the optimized guide again to ensure that it meets the design requirements and surgical needs, generate and confirm the final 3D planned optimized guide, and prepare for actual surgery.

[0036] Preferably, step S1 comprises the following steps: Step S11: obtaining a CT image of the patient's lesion tissue and an MRI image of the patient's lesion tissue by computer tomography technology; Step S12: performing image preprocessing on the patient's lesion tissue CT image and the patient's lesion tissue MRI image to generate a standard patient's lesion tissue CT image and a standard patient's lesion tissue MRI image, wherein the image preprocessing includes image filtering, image brightness enhancement and image gray value conversion; Step S13: extracting high-density structural tissue from the CT image of the lesion tissue of the standard patient to obtain a high-density structural tissue region image of the CT image; identifying the vascular fuzzy region of the high-density structural tissue region image of the CT image to generate vascular fuzzy region data of the CT image; Step S14: performing angiography enhancement on the CT image blood vessel blurred area data to generate CT image blood vessel enhanced area data; performing blood vessel blurred area image replacement on the CT image high-density structure tissue area according to the CT image blood vessel enhanced area data to generate a standard patient lesion tissue optimized CT image; Step S15: Perform soft tissue identification on the MRI image of the patient's lesion tissue to generate an MRI image soft tissue region image; merge the standard patient lesion tissue optimized CT image and the MRI image soft tissue region image to generate a patient lesion tissue characteristic image.

[0037] In an embodiment of the present invention, a CT image of a patient's lesion tissue is obtained by computer tomography (CT). An MRI image of a patient's lesion tissue is obtained by magnetic resonance imaging (MRI). The CT and MRI images are subjected to noise filtering to remove noise from the images. The images are subjected to brightness enhancement processing to improve the contrast of the images so that the lesion tissue is more clearly visible. The grayscale value range of the image is adjusted to make the details of the image more obvious. The CT image of the lesion tissue of a standard patient is analyzed to extract image data of a high-density structural tissue area. In the image of the high-density structural tissue area, a vascular fuzzy area is identified and related data is generated. The vascular fuzzy area data is subjected to contrast enhancement processing to improve the visibility of the blood vessels. According to the enhanced vascular data, the vascular fuzzy area in the high-density structural tissue area is replaced with an image to generate an optimized CT image. The MRI image of the lesion tissue of a standard patient is analyzed to generate a soft tissue area image, and the optimized CT image and the soft tissue area image in the MRI image are merged to generate a comprehensive image containing the characteristics of the lesion tissue.

[0038] Preferably, performing angiography enhancement on the CT image blood vessel blurred area data includes: Perform regional resolution detection on the CT image blood vessel blur region data to generate blood vessel blur region resolution detection data; perform image resolution improvement on the CT image blood vessel blur region data through the blood vessel blur region resolution detection data to generate blood vessel blur optimization region data; The vascular morphological features are extracted from the vascular fuzzy optimized area data to generate vascular morphological feature data; the vascular position is confirmed on the vascular fuzzy optimized area data based on the vascular morphological feature data to generate fuzzy area vascular position information data; the vascular fuzzy optimized area data is regionally enhanced by angiography based on the fuzzy area vascular position information data to generate CT image angiography enhanced area data.

[0039] In the embodiment of the present invention, by performing resolution detection on the vascular blurred area data in the CT image, the resolution detection data of the vascular blurred area is generated. This step is intended to identify and evaluate the resolution of the vascular blurred area to determine whether further optimization is required. According to the resolution detection data, the vascular blurred area data of the CT image is subjected to image resolution enhancement processing to generate vascular blurred optimized area data. This step uses image processing technology to improve the clarity of the blurred area and make the vascular structure easier to identify. The vascular morphological features of the vascular blurred optimized area data are extracted to generate vascular morphological feature data. This step uses an image analysis algorithm to extract the morphological features of the blood vessels, such as shape, size and position. According to the extracted vascular morphological feature data, the vascular position of the vascular blurred optimized area data is confirmed to generate vascular position information data of the blurred area. The purpose of this step is to accurately locate the specific position of the blood vessel in the blurred area. According to the vascular position information data of the blurred area, the vascular blurred optimized area data is subjected to regional vascular contrast enhancement to generate vascular contrast enhancement area data of the CT image. This step makes the blood vessels more clearly visible in the image by applying contrast enhancement technology.

[0040] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes: Step S21: Perform preoperative planning based on the patient's lesion tissue characteristic image to obtain preoperative planning data; Step S22: performing text semantic analysis on the preoperative planning data to generate preoperative planning text semantic data; performing keyword search on the preoperative planning data according to the preoperative planning text semantic data to generate preoperative planning keywords; Step S23: performing preoperative planning image simulation on the preoperative planning data according to the preoperative planning keywords to generate a preoperative planning image; performing image matching results on the preoperative planning image and the patient's lesion tissue characteristic image to determine if the image matching result between the preoperative planning image and the patient's lesion tissue characteristic image is true, performing AI tissue segmentation on the preoperative planning image based on the patient's lesion tissue characteristic image to generate AI tissue segmentation data; Step S24: When the image matching result between the preoperative planning image and the patient's lesion tissue feature image is false, the preoperative planning image is replanned.

[0041] In an embodiment of the present invention, characteristic images of the patient's lesion tissue are obtained by using CT, MRI or other imaging technologies. The image data is imported into professional medical image processing software. The lesion tissue is analyzed using image processing software to mark the lesion area and the surrounding important anatomical structures. Combined with the patient's medical history, clinical examination results and other information, the nature, location, size, etc. of the lesion are comprehensively evaluated. Based on the image analysis results, a detailed preoperative plan is formulated, including the location of the surgical incision, the surgical path, key steps and possible risks. A 3D modeling tool is used to create a three-dimensional model of the lesion tissue and surrounding structures to facilitate surgical planning and simulation. The above analysis and planning results are used to generate digital preoperative planning data, including detailed information such as image data, 3D models, and surgical steps. Natural language processing (NLP) technology is used to perform semantic analysis on the text information in the preoperative planning data to extract key content and contextual relationships. NLP technology can include vocabulary analysis, syntactic analysis, entity recognition, etc. to ensure that the extracted text information is accurate and has medical significance. The parsed text information is organized into structured text semantic data for subsequent processing. Based on text semantic data, use keyword extraction algorithms (such as TF-IDF, keyword extraction models, etc.) to extract important preoperative planning keywords from preoperative planning data. Ensure that the extracted keywords can accurately represent the key content of the preoperative plan and the key points of the operation. Use image processing and simulation tools to generate preoperative planning images based on the extracted preoperative planning keywords. Image simulation should reproduce the situation encountered during the operation as realistically as possible to help doctors pre-evaluate the difficulties and key steps of the operation. Use image matching algorithms (such as feature point-based matching algorithms, deep learning-based matching algorithms, etc.) to match the preoperative planning images with the patient's lesion tissue feature images. Determine whether the image matching result is true to ensure that the preoperative planning images are highly consistent with the actual lesion images. If the image matching result is true, use AI algorithms (such as convolutional neural networks, U-Net, etc.) to perform tissue segmentation on the preoperative planning images and generate AI tissue segmentation data. Ensure that the segmentation results can accurately identify the lesion tissue and surrounding important structures, and provide reliable surgical navigation information. When the image matching result is false, analyze the reasons for the matching failure, including image quality problems, changes in lesion characteristics, etc. Adjust and optimize the preoperative planning images, and re-plan the surgical incision, path, and key steps. Regenerate the adjusted preoperative planning images, and perform image matching and identification again. Through multiple iterations, gradually optimize the preoperative plan to ensure that the final preoperative planning images are highly consistent with the patient's lesion tissue feature images.

[0042] Preferably, when the image matching result between the preoperative planning image and the patient's lesion tissue characteristic image is true, performing AI tissue segmentation on the preoperative planning image based on the patient's lesion tissue characteristic image includes: When the image matching result between the preoperative planning image and the patient's lesion tissue characteristic image is true, the patient's lesion tissue characteristic is subjected to tissue depth analysis to generate tissue depth analysis data; the patient's lesion tissue characteristic image is subjected to tissue depth level division through the tissue depth analysis data to generate lesion tissue depth level division data; Based on the lesion tissue depth level division data, the patient's lesion tissue characteristic images are separated by tissue level to generate lesion tissue level separation data; machine learning methods are used to perform tissue segmentation and annotation on preoperative planning images through lesion tissue level separation data to generate AI tissue segmentation data.

[0043] In an embodiment of the present invention, the patient's lesion tissue feature image is preprocessed, including denoising, standardization, enhancement, etc., to improve the accuracy of the analysis. A deep learning model (such as a deep convolutional neural network) or other advanced image processing technology is used to perform a deep analysis of the lesion tissue. The deep feature information of the tissue is extracted, including the internal structure, hierarchy and boundaries of the tissue. The extracted deep feature information is organized into structured tissue depth analysis data to reflect the hierarchy and structural information of the tissue. Based on the tissue depth analysis data, the tissue features and differences at different depth levels are analyzed. The tissue depth features are hierarchically divided using a hierarchical algorithm (such as cluster analysis, hierarchical classification algorithm, etc.). According to the division results, lesion tissue depth hierarchical division data is generated, including depth information and feature descriptions of each level. The lesion tissue depth hierarchical division data is processed to clarify the boundaries and separations between different levels. The tissue is hierarchically separated using an image segmentation technique (such as a region growing algorithm, an image segmentation network, etc.), and the lesion tissue hierarchical separation data is output, including separation information and labels of each level of tissue. The generated lesion tissue hierarchical separation data is used to train the machine learning model. Select appropriate machine learning algorithms (such as convolutional neural networks, U-Net, deep generative adversarial networks, etc.) for training to ensure that the model can accurately identify and segment different tissue levels. Apply the trained model to preoperative planning images, perform tissue segmentation annotation on the images, generate AI tissue segmentation data, and identify the various tissue levels and structures in the preoperative planning images.

[0044] Preferably, using a machine learning method to perform tissue segmentation and annotation on preoperative planning images through lesion tissue level separation data includes: The lesion tissue level separation data and the preset medical image resource library are accurately annotated to generate tissue annotation data; the tissue annotation data is enhanced to generate tissue annotation enhanced data; Divide the tissue annotation enhanced data into data sets to generate model training sets and model test sets; train the model training set through the U-Net deep learning model architecture to generate a tissue segmentation training model; use the model test set to evaluate and optimize the tissue segmentation training model to generate a tissue segmentation intelligent model; Import the preoperative planning images into the tissue segmentation intelligent model for intelligent tissue annotation to generate preoperative planning annotation images, and perform tissue segmentation on the preoperative planning annotation images to generate AI tissue segmentation data.

[0045] In an embodiment of the present invention, by collecting lesion tissue level separation data and image data in a medical image resource library, a labeling standard is defined, including tissue categories, labeling methods, and labeling details. Medical image labeling tools (such as ITK-SNAP, Labelbox, etc.) are used to label images. The lesion tissue level separation data is combined with the image resource library to generate accurate tissue labeling data, including labels for each tissue level and structure. Data enhancement techniques such as rotation, translation, scaling, flipping, noise addition, etc. are used to enhance the diversity and robustness of the data. Data enhancement libraries (such as Albumentations, TensorFlow's tf.image, etc.) are used for processing. The tissue labeling data is enhanced to generate tissue labeling enhanced data to expand the scale and diversity of the training data set. The enhanced data set is divided into a training set and a test set, usually divided in a ratio of 80%:20% or 70%:30%. Data set partitioning tools (such as Scikit-learn's train_test_split, etc.) are used to ensure the randomness and representativeness of the data. The model training set contains labeled data for training the U-Net model. The model test set contains annotated data for evaluating and optimizing the model. Build the U-Net deep learning model architecture, including encoder, decoder, and skip connections. Set the model's hyperparameters, such as learning rate, batch size, loss function, etc. Use the model training set to train the U-Net model. Monitor the loss function and evaluation indicators during training, adjust the model parameters to improve performance, and use deep learning frameworks such as TensorFlow, PyTorch, etc. for training. Use the model test set to evaluate the trained model and calculate indicators such as accuracy, recall, and F1 score. Optimize the model based on the evaluation results, such as adjusting hyperparameters, adding regularization, using a more complex model architecture, etc. Save the optimized tissue segmentation intelligent model for subsequent prediction and annotation. Import the preoperative planning images into the tissue segmentation intelligent model for processing, and perform necessary preprocessing on the imported images, such as normalization and resizing. Use the tissue segmentation intelligent model to predict the preoperative planning images, generate tissue intelligent annotation data, generate preoperative planning annotation images, and display the tissue areas annotated by the model. Label the images according to the preoperative plan, segment the tissues in the images, and generate AI tissue segmentation data. Verify the accuracy of the generated AI tissue segmentation data, and adjust and optimize it as needed.

[0046] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes: Step S31: screening the puncture needle track and the marked dose range of the preoperative planning image by using AI tissue segmentation data to obtain puncture needle track data and marked dose range data; Step S32: using a 3D compiler to perform entity conversion on the puncture needle track data and the marked dose range data to generate puncture needle track entity data and marked dose range entity data; Step S33: selecting local skin tissue from the puncture needle track entity data by marking the dose range entity data to obtain local skin tissue data; Step S34: Perform local skin entity conversion on the local skin tissue data to generate local skin entity conversion data; perform vascular projection entity conversion on the local skin entity conversion data to generate vascular mapping entity data; perform three-dimensional reconstruction on the puncture needle track entity data, local skin entity conversion data and vascular mapping entity data to generate an initial three-dimensional planning guide.

[0047] In an embodiment of the present invention, the preoperative planning image is processed by using AI tissue segmentation data as input. According to the surgical plan and tissue segmentation results, the position and path of the puncture needle track are determined. Consider the tissue structure, lesion location, and the linearity and accessibility of the needle track. Mark and generate puncture needle track data, including information such as the starting point, end point, path, and angle of the needle track. According to the marked dose range requirements in the preoperative plan, the distribution of the marked dose in the tissue is analyzed, the boundary, size, and shape of the dose range are determined, and the corresponding data set is generated. Use 3D modeling software or a compiler (such as Blender, 3dsMax, MeshLab, etc.) to perform data conversion, and import the puncture needle track data and the marked dose range data into the 3D compiler. In the 3D compiler, a three-dimensional solid model of the needle track is generated according to the puncture needle track data. Generate a corresponding three-dimensional solid model according to the marked dose range data. Correct and optimize the generated solid model to ensure that it is consistent with the actual needs, and export the optimized solid data to a standard 3D model format (such as STL, OBJ, etc.). In the 3D compiler, the puncture needle track entity data is screened using the marked dose range entity data, the area where the puncture needle track entity data overlaps with the marked dose range is selected, the location and range of the local skin tissue is determined, and the local skin tissue data related to the marked dose range is extracted and saved. The local skin tissue data is entity transformed using a 3D modeling tool to generate a local skin entity model, and the local skin entity data is refined to ensure that it accurately represents the actual shape of the skin tissue. The vascular data (which needs to be extracted from medical images or obtained by other means) is obtained, and the local skin entity data is entity transformed into a vascular projection in the 3D compiler to generate vascular mapping entity data. The puncture needle track entity data, local skin entity transformation data, and vascular mapping entity data are integrated, and 3D reconstruction is performed using 3D modeling software to generate an initial 3D planning guide. The reconstruction process should include integrating all data into a unified 3D model and optimizing the generated 3D planning guide to ensure that it meets the needs and accuracy requirements of surgical planning.

[0048] Preferably, step S34 includes the following steps: Step S341: discretizing the skin tissue surface of the local skin tissue data to generate discrete points on the skin tissue surface; performing point cloud conversion on the discrete points on the skin tissue surface to generate point cloud data on the skin tissue surface; Step S342: reconstructing the skin tissue surface point cloud data to generate skin tissue surface data; meshing the skin tissue surface data to generate skin tissue surface meshed data; performing normal extension and offset on the skin tissue surface meshed data to generate skin tissue surface offset data; Step S343: combining the skin tissue surface data and the skin tissue surface offset data to generate local skin entity transformation data; performing normal direction blood vessel projection on the local skin entity transformation data to generate blood vessel projection entity data; Step S344: Perform Boolean operation on the blood vessel transillumination entity data to generate blood vessel mapping entity data; perform three-dimensional reconstruction on the puncture needle track entity data, local skin entity conversion data and blood vessel mapping entity data to generate an initial three-dimensional planning guide.

[0049] In an embodiment of the present invention, surface discretization is performed by using a three-dimensional model of local skin tissue data. A suitable discretization algorithm (such as uniform sampling, grid point sampling, etc.) is selected to divide the skin tissue surface into a discrete set of points, and the discrete point data of the skin tissue surface is output as the basis for point cloud generation. Point cloud processing tools (such as PCL library, MeshLab, etc.) are used to convert the discrete point data on the skin tissue surface into point cloud data, and point cloud data on the skin tissue surface is generated, including the coordinate information and additional attributes (such as color, density, etc.) of each point. Surface reconstruction algorithms (such as Marching Cubes, Poisson Surface Reconstruction, etc.) are used to convert point cloud data into surface data, and skin tissue surface data is output, including the geometric shape and structure of the skin surface. Surface data is meshed using a meshing algorithm (such as Delaunay triangulation, Voronoi diagram, etc.), and skin tissue surface meshing data is output, including vertex, edge and face information of the mesh. The normal vector of the skin tissue surface meshing data is calculated. Offset processing is performed in the normal direction to generate skin tissue surface offset data. The offset can be adjusted according to specific needs, and the skin tissue surface offset data is output to represent the extended shape of the skin tissue surface. Use 3D modeling software or compilers (such as Blender, Fusion 360, etc.) to combine the skin tissue surface data with the skin tissue surface offset data. The combined model represents the actual local skin tissue, including the offset skin surface. Use vascular data to perform vascular projection in the normal direction of the local skin entity. Vascular data can be obtained through CT or MRI images, or generated based on other medical data. The local skin entity data is processed for vascular projection to generate vascular projection entity data to represent the mapping of blood vessels on the skin surface. Use 3D modeling tools to perform Boolean operations (such as union, intersection, difference, etc.) to process the vascular projection entity data. Generate vascular mapping entity data through Boolean operations to represent the actual position and shape of the blood vessels in the skin tissue. Integrate the puncture needle track entity data, local skin entity transformation data, and vascular mapping entity data into a 3D model. Use 3D modeling software to perform 3D reconstruction to generate an initial 3D planning guide. Ensure that the model accurately reflects the puncture needle track, skin tissue and blood vessel distribution, and optimize and verify the generated three-dimensional planning guide to ensure that it meets the surgical needs and accuracy requirements.

[0050] As an example of the present invention, refer to Figure 4 As shown, in this example, step S4 includes: Step S41: cutting the initial three-dimensional planned guide plate based on the preset needle track cutting distance to generate a three-dimensional planned cutting guide plate; Step S42: 3D printing the three-dimensional planned cutting guide to generate a three-dimensional planned printing guide; Step S43: performing a vascular function matching test on the three-dimensional plan printing guide to generate guide vascular function matching test data; optimizing the three-dimensional plan printing guide based on the guide vascular function matching test data to generate a three-dimensional plan optimized guide.

[0051] In an embodiment of the present invention, a preset needle track cutting distance is determined, that is, the minimum distance from the needle track starting point or path to the guide plate boundary. The cutting distance parameters are set in the 3D modeling software to ensure that the cutting operation meets the surgical requirements and safety standards. A 3D modeling tool (such as MeshLab, Fusion 360, Blender, etc.) is used to cut the guide plate. According to the preset cutting distance, the initial 3D planned guide plate is cut to remove unnecessary parts. Ensure that the needle track path and related structures are retained, and the guide plate is adapted to the surgical needs, and the cut 3D planned cutting guide plate model is output as the basis for subsequent printing and testing. The 3D planned cutting guide plate data is converted into a format suitable for a 3D printer (such as STL, OBJ, etc.). Configure the parameters of the 3D printer, such as material type (PLA, ABS, etc.), printing resolution, filling density, support structure, etc., import the prepared 3D planned cutting guide plate data into the 3D printer, start the printing process, and monitor the working status of the printer during the printing process to ensure the printing quality and accuracy. After printing is completed, necessary post-processing is performed, such as removing support structures, polishing surfaces, cleaning, etc., to ensure that the printed guide meets the surgical requirements, obtain the 3D planned printed guide, and prepare for functional testing. The 3D planned printed guide is tested with real medical images (such as CT, MRI, etc.) to check the matching between the guide and the actual vascular structure. A biological model or a virtual simulation environment can be used to perform vascular function matching tests to evaluate the actual application effect of the guide. The test results are recorded and the guide vascular function matching test data is generated, including matching accuracy, possible deviations, and improvement suggestions. The vascular function matching test data is analyzed to identify the deficiencies or room for improvement in the guide. According to the test results, the 3D planned printed guide is optimized and adjusted, including redesigning the needle path, adjusting the guide thickness, optimizing the vascular matching area, etc., updating and optimizing the guide design, and generating a 3D planned optimized guide.

[0052] Preferably, step S43 includes the following steps: Step S431: performing vascular fitting virtual matching on the three-dimensional planned printed guide plate to generate guide plate vascular matching data; Step S432: Performing a functional evaluation on the three-dimensional planned printed guide plate through the guide plate blood vessel matching data to generate guide plate functional evaluation data, wherein the functional evaluation includes blood vessel patency, hemodynamic changes, and the degree of fit between the guide plate and the blood vessel wall; Step S433: Score the site adaptability of the guide functional evaluation data to generate guide vascular function matching test data; identify guide problems on the guide vascular function matching test data to generate guide design problem data; optimize the three-dimensional planned printing guide using the guide design problem data to generate a three-dimensional planned optimized guide.

[0053] In an embodiment of the present invention, a virtual surgical environment is established by importing the three-dimensional planned printed guide data and related vascular image data into virtual simulation software to ensure that the guide and vascular data can be accurately connected. Use a virtual matching algorithm (such as point cloud-based matching, shape matching, etc.) to analyze the fit between the guide and the blood vessel. Record the fit between the guide and the blood vessel, and generate guide blood vessel matching data, including fit, deviation, and potential problems. Determine the evaluation criteria, including vascular patency, hemodynamic changes, and the fit between the guide and the blood vessel wall. Analyze whether the guide affects the patency of the blood vessel, and evaluate whether the spatial relationship between the guide and the blood vessel is appropriate. Simulate the impact of the guide on blood flow and evaluate hemodynamic changes (such as flow rate, pressure, etc.). Measure the fit between the guide and the blood vessel wall to check whether there is a mismatch or excessive compression. Record the results of the functional evaluation and generate guide functional evaluation data, including detailed information and comprehensive evaluation of each indicator. Set scoring criteria for site adaptability, such as fit accuracy, blood flow impact, etc., score the guide functional evaluation data, and generate guide blood vessel functional matching test data. Record the site fit score to evaluate the guide's performance in actual application. Analyze the guide vascular function matching test data, identify problems in the design (such as mismatch, compression, blood flow problems, etc.), and record the guide design problem data, including the specific problems identified and the recommended improvement measures. Develop an optimization plan based on the guide design problem data, including redesigning the guide structure, adjusting the guide size, optimizing the vascular fitting area, etc., optimize the 3D planned printed guide, adjust the design to solve the identified problems, and output the optimized 3D planned guide to ensure that it better meets the surgical requirements and functional needs.

[0054] The beneficial effect of the present invention is that by acquiring CT and MRI images of the patient's lesion tissue, fuzzy vascular area angiography optimization and soft tissue identification are performed, and the generated patient lesion tissue characteristic images can fully display the structural characteristics and soft tissue conditions of the lesion. This comprehensive image data merger provides a detailed and accurate reference for subsequent preoperative planning, ensuring the accuracy of the surgical plan. The preoperative planning images generated based on the patient's lesion tissue characteristic images can help doctors plan surgical paths and strategies in advance. This step can reduce intraoperative uncertainty, and the image matching results are judged. When the matching result is true, AI is used for tissue segmentation. The generated AI tissue segmentation data can accurately mark various types of tissues in the surgical area and provide detailed tissue information for the operation; when the matching result is false, re-planning is performed to ensure the accuracy and applicability of the surgical plan. The preoperative planning images are screened through AI tissue segmentation data, and the obtained puncture needle track data and marked dose range data can accurately determine the surgical path and dose range. The puncture needle track data and marked dose range data are physically converted, and the generated three-dimensional planning guide provides an intuitive three-dimensional structural model. The initial three-dimensional planning guide is trimmed based on the preset needle track cutting distance and 3D printed. The generated three-dimensional planning printed guide has precise size and shape, which is suitable for actual surgical needs and convenient for doctors to use. The vascular function matching test of the three-dimensional planning printed guide can verify the adaptability of the guide to the patient's vascular structure and ensure the effectiveness and safety of the guide in actual use. The three-dimensional planning printed guide is optimized through test data, and the generated three-dimensional planning optimized guide can further improve the functionality and applicability of the guide. Therefore, the present invention improves the automatic generation accuracy and reliability of the three-dimensional vascular mapping guide through multimodal image fusion, AI automatic segmentation and functional testing.

[0055] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0056] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for automatically generating a human vascular mapping guide based on three-dimensional modeling. It is characterized in that The following steps are involved: Step S1: obtaining a CT image of the patient's lesion tissue and an MRI image of the patient's lesion tissue; performing fuzzy vascular area angiography optimization on the CT image of the patient's lesion tissue to generate a standard patient's lesion tissue optimized CT image; performing soft tissue identification on the MRI image of the patient's lesion tissue to generate an MRI image of the soft tissue area; merging the standard patient's lesion tissue optimized CT image and the patient's lesion tissue MRI image to generate a patient's lesion tissue characteristic image; Step S2: performing preoperative planning image simulation based on the patient's lesion tissue characteristic image to generate a preoperative planning image; performing image matching results on the preoperative planning image and the patient's lesion tissue characteristic image, and when the image matching result between the preoperative planning image and the patient's lesion tissue characteristic image is true, performing AI tissue segmentation on the preoperative planning image based on the patient's lesion tissue characteristic image to generate AI tissue segmentation data; when the image matching result between the preoperative planning image and the patient's lesion tissue characteristic image is false, re-planning the preoperative planning image; Step S3: Screen the puncture needle track and the marked dose range of the preoperative planning image through AI tissue segmentation data to obtain the puncture needle track data and the marked dose range data; perform entity conversion on the puncture needle track data and the marked dose range data to generate the puncture needle track entity data and the marked dose range entity data; perform vascular projection entity conversion on the puncture needle track entity data through the marked dose range entity data to generate vascular mapping entity data; perform three-dimensional reconstruction on the puncture needle track entity data, the local skin entity conversion data and the vascular mapping entity data to generate an initial three-dimensional planning guide; Step S4: cutting and printing the initial three-dimensional planned guide based on the preset needle channel cutting distance to generate a three-dimensional planned printed guide; performing a vascular function matching test on the three-dimensional planned printed guide to generate guide vascular function matching test data; optimizing the three-dimensional planned printed guide using the guide vascular function matching test data to generate a three-dimensional planned optimized guide.

2. The method for automatically generating a human blood vessel mapping guide based on three-dimensional modeling according to claim 1, It is characterized in that Step S1 includes the following steps: Step S11: obtaining a CT image of the patient's lesion tissue and an MRI image of the patient's lesion tissue by computer tomography technology; Step S12: performing image preprocessing on the patient's lesion tissue CT image and the patient's lesion tissue MRI image to generate a standard patient's lesion tissue CT image and a standard patient's lesion tissue MRI image, wherein the image preprocessing includes image filtering, image brightness enhancement and image gray value conversion; Step S13: extracting high-density structural tissue from the CT image of the lesion tissue of the standard patient to obtain a high-density structural tissue region image of the CT image; identifying the vascular fuzzy region of the high-density structural tissue region image of the CT image to generate vascular fuzzy region data of the CT image; Step S14: performing angiography enhancement on the CT image blood vessel blurred area data to generate CT image blood vessel enhanced area data; performing blood vessel blurred area image replacement on the CT image high-density structure tissue area according to the CT image blood vessel enhanced area data to generate a standard patient lesion tissue optimized CT image; Step S15: Perform soft tissue identification on the MRI image of the patient's lesion tissue to generate an MRI image soft tissue region image; merge the standard patient lesion tissue optimized CT image and the MRI image soft tissue region image to generate a patient lesion tissue characteristic image.

3. The method for automatically generating a human blood vessel mapping guide based on three-dimensional modeling according to claim 2, It is characterized in that The angiography enhancement of the CT image vascular blurred area data includes: Perform regional resolution detection on the CT image blood vessel blur region data to generate blood vessel blur region resolution detection data; perform image resolution improvement on the CT image blood vessel blur region data through the blood vessel blur region resolution detection data to generate blood vessel blur optimization region data; The vascular morphological features are extracted from the vascular fuzzy optimized area data to generate vascular morphological feature data; the vascular position is confirmed on the vascular fuzzy optimized area data based on the vascular morphological feature data to generate fuzzy area vascular position information data; the vascular fuzzy optimized area data is regionally enhanced by angiography based on the fuzzy area vascular position information data to generate CT image angiography enhanced area data.

4. The method for automatically generating a human blood vessel mapping guide based on three-dimensional modeling according to claim 1, It is characterized in that Step S2 includes the following steps: Step S21: Perform preoperative planning based on the patient's lesion tissue characteristic image to obtain preoperative planning data; Step S22: performing text semantic analysis on the preoperative planning data to generate preoperative planning text semantic data; performing keyword search on the preoperative planning data according to the preoperative planning text semantic data to generate preoperative planning keywords; Step S23: performing preoperative planning image simulation on the preoperative planning data according to the preoperative planning keywords to generate a preoperative planning image; performing image matching results on the preoperative planning image and the patient's lesion tissue characteristic image to determine if the image matching result between the preoperative planning image and the patient's lesion tissue characteristic image is true, performing AI tissue segmentation on the preoperative planning image based on the patient's lesion tissue characteristic image to generate AI tissue segmentation data; Step S24: When the image matching result between the preoperative planning image and the patient's lesion tissue feature image is false, the preoperative planning image is replanned.

5. The method for automatically generating a human blood vessel mapping guide based on three-dimensional modeling according to claim 4, It is characterized in that When the image matching result between the preoperative planning image and the patient's lesion tissue characteristic image is true, AI tissue segmentation of the preoperative planning image based on the patient's lesion tissue characteristic image includes: When the image matching result between the preoperative planning image and the patient's lesion tissue characteristic image is true, the patient's lesion tissue characteristic is subjected to tissue depth analysis to generate tissue depth analysis data; the patient's lesion tissue characteristic image is subjected to tissue depth level division through the tissue depth analysis data to generate lesion tissue depth level division data; Based on the lesion tissue depth level division data, the patient's lesion tissue characteristic images are separated by tissue level to generate lesion tissue level separation data; machine learning methods are used to perform tissue segmentation and annotation on preoperative planning images through lesion tissue level separation data to generate AI tissue segmentation data.

6. The method for automatically generating a human blood vessel mapping guide based on three-dimensional modeling according to claim 5, It is characterized in that Using machine learning methods to segment and annotate preoperative planning images through lesion tissue level separation data includes: The lesion tissue level separation data and the preset medical image resource library are accurately annotated to generate tissue annotation data; the tissue annotation data is enhanced to generate tissue annotation enhanced data; Divide the tissue annotation enhanced data into data sets to generate model training sets and model test sets; train the model training set through the U-Net deep learning model architecture to generate a tissue segmentation training model; use the model test set to evaluate and optimize the tissue segmentation training model to generate a tissue segmentation intelligent model; Import the preoperative planning images into the tissue segmentation intelligent model for intelligent tissue annotation to generate preoperative planning annotation images, and perform tissue segmentation on the preoperative planning annotation images to generate AI tissue segmentation data.

7. The method for automatically generating a human blood vessel mapping guide based on three-dimensional modeling according to claim 1, It is characterized in that Step S3 includes the following steps: Step S31: screening the puncture needle track and the marked dose range of the preoperative planning image by using AI tissue segmentation data to obtain puncture needle track data and marked dose range data; Step S32: using a 3D compiler to perform entity conversion on the puncture needle track data and the marked dose range data to generate puncture needle track entity data and marked dose range entity data; Step S33: selecting local skin tissue from the puncture needle track entity data by marking the dose range entity data to obtain local skin tissue data; Step S34: Perform local skin entity conversion on the local skin tissue data to generate local skin entity conversion data; perform vascular projection entity conversion on the local skin entity conversion data to generate vascular mapping entity data; perform three-dimensional reconstruction on the puncture needle track entity data, local skin entity conversion data and vascular mapping entity data to generate an initial three-dimensional planning guide.

8. The method for automatically generating a human blood vessel mapping guide based on three-dimensional modeling according to claim 7, It is characterized in that Step S34 includes the following steps: Step S341: discretizing the skin tissue surface of the local skin tissue data to generate discrete points on the skin tissue surface; performing point cloud conversion on the discrete points on the skin tissue surface to generate point cloud data on the skin tissue surface; Step S342: reconstructing the skin tissue surface point cloud data to generate skin tissue surface data; meshing the skin tissue surface data to generate skin tissue surface meshed data; performing normal extension and offset on the skin tissue surface meshed data to generate skin tissue surface offset data; Step S343: combining the skin tissue surface data and the skin tissue surface offset data to generate local skin entity transformation data; performing normal direction blood vessel projection on the local skin entity transformation data to generate blood vessel projection entity data; Step S344: Perform Boolean operation on the blood vessel transillumination entity data to generate blood vessel mapping entity data; perform three-dimensional reconstruction on the puncture needle track entity data, local skin entity conversion data and blood vessel mapping entity data to generate an initial three-dimensional planning guide.

9. The method for automatically generating a human blood vessel mapping guide based on three-dimensional modeling according to claim 1, It is characterized in that Step S4 includes the following steps: Step S41: cutting the initial three-dimensional planned guide plate based on the preset needle track cutting distance to generate a three-dimensional planned cutting guide plate; Step S42: 3D printing the three-dimensional planned cutting guide to generate a three-dimensional planned printing guide; Step S43: performing a vascular function matching test on the three-dimensional plan printing guide to generate guide vascular function matching test data; optimizing the three-dimensional plan printing guide based on the guide vascular function matching test data to generate a three-dimensional plan optimized guide.

10. The method for automatically generating a human blood vessel mapping guide based on three-dimensional modeling according to claim 1, It is characterized in that Step S43 includes the following steps: Step S431: performing vascular fitting virtual matching on the three-dimensional planned printed guide plate to generate guide plate vascular matching data; Step S432: Performing a functional evaluation on the three-dimensional planned printed guide plate through the guide plate blood vessel matching data to generate guide plate functional evaluation data, wherein the functional evaluation includes blood vessel patency, hemodynamic changes, and the degree of fit between the guide plate and the blood vessel wall; Step S433: Score the site adaptability of the guide functional evaluation data to generate guide vascular function matching test data; identify guide problems on the guide vascular function matching test data to generate guide design problem data; optimize the three-dimensional planned printing guide using the guide design problem data to generate a three-dimensional planned optimized guide.

Images