Gynecological pelvic and abdominal lymphatic cleaning prediction model training method and lymph node model

Abstract

The invention relates to a gynecological pelvic and abdominal lymphatic cleaning prediction model training method and a lymph node model.The training method comprises the steps that preoperative image data and intraoperative instrument operation parameters of historical patients are subjected to modeling processing, and a dynamic mechanics-fluid coupling lymph node model based on bionic materials is constructed; mechanical deformation and fluid flow of a lymphatic system in various clinical scenes are simulated; performing data acquisition processing on the used lymph node model, and quantifying the dynamic influence of the surgical operation on the lymphatic system; the relevance between historical operation and injury risk is deeply learned, and the lymphatic vessel rupture probability is output; real-time feedback control processing is carried out based on the lymphatic vessel rupture probability, a training correction instruction and a risk thermodynamic diagram are generated, and medical workers are guided to optimize surgical skills. Through fusion analysis of the high-fidelity bionic model and multi-modal data, dynamic simulation of the mechanical-fluid coupling effect in the operation and real-time quantification of the operation risk are achieved, and the accuracy and clinical transformation value of operation training are improved.

Description

Technical Field

[0001] This invention relates to the field of medical surgical simulation technology, specifically to a training method for a gynecological pelvic and abdominal lymph node dissection prediction model and a lymph node model. Background Technology

[0002] Gynecological pelvic and abdominal lymph node dissection is a key technique in the treatment of gynecological malignancies. Its core lies in the precise removal of affected lymph nodes while maximally protecting surrounding blood vessels, nerves, and the lymphatic system. Due to the complex anatomy of the pelvic and abdominal cavity, with its intricate distribution of lymphatic vessels and blood vessels, surgical procedures are highly susceptible to lymphatic vessel rupture due to improper traction or thermal diffusion from energy instruments, leading to postoperative complications such as lymphorrhea and infection. Therefore, standardized surgical training for medical personnel is crucial for improving clinical efficacy, and the construction of high-fidelity surgical simulation models and the establishment of dynamic feedback mechanisms are core technological requirements in this field.

[0003] Current clinical surgical training methods primarily rely on ex vivo animal tissues or static anatomical models, which struggle to simulate the dynamic interactions of intraoperative tissue deformation, lymphatic flow, and the thermal effects of energy instruments. Traditional physical models lack real-time responses to manipulation force and fluid resistance, making it impossible to quantify operational risks. While virtual simulation systems can partially simulate mechanical feedback, their limitations in tactile realism and anatomical fidelity make it difficult to reproduce the tissue mechanics and fluid dynamics behaviors under complex clinical scenarios. Furthermore, current technologies have not yet achieved personalized modeling based on historical surgical data, failing to provide adaptive training programs for different patient anatomical variations. Summary of the Invention

[0004] Based on this, the purpose of this invention is to provide a training method for a gynecological pelvic and abdominal lymph node dissection prediction model that can dynamically simulate intraoperative mechanical-fluid coupling effects, quantify operational risks in real time, and provide visual feedback.

[0005] The objective of this invention is achieved through the following solution:

[0006] In a first aspect, the present invention provides a method for training a predictive model for gynecological pelvic and abdominal lymph node dissection, comprising the following steps:

[0007] S1: Dynamic modeling and processing of preoperative imaging data and intraoperative instrument operation parameters of historical patients. By integrating the anatomical structure and operation parameters of historical cases, a dynamic mechanical-fluid coupling lymph node model based on biomimetic materials is constructed. The lymph node model is made of magnetorheological elastomer and thermosensitive hydrogel composite to simulate the mechanical deformation and fluid flow of the lymphatic system under various clinical scenarios.

[0008] S2: Multimodal data acquisition and processing of lymph node models used by medical staff, recording the operation force and fluid state through force feedback devices and optical sensors, extracting multi-source surgical features containing mechanical features, fluid flow features and energy device thermal diffusion features, and using multi-source surgical features to quantify the dynamic impact of surgical operations on the lymphatic system;

[0009] S3: Perform multimodal fusion analysis on multi-source surgical features, use a deep learning model to learn the correlation between historical operations and injury risks, and output the probability of lymphatic vessel rupture. The probability of lymphatic vessel rupture is used to assess the operational risks of medical staff.

[0010] S4: Real-time feedback control based on lymphatic vessel rupture probability. By dynamically adjusting operating parameters and generating visual signals, training correction instructions and risk heatmaps are generated. The training correction instructions are used to guide medical staff to optimize surgical skills.

[0011] In one embodiment, S1 of the gynecological pelvic and abdominal lymph node dissection prediction model training method provided by the present invention specifically includes the following steps:

[0012] S11: Pelvic lymph nodes and blood vessels are segmented from the preoperative imaging data of historical patients. A three-dimensional anatomical framework of biomimetic materials is generated through 3D printing technology. The three-dimensional anatomical framework includes a magnetorheological elastomer skeleton and hydrogel-filled lymphatic microchannels.

[0013] S12: Dynamic modeling is performed based on the intraoperative instrument operation parameters of historical patients. Combined with magnetic field control and operation force changes, dynamic stiffness function and fluid resistance function are generated. The dynamic stiffness function is used to quantify the real-time changes in material stiffness with magnetic field and operation force, and the fluid resistance function is used to simulate the dynamic adjustment of lymph flow resistance.

[0014] S13: Perform parameter fusion processing on three-dimensional anatomical structure data, dynamic stiffness function and fluid resistance function, and map mechanical parameters to anatomical structure through finite element mesh to construct a lymph node model based on biomimetic materials with dynamic mechanical-fluid coupling.

[0015] In one embodiment, S12 of the gynecological pelvic and abdominal lymph node dissection prediction model training method provided by the present invention specifically includes the following steps:

[0016] S121: Process the magnetic field strength and manipulation force in the intraoperative instrument operation parameters of historical patients, and combine them with the basic elastic modulus of the magnetorheological elastomer to generate a dynamic stiffness function. The expression of the dynamic stiffness function is as follows:

[0017]

[0018] Where E(t) is the dynamic stiffness function, representing the value of the dynamic stiffness function at time t, E0 is the basic elastic modulus of the magnetorheological elastomer, k is the response coefficient of the magnetorheological elastomer, α and β are weighting factors calibrated using historical surgical data, H(τ) is the magnetic field strength, and F is the operating force.

[0019] S122: The operating force and initial fluid resistance of the hydrogel microchannels in the intraoperative instrument operation parameters of historical patients are processed, and combined with the dynamic stiffness function, a fluid resistance function is generated. The expression of the fluid resistance function is:

[0020]

[0021] Where R(t) is the fluid resistance function, representing the fluid resistance function value at time t, R0 is the initial fluid resistance of the hydrogel microchannel, and γ is the hydrogel permeability correction coefficient.

[0022] In one embodiment, S13 of the gynecological pelvic and abdominal lymph node dissection prediction model training method provided by the present invention specifically includes the following steps:

[0023] S131: Based on the spatial parameters of lymphatic microchannels in a three-dimensional anatomical framework, fluorescent dye is injected into hydrogel-filled microchannels. The dye concentration and flow rate are controlled by a microfluidic pump to generate a visualized fluid flow path. The visualized fluid flow path is used to observe the flow status of lymph fluid in real time.

[0024] S132: Based on dynamic stiffness function and magnetic field control parameters, the magnetorheological elastomer skeleton of the three-dimensional anatomical structure is subjected to dynamic magnetic field loading treatment. The local magnetic field strength is adjusted in real time through electromagnetic coil array to generate biomimetic material hardness change synchronized with surgical operation.

[0025] S133: Based on the fluid resistance function and hydrogel permeability parameters, the fluid resistance of the lymphatic microchannels in the three-dimensional anatomical framework is dynamically adjusted. The cross-sectional area of ​​the channel is adjusted by a variable aperture valve to generate lymph flow velocity feedback that matches the operating force.

[0026] S134: Based on the visualized fluid flow path, biomimetic material hardness changes and flow velocity feedback, a multi-physics field coupling verification process is performed on the three-dimensional anatomical structure framework. By comparing historical surgical data to calibrate the model response, a high-fidelity biomimetic material dynamic mechanical-fluid coupling lymph node model is generated.

[0027] In one embodiment, S2 of the gynecological pelvic and abdominal lymph node dissection prediction model training method provided by the present invention specifically includes the following steps:

[0028] S21: Based on the force feedback data of medical staff training obtained from the lymph node model, the clamping force output by the force feedback device is sampled and processed with high precision. The time-series signal is segmented by an equal-interval time window to generate a mechanical feature vector. The mechanical feature vector is used to quantify the temporal changes in the force of the medical staff's operation.

[0029] S22: The fluid flow path visualization of lymphatic microchannels based on lymph node model optically tracks the displacement of fluorescent dyes, calculates the flow velocity and flow resistance change rate through high-speed camera and particle image velocimetry, and generates fluid dynamic parameters. The fluid dynamic parameters are used to evaluate the integrity of lymphatic vessels.

[0030] S23: Based on the thermal response characteristics of biomimetic materials, infrared thermal imaging is used to process the thermal diffusion range of the electrosurgical unit used by medical staff. Thermal damage distribution characteristics are generated by temperature gradient mapping. These thermal damage distribution characteristics are used to quantify the thermal impact of energy devices on biomimetic materials.

[0031] In one embodiment, S3 of the gynecological pelvic and abdominal lymph node dissection prediction model training method provided by the present invention specifically includes the following steps:

[0032] S31: The mechanical feature vectors, fluid dynamic parameters and thermal damage distribution characteristics of multi-source surgery features are divided into time windows. The continuous data is divided into independent time periods by sliding windows to generate a multimodal time series dataset.

[0033] S32: Based on temporal convolutional networks, perform correlation processing on multimodal temporal datasets, extract local and global features through multi-layer dilated convolutional kernels, and generate high-dimensional spatiotemporal features;

[0034] S33: Based on long short-term memory networks, the temporal dependencies of high-dimensional spatiotemporal features are modeled and processed. Key temporal information is filtered through a gating mechanism, and the probability of lymphatic vessel rupture is output.

[0035] In one embodiment, S4 of the gynecological pelvic and abdominal lymph node dissection prediction model training method provided by the present invention specifically includes the following steps:

[0036] S41: Based on the probability of lymphatic vessel rupture and a preset clinical safety threshold, the risk of injury is dynamically assessed and processed. By comparing with the threshold in real time, a fluorescent warning signal is generated during high-risk operations.

[0037] S42: Based on the characteristics of multi-source surgery and the probability of lymphatic vessel rupture, gradient optimization and boundary constraint processing are performed on the operating force of medical staff to generate a corrected operating force.

[0038] S43: Based on the fluorescent warning signal and the strength of the correction operation, perform visualization rendering on the risk area to generate training correction instructions and risk heat map.

[0039] Preferably, the present invention also provides a lymph node model for a gynecological pelvic and abdominal lymph node dissection prediction model, comprising:

[0040] The biomimetic lymph node unit is a replica of the anatomical structure of pelvic lymph nodes and para-aortic lymph nodes using 3D printing technology. The biomimetic lymph node unit is made of magnetorheological elastomer, and the stiffness of the magnetorheological elastomer is adjusted in real time by an external magnetic field to simulate the tissue deformation caused by traction during surgery.

[0041] The biomimetic lymphatic network is a microchannel structure made of temperature-sensitive hydrogel. The biomimetic lymphatic network is filled with fluorescently labeled fluid that simulates lymph fluid. The permeability of the biomimetic lymphatic network is dynamically adjusted according to the operating pressure of medical staff.

[0042] Bionic vascular stents are vascular skeletons formed by 3D printing of flexible polymer materials. The bionic vascular stents and bionic lymphatic vessels are distributed in a network-like interlacing pattern.

[0043] The dynamic response module includes a mechanical sensor and an internal flow sensor integrated into the biomimetic lymph node unit. The dynamic response module is used to collect operational force and fluid dynamic data in real time.

[0044] A fluorescent labeling feedback module, comprising a fluorescent labeling fluid and a fluorescent sensor filled within a biomimetic vascular stent;

[0045] Among them, the fluorescence sensor is used to trigger a lymphatic vessel rupture alarm signal when the operating force of medical staff exceeds a preset safety threshold. The lymphatic vessel rupture alarm signal is used to control the local fluorescence intensity enhancement of the fluorescently labeled fluid.

[0046] Thirdly, this application provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement any of the above-mentioned training methods for gynecological pelvic and abdominal lymph node dissection prediction models.

[0047] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-mentioned methods for training a gynecological pelvic and abdominal lymph node dissection prediction model.

[0048] In summary, the gynecological pelvic and abdominal lymph node dissection prediction model training method provided by this invention constructs a dynamic mechanics-fluid coupling lymph node model based on biomimetic materials. This enables full-dimensional simulation of the complex biomechanical behavior during gynecological pelvic and abdominal lymph node dissection surgery, effectively overcoming the technical limitations of traditional static models in reproducing intraoperative tissue dynamic deformation, lymphatic fluid flow, and the thermal effects of energy instruments. By integrating anatomical structures and operational parameters from historical cases, this lymph node model can reproduce the multi-physics response characteristics of the lymphatic system under traction, cutting, and thermal effects, allowing medical personnel to grasp the dynamic interaction between tissue mechanics boundaries and fluid dynamics in a highly simulated clinical setting. Based on multimodal data acquisition and deep learning-driven risk quantification analysis, the system overcomes the shortcomings of existing training methods that lack real-time operational feedback and personalized assessment. By dynamically analyzing the nonlinear correlation between operational force, fluid state, and thermal diffusion characteristics, it constructs a mapping model between surgical operation and lymphatic vessel damage risk, improving the accuracy and timeliness of intraoperative risk warning. By further integrating real-time feedback control mechanisms with visual correction instructions, dynamic optimization guidance can be provided for operating force, instrument movement trajectory, and energy release parameters. This reduces the risk of complications such as lymphatic vessel rupture and thermal damage spread caused by improper operation, and provides medical staff with a standardized training system that combines anatomical fidelity and interactive realism. Ultimately, this achieves the core goal of shortening the surgical learning curve and improving the safety of clinical operations.

[0049] To better understand and implement this invention, the following detailed description is provided in conjunction with the accompanying drawings. Attached Figure Description

[0050] Figure 1 A flowchart illustrating a method for training a gynecological pelvic and abdominal lymph node dissection prediction model, as provided in an embodiment of this application.

[0051] Figure 2 A schematic diagram of the process for generating a dynamic mechanical-fluid coupled lymph node model provided in an embodiment of this application;

[0052] Figure 3a A schematic diagram of the structure of a lymph node model with removed human simulation skin provided in an embodiment of this application;

[0053] Figure 3b A schematic diagram of the structure of a lymph node model retaining human simulated skin, provided in an embodiment of this application;

[0054] Figure 4 This is a schematic diagram illustrating the process of generating training correction instructions and risk heatmaps provided in the embodiments of this application. Detailed Implementation

[0055] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Preferred embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a thorough and complete understanding of the disclosure of the invention.

[0056] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0057] In one embodiment, such as Figure 1 As shown, a training method for a gynecological pelvic and abdominal lymph node dissection prediction model is provided. This embodiment illustrates the application of this method to a terminal. It is understood that this method can also be applied to a server, and can also be applied to a system including a terminal and a server, and implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0058] S1: Dynamic modeling and processing of preoperative imaging data and intraoperative instrument operation parameters of historical patients are performed. By integrating the anatomical structure and operation parameters of historical cases, a dynamic mechanical-fluid coupling lymph node model based on biomimetic materials is constructed. The lymph node model is made of magnetorheological elastomer and thermosensitive hydrogel composite and is used to simulate the mechanical deformation and fluid flow of the lymphatic system under various clinical scenarios.

[0059] Specifically, medical staff collect a large amount of preoperative imaging data from patients undergoing gynecological pelvic and abdominal lymph node dissection surgery and input it into the system. This preoperative imaging data includes high-resolution magnetic resonance imaging (MRI), computed tomography (CT), and other multimodal imaging data. These images provide detailed anatomical information about the lymph nodes and surrounding tissues, such as the location, size, and relationship of each group of lymph nodes within the pelvic and abdominal cavities to surrounding blood vessels and nerves. Simultaneously, the system collects intraoperative instrument operation parameters, including the movement trajectory of the surgical instruments, the force applied, the energy output intensity, and the duration of action. Through analysis of these parameters, the system establishes a dynamic model to simulate the interaction between surgical instruments and the lymphatic system, including the traction and cutting of lymph nodes by the instruments and the thermal effects of the energy instruments on surrounding tissues.

[0060] Building upon this foundation, the system integrates anatomical structures and operational parameters from historical cases to construct a dynamic mechanical-fluid coupling lymph node model based on biomimetic materials. This lymph node model is fabricated using a composite of magnetorheological elastomers and thermosensitive hydrogels. Magnetorheological elastomers possess excellent mechanical properties, capable of simulating the mechanical deformation of lymph nodes under different stress conditions, such as elastic deformation under tension and tissue separation during cutting. Thermosensitive hydrogels simulate the flow characteristics of lymphatic fluid and its behavior under temperature changes, such as the alteration of lymphatic fluid flow velocity and direction under the influence of heat generated during surgical procedures. By adjusting the ratio and composite method of the magnetorheological elastomer and thermosensitive hydrogel, this lymph node model can accurately simulate the mechanical deformation and fluid flow of the lymphatic system in various clinical scenarios, such as different surgical stages and different patient anatomical variations, providing a highly realistic physical basis for subsequent surgical training.

[0061] S2: Multimodal data acquisition and processing of lymph node models used by medical staff. The force and fluid state are recorded by force feedback devices and optical sensors. Multi-source surgical features containing mechanical characteristics, fluid flow characteristics and energy device thermal diffusion characteristics are extracted. Multi-source surgical features are used to quantify the dynamic impact of surgical operations on the lymphatic system.

[0062] Specifically, during the training of the gynecological pelvic and abdominal lymph node dissection prediction model, when medical staff perform simulated surgical operations using the lymph node model, the system simultaneously initiates multimodal data acquisition of the lymph node model. This process primarily relies on force feedback devices and optical sensors installed on the lymph node model to record key information during the surgical procedure. The force feedback device is tightly integrated into the surgical simulation instrument, enabling it to sense and accurately record various operational forces applied to the lymph node model by medical staff in real time during the surgical operation. These operational forces encompass the complex force changes generated when the instruments compress, pull, and cut the lymph nodes. The acquisition process is highly continuous and accurate, capable of meticulously capturing every subtle force fluctuation, thus providing a detailed and accurate mechanical data foundation for subsequent analysis. The optical sensor is used to monitor the fluid state of the lymph, capturing the flow velocity, flow rate changes, pressure gradient, and dynamic behavior of the lymph under the influence of the thermal effects of the energy instruments, such as changes in the local lymph flow direction caused by thermal diffusion. Leveraging advanced image processing algorithms, the system can efficiently analyze and process the vast amounts of image data acquired by optical sensors. By employing techniques such as image segmentation, feature extraction, and pattern recognition, the system can accurately extract mechanical and hydrodynamic characteristic parameters of the lymph node model during surgical procedures, including volume changes, surface displacement, and the velocity and direction fields of fluid flow. Furthermore, the system can also accurately analyze and quantify thermal effect characteristics, such as temperature gradients caused by thermal diffusion from energy instruments, using thermal imaging data captured by the optical sensors.

[0063] Furthermore, the system performs in-depth analysis and processing of the collected multi-dimensional data to extract multi-source surgical features, including mechanical characteristics, fluid flow characteristics, and energy device thermal diffusion characteristics. Mechanical characteristics specifically include parameters such as the magnitude, rate of change, location of application, and direction of force, reflecting the mechanical impact of surgical procedures on lymph nodes and surrounding tissues. Fluid flow characteristics encompass lymphatic fluid velocity, flow rate, pressure changes, and flow patterns, comprehensively quantifying the degree of interference of surgical procedures on lymphatic circulation. Energy device thermal diffusion characteristics involve the range of thermal diffusion, rate of temperature change, and duration of thermal action, used to assess the risk of thermal damage to the lymphatic system and surrounding tissues by energy devices during surgery. These multi-source surgical features collectively constitute a comprehensive quantitative description of the dynamic impact of surgical procedures, providing rich and accurate data support for subsequent risk assessment and skills training.

[0064] S3: Perform multimodal fusion analysis on multi-source surgical features, use a deep learning model to learn the correlation between historical operations and injury risks, and output the probability of lymphatic vessel rupture. The probability of lymphatic vessel rupture is used to assess the operational risks of medical staff.

[0065] Specifically, in the construction of the deep learning model, a large amount of historical surgical case data is input, including surgical operation characteristics such as manipulation force, fluid flow characteristics, and thermal diffusion characteristics of energy instruments, as well as the occurrence of corresponding postoperative complications such as lymphatic vessel rupture and lymphorrhea. Through the model's multi-layer neural network structure, the system automatically learns the implicit patterns and rules in the data, establishing a non-linear mapping relationship between surgical operation characteristics and the risk of lymphatic vessel rupture. In this embodiment, when the system inputs multi-source feature data of the current surgical operation, the deep learning model can quickly and accurately output the probability value of lymphatic vessel rupture under the current operation based on the learned patterns. This probability value quantitatively and intuitively reflects the degree of risk of lymphatic vessel rupture caused by the surgical operation, providing medical staff with real-time and accurate risk assessment results. This helps them adjust their operating strategies in a timely manner during surgical training, reducing the risk of complications.

[0066] S4: Real-time feedback control based on lymphatic vessel rupture probability. By dynamically adjusting operating parameters and generating visual signals, training correction instructions and risk heatmaps are generated. The training correction instructions are used to guide medical staff to optimize surgical skills.

[0067] Specifically, the system dynamically adjusts the operating parameters of surgical instruments based on risk assessment results. For example, when it detects that excessive force may increase the risk of lymphatic vessel rupture, the system automatically reduces the output power of the energy instruments to minimize potential damage to the lymphatic system from heat diffusion; or it adjusts the movement trajectory of the instruments to avoid high-risk areas, thereby reducing the possibility of lymphatic vessels being pulled or cut. Simultaneously, the system generates intuitive visual signals, such as using different colors to distinguish high-risk areas in the surgical simulation interface—for example, using red indicator lights to represent high risk and green to represent relative safety. It also uses graphical methods to display the dynamic changes in lymphatic fluid flow, providing real-time alerts to medical staff regarding the current risk level of the operation, enabling them to quickly perceive and adjust their actions.

[0068] The training correction instructions, combined with clinical surgical procedure guidelines, provide improvement suggestions for specific operational details of medical staff. For example, if medical staff frequently over-traction lymph nodes during a certain procedure, the system will prompt them to adjust the traction angle and force, adopting a more precise and gentler technique to reduce mechanical damage to the lymphatic system. The risk heatmap visually displays the risk distribution of the entire surgical area, using color variations to intuitively show the risk differences under different locations and procedures. This helps medical staff gain a deeper understanding of the risk factors in surgical procedures during training, enabling targeted skills training and optimization, and gradually improving their operational level and safety in gynecological pelvic and abdominal lymph node dissection surgery.

[0069] In summary, the gynecological pelvic and abdominal lymph node dissection prediction model training method provided by this invention constructs a dynamic mechanics-fluid coupling lymph node model based on biomimetic materials. This enables full-dimensional simulation of the complex biomechanical behavior during gynecological pelvic and abdominal lymph node dissection surgery, effectively overcoming the technical limitations of traditional static models in reproducing intraoperative tissue dynamic deformation, lymphatic fluid flow, and the thermal effects of energy instruments. By integrating anatomical structures and operational parameters from historical cases, this lymph node model can reproduce the multi-physics response characteristics of the lymphatic system under traction, cutting, and thermal effects, allowing medical personnel to grasp the dynamic interaction between tissue mechanics boundaries and fluid dynamics in a highly simulated clinical setting. Based on multimodal data acquisition and deep learning-driven risk quantification analysis, the system overcomes the shortcomings of existing training methods that lack real-time operational feedback and personalized assessment. By dynamically analyzing the nonlinear correlation between operational force, fluid state, and thermal diffusion characteristics, it constructs a mapping model between surgical operation and lymphatic vessel damage risk, improving the accuracy and timeliness of intraoperative risk warning. By further integrating real-time feedback control mechanisms with visual correction instructions, dynamic optimization guidance can be provided for operating force, instrument movement trajectory, and energy release parameters. This reduces the risk of complications such as lymphatic vessel rupture and thermal damage spread caused by improper operation, and provides medical staff with a standardized training system that combines anatomical fidelity and interactive realism. Ultimately, this achieves the core goal of shortening the surgical learning curve and improving the safety of clinical operations.

[0070] In one embodiment, such as Figure 2 As shown, S1 of the gynecological pelvic and abdominal lymph node dissection prediction model training method provided by the present invention specifically includes the following steps:

[0071] S11: Pelvic lymph nodes and blood vessels are segmented from preoperative imaging data of historical patients. A three-dimensional anatomical framework of biomimetic materials is generated using 3D printing technology. The three-dimensional anatomical framework includes a magnetorheological elastomer skeleton and hydrogel-filled lymphatic microchannels.

[0072] Specifically, the system extracts preoperative imaging data from historical patients from a medical imaging database. This data comes from high-resolution CT and MRI scans, which provide detailed anatomical information about pelvic lymph nodes, blood vessels, and surrounding tissues. CT scans, due to their high-density resolution, can clearly display the morphology and location of tissues such as bones, lymph nodes, and blood vessels; MRI provides even higher precision in soft tissue contrast, allowing for clearer differentiation of the boundaries between different soft tissues, thus providing a more accurate basis for the identification of lymph nodes and blood vessels.

[0073] During data acquisition, the system needs to perform preliminary sorting and screening of a large amount of image data, removing duplicate, blurry, or unsuitable images to ensure the data quality for subsequent processing. Simultaneously, the system will standardize the data, including converting images to DICOM format and adjusting grayscale and contrast to meet the requirements of subsequent segmentation. Furthermore, to protect patient privacy, the system will anonymize patient identification information in the image data to ensure data security and compliance. Through this series of data acquisition and preprocessing operations, the system provides a high-quality, standardized image data foundation for subsequent segmentation.

[0074] Specifically, the system utilizes image segmentation algorithms to segment preoperative image data, separating key structures such as lymph nodes and blood vessels within the pelvic cavity. Preferably, the system employs a deep learning-based U-Net architecture, using convolutional neural networks to extract features and segment the image data. This allows the system to automatically learn the features of target structures like lymph nodes and blood vessels, achieving automated segmentation. During segmentation, the system first preprocesses the image data, including image enhancement and normalization, to improve segmentation accuracy. Then, the trained U-Net model performs pixel-by-pixel classification of the image data, separating target structures such as lymph nodes and blood vessels from the background tissue. To further enhance the accuracy of the segmentation results, the system incorporates post-processing techniques, such as Conditional Random Fields (CRF), to optimize the results and remove small isolated regions and noise points. The system also visualizes the segmentation results for doctors or technicians to evaluate and verify the effectiveness. When necessary, medical staff can manually correct the segmentation results to ensure that the segmented lymph nodes and blood vessels closely match the actual structures. By segmenting historical patient image data, the system can acquire a wealth of information on the morphology and location of lymph nodes and blood vessels, providing an accurate digital model for the subsequent construction of a three-dimensional anatomical framework. The deep learning-based segmentation algorithm boasts high accuracy and efficiency, providing reliable data support for model construction.

[0075] After segmenting and processing the image data, the system utilizes 3D printing technology to print a biomimetic three-dimensional anatomical framework based on the generated three-dimensional digital model and biomimetic materials. This framework uses magnetorheological elastomers as the skeleton material. These materials exhibit significant changes in mechanical properties under a magnetic field, simulating the mechanical response of lymph node tissue during surgical procedures. For example, a magnetorheological elastomer with carbonyl iron powder uniformly dispersed in a polyborosiloxane matrix can achieve a relative change in shear modulus of up to 878%, demonstrating excellent magnetically controlled stiffness. Simultaneously, hydrogel-filled lymphatic microchannels are constructed within the framework. Hydrogel materials possess good biocompatibility and thermosensitivity, undergoing a sol-gel transition at specific temperatures to simulate the flow characteristics of lymphatic fluid. For instance, chitosan (CS) thermosensitive hydrogel forms a hydrogel above 37°C and can be used to simulate lymphatic fluid flow. During the 3D printing process, the system needs to control the printing parameters to ensure that the anatomical framework is highly consistent with the real anatomical structure in terms of morphology, size, and physical properties, providing a solid foundation for subsequent surgical simulation training.

[0076] S12: Dynamic modeling is performed based on the intraoperative instrument operation parameters of historical patients. Combined with magnetic field control and operation force changes, dynamic stiffness function and fluid resistance function are generated. The dynamic stiffness function is used to quantify the real-time changes in material stiffness with magnetic field and operation force, and the fluid resistance function is used to simulate the dynamic adjustment of lymph flow resistance.

[0077] Specifically, the dynamic stiffness function is constructed based on the magnetorheological effect principle of magnetorheological elastomers. By studying the intrinsic relationship between magnetic field strength, operating force, and material stiffness, a mathematical model is established to quantify the real-time changes in material stiffness. For example, the elastic modulus of a magnetorheological elastomer can change significantly under different magnetic field strengths. The system utilizes this characteristic to enable the model to adjust its stiffness in real time according to changes in the magnetic field and force during surgical operations, thereby realistically simulating the mechanical properties of lymph node tissue. The fluid resistance function focuses on simulating the changes in flow resistance of lymph fluid within lymphatic microchannels. The system analyzes the rheological properties of lymph fluid and the geometry of lymphatic microchannels, and combines the rheological parameters of the hydrogel material to construct a function model that can dynamically adjust fluid resistance. This function calculates the lymph fluid flow resistance in real time based on factors such as surgical operating force and instrument movement speed, achieving a high degree of simulation of the lymph fluid flow process. Preferably, the dynamic stiffness function and the fluid resistance function are generated through the following steps:

[0078] S121: Process the magnetic field strength and operating force in the intraoperative instrument operation parameters of historical patients, and generate a dynamic stiffness function by combining the basic elastic modulus of the magnetorheological elastomer.

[0079] Specifically, the system extracts parameters such as magnetic field strength and manipulation force from a large amount of historical surgical data. These parameters cover the use of surgical instruments in different types and stages of surgery, including instrument type, magnitude and direction of manipulation force, manipulation frequency, and power and heat diffusion range of energy instruments. Through statistical analysis and mining of this historical data, the system can gain a deep understanding of the complex interaction between instruments and lymph node tissue during surgical operations, thus providing rich data support for subsequent dynamic modeling. Simultaneously, the system combines the fundamental elastic modulus of the magnetorheological elastomer with the extracted magnetic field strength and manipulation force data to generate a dynamic stiffness function. A magnetorheological elastomer is a smart material with a unique magnetorheological effect; its fundamental elastic modulus is the inherent elastic property parameter of the material in the absence of a magnetic field. When the magnetic field strength or manipulation force changes, the mechanical properties of the magnetorheological elastomer will change significantly. Preferably, the expression for the dynamic stiffness function is:

[0080]

[0081] Where E(t) is the dynamic stiffness function, representing the value of the dynamic stiffness function at time t, E0 is the fundamental elastic modulus of the magnetorheological elastomer, k is the response coefficient of the magnetorheological elastomer, α and β are weighting factors calibrated using historical surgical data, H(τ) is the magnetic field strength, and F is the operating force. This function can quantify the changes in the mechanical properties of the magnetorheological elastomer during surgical operations in real time, providing accurate mechanical feedback for surgical simulation training. For example, a magnetorheological elastomer with carbonyl iron powder uniformly dispersed in a polyborosilicate elastomer can exhibit a relative change in shear modulus of up to 878%, demonstrating excellent magnetically controlled stiffness performance. In this way, the system can simulate the mechanical response of lymph node tissue during surgical operations, such as the change in the elastic modulus of lymph node tissue under different magnetic field strengths and operating forces, thereby providing medical staff with a highly realistic surgical operation experience.

[0082] S122: Process the operating force and initial fluid resistance of the hydrogel microchannel in the intraoperative instrument operation parameters of historical patients, and generate a fluid resistance function by combining the dynamic stiffness function.

[0083] Specifically, the initial fluid resistance of the hydrogel microchannel is calculated based on factors such as the microchannel geometry, the rheological properties of the hydrogel material, and the rheological parameters of the lymph fluid. During surgical procedures, changes in the force applied directly affect the flow state of the lymph fluid within the microchannel, thereby altering the fluid resistance. The system can simulate the dynamic changes in lymph fluid flow resistance in real time by constructing a fluid resistance function. Preferably, the expression for the fluid resistance function is:

[0084]

[0085] Where R(t) is the fluid resistance function, representing the fluid resistance function value at time t, R0 is the initial fluid resistance of the hydrogel microchannel, and γ is the hydrogel permeability correction coefficient. This function comprehensively considers the influence of the rate of change of operating force and the change of material stiffness on fluid resistance, enabling a high degree of simulation of lymph flow resistance. For example, when surgical instruments apply a large operating force to the lymph node, the flow velocity and direction of lymph in the microchannel will change, and the fluid resistance will increase accordingly. Through the fluid resistance function, the system can calculate this change in real time, providing medical staff with accurate fluid dynamics feedback, helping them to better understand and master the flow characteristics of lymph during surgical operations, thereby effectively improving the safety and precision of surgical operations.

[0086] S13: Perform parameter fusion processing on three-dimensional anatomical structure data, dynamic stiffness function and fluid resistance function, and map mechanical parameters to anatomical structure through finite element mesh to construct a lymph node model based on biomimetic materials with dynamic mechanical-fluid coupling.

[0087] Specifically, the finite element mesh mapping technology maps mechanical parameters onto a three-dimensional anatomical framework using finite element meshes. This allows the system to precisely allocate parameters from the dynamic stiffness function and fluid resistance function to various regions and parts of the model, ensuring that each part of the model exhibits mechanical and fluid properties that match the actual physiological environment based on its anatomical location and structural characteristics. Preferably, the lymph node model is constructed through the following steps:

[0088] S131: Based on the spatial parameters of lymphatic microchannels in a three-dimensional anatomical framework, fluorescent dye is injected into hydrogel-filled microchannels. The dye concentration and flow rate are controlled by a microfluidic pump to generate a visualized fluid flow path. The visualized fluid flow path is used to observe the flow status of lymph fluid in real time.

[0089] Specifically, the system utilizes a high-precision microfluidic pump to control the concentration and flow rate of the dye, ensuring stable and controllable dye flow within the microchannels. The selection of fluorescent dyes must meet conditions such as good biocompatibility, high fluorescence intensity, and resistance to fading, ensuring clear observation of lymph flow during simulation. For example, FITC-labeled dextran is an ideal fluorescent dye, stably simulating lymph flow at 37°C. The microfluidic pump precisely controls the injection speed and pressure, creating a uniform flow path for the dye within the microchannels. These visualized fluid flow paths not only reflect the flow direction and velocity of lymph in real time but also help medical personnel intuitively observe the dynamic changes of lymph under different operating conditions, providing crucial visual feedback for surgical simulation training. Through fluorescence imaging technology, the system can capture these flow paths in real time and overlay them onto a three-dimensional anatomical framework, allowing medical personnel to clearly see the lymph flow during simulated surgery, thereby better understanding and mastering the impact of surgical procedures on the lymphatic system.

[0090] S132: Based on dynamic stiffness function and magnetic field control parameters, dynamic magnetic field loading is applied to the magnetorheological elastomer skeleton of the three-dimensional anatomical structure framework. The local magnetic field strength is adjusted in real time through an electromagnetic coil array to generate biomimetic material hardness changes synchronized with surgical operations.

[0091] Specifically, an array of electromagnetic coils is arranged around a three-dimensional anatomical framework, and the strength and distribution of the magnetic field they generate can be precisely adjusted by controlling the magnitude and direction of the current. A dynamic stiffness function provides a mathematical model of how material stiffness changes with the magnetic field and the applied force. Using this model, combined with real-time monitored surgical parameters, the system dynamically adjusts the current in the electromagnetic coil array, thereby precisely controlling the stiffness changes of the magnetorheological elastomer framework. For example, when surgical instruments apply significant force to a lymph node, the system increases the magnetic field strength accordingly, increasing the stiffness of the magnetorheological elastomer and simulating the mechanical response of lymph node tissue during actual surgery. In this way, the system can simulate the changes in the mechanical properties of lymph nodes under different surgical conditions in real time, providing medical personnel with a highly realistic surgical experience, helping them better master surgical techniques, and improving the safety and precision of surgery.

[0092] S133: Based on the fluid resistance function and hydrogel permeability parameters, the fluid resistance of the lymphatic microchannels in the three-dimensional anatomical framework is dynamically adjusted. The cross-sectional area of ​​the flow channel is adjusted by a variable aperture valve to generate lymph flow velocity feedback that matches the operating force.

[0093] Specifically, the valve's opening and closing degree is dynamically adjusted by the control system based on real-time operating force and fluid resistance function, enabling continuous regulation from fully open to fully closed. For example, during surgical simulation, when the operating force increases, the system increases fluid resistance by decreasing the valve orifice, thereby reducing the flow rate; conversely, it increases the orifice to reduce resistance and restore the flow rate. This dynamic adjustment mechanism can simulate the real-world flow rate changes of lymph fluid under different operating forces, making the surgical simulation more realistic. By precisely controlling fluid resistance, the system provides medical staff with rich fluid dynamics feedback, helping them to more accurately grasp the fluid flow characteristics during surgical procedures, thereby improving the safety and effectiveness of the surgical operation.

[0094] S134: Based on the visualized fluid flow path, biomimetic material hardness changes and flow velocity feedback, a multi-physics field coupling verification process is performed on the three-dimensional anatomical structure framework. By comparing historical surgical data to calibrate the model response, a high-fidelity biomimetic material dynamic mechanical-fluid coupling lymph node model is generated.

[0095] Specifically, in the multiphysics coupling verification process, the system integrates information from multiple physical fields, including mechanics, fluid dynamics, and electromagnetics, to ensure the model's response consistency under different physical conditions. For example, the system compares the visualized fluid flow path with images of lymph flow in actual surgery to verify the accuracy of the fluid dynamics simulation; simultaneously, it compares the hardness changes of the biomimetic material with mechanical feedback data from surgical operations to ensure the realism of the mechanical simulation. In this way, the system can comprehensively calibrate and optimize the model, generating a high-fidelity dynamic mechanical-fluid coupled lymph node model of biomimetic materials. The constructed lymph node model is shown below. Figure 3a and Figure 3b As shown, where, Figure 3b This is a lymph node model with a human-like skin in a non-surgical state. Figure 3a This model, created by removing the human body's simulated skin from the lymph node model, not only closely resembles the real lymph node in morphology but also simulates the actual surgical scenario in terms of mechanical properties and hydrodynamic behavior. It can realistically reflect various physical phenomena and tissue responses during surgical operations, providing medical staff with a high-quality surgical simulation training environment, which helps improve the success rate of surgery and reduce the risk of postoperative complications.

[0096] The aforementioned training method for a predictive model of gynecological pelvic and abdominal lymph node dissection integrates preoperative imaging data and intraoperative operational parameters from historical patients to construct a biomimetic lymph node model that combines high anatomical fidelity with dynamic response characteristics. This solves the technical challenge of traditional ex vivo tissue or static models failing to simulate the complex mechanical-fluid coupling effects during pelvic and abdominal lymph node dissection. The biomimetic lymph node model is based on a composite structure design of magnetorheological elastomer and hydrogel. Through 3D printing technology, it accurately reproduces the individualized pelvic vascular-lymphatic vessel topology network, enabling the anatomical framework to not only highly replicate the spatial configuration of real tissue but also achieve dynamic adaptive adjustment of material stiffness and fluid resistance through magnetic field control and mechanical parameter mapping. Based on a cross-scale parameter fusion mechanism using finite element meshes, the model can respond in real time to the traction, cutting, and energy effects of surgical instruments, simultaneously simulating the interaction effects of lymphatic vessel deformation and fluid flow. This overcomes the inherent limitations of existing training models in reproducing intraoperative tissue viscoelastic deformation, lymphatic fluid seepage, and thermal diffusion behavior. By further integrating the dynamic stiffness function and fluid resistance function generated from historical surgical data, the model can autonomously adjust its biomechanical properties according to changes in magnetic field strength and operational force. This provides medical staff with a highly realistic multi-physics interactive training environment, which can significantly improve their understanding of the correlation between surgical operations and tissue dynamic responses. It can effectively avoid the risk of lymphatic vessel mechanical rupture or thermal damage caused by deviations in traction angles or uncontrolled energy release, ultimately forming a standardized surgical training system that combines anatomical variation adaptability with real-time risk feedback.

[0097] In one embodiment, S2 of the gynecological pelvic and abdominal lymph node dissection prediction model training method provided by the present invention specifically includes the following steps:

[0098] S21: Based on the force feedback data of medical staff training obtained from the lymph node model, the clamping force output by the force feedback device is sampled and processed with high precision. The time-series signal is segmented by an equal-interval time window to generate a mechanical feature vector. The mechanical feature vector is used to quantify the temporal changes in the force of the medical staff's operation.

[0099] Specifically, the force feedback device is installed at the end of the surgical simulation instrument, closely attached to the lymph node model. Its core component, a six-axis force sensor, can sense the clamping force applied to the lymph node model by medical staff during surgical operations in real time, ensuring high fidelity of the force feedback data. Specifically, the system performs high-precision sampling processing on this data to capture subtle fluctuations in force changes. The time-series signal is segmented by equally spaced time windows, the length of which is adjusted according to the characteristics of the surgical operation, transforming the continuous force feedback signal into multiple discrete data segments. Further processing extracts key feature parameters, including the mean force, variance, peak value, and rate of force change, generating a mechanical feature vector. For example, during lymph node dissection, the time window is set to 50 milliseconds, allowing the system to accurately capture every minute force adjustment by the medical staff; while during larger tissue traction operations, the time window is extended to 200 milliseconds to accommodate relatively slow force changes. These mechanical feature vectors comprehensively quantify the temporal changes in force exerted by the medical staff, providing crucial evidence for subsequent accurate assessment of surgical operation risks.

[0100] S22: The fluid flow path visualization of lymphatic vessel microchannels based on the lymph node model performs optical tracking processing on the displacement of fluorescent dyes, calculates the flow velocity and flow resistance change rate through high-speed imaging and particle image velocimetry, and generates fluid dynamic parameters, which are used to evaluate the integrity of lymphatic vessels.

[0101] Specifically, a high-speed camera is mounted directly above the lymph node model, its field of view precisely covering the entire microchannel region, enabling it to capture the rapid flow of fluorescent dye within the microchannels at a high frame rate. Simultaneously, Particle Image Velocimetry (PIV) technology accurately calculates the fluid velocity field by analyzing the displacement of fluorescent dye particles in adjacent frames. More specifically, the system combines the spatial parameters of the lymphatic microchannels with fluid dynamics principles to further calculate the rate of change of flow resistance. For example, when the operation of medical personnel causes localized pressure deformation in the microchannel, the flow velocity in that area increases significantly, and the flow resistance increases accordingly. Through optical tracking processing, the system monitors these changes in real time and generates fluid dynamic parameters. These parameters not only reflect the flow state of the lymph fluid but also accurately assess the integrity of the lymphatic vessels. During surgical procedures, the integrity of the lymphatic vessels is crucial for preventing postoperative complications such as lymphorrhea. By analyzing the fluid dynamic parameters, the system can promptly detect potential risks of lymphatic vessel damage, providing vital feedback to medical personnel during surgical procedures.

[0102] S23: Based on the thermal response characteristics of biomimetic materials, infrared thermal imaging is used to process the thermal diffusion range of the electrosurgical unit used by medical staff. Thermal damage distribution characteristics are generated by temperature gradient mapping. These thermal damage distribution characteristics are used to quantify the thermal impact of energy devices on biomimetic materials.

[0103] Specifically, the infrared thermal imager is installed to the side of the surgical simulation space. Its field of view is precisely adjusted to completely cover the area affected by the electrosurgical unit and the surrounding biomimetic material tissue. This thermal imager features high spatial resolution and rapid imaging, with a temperature resolution of 0.1℃ and a spatial resolution typically between tens and hundreds of micrometers. During the surgical simulation, when the electrosurgical unit contacts the biomimetic material, the infrared thermal imager captures the temperature changes on the material surface in real time. The system uses a temperature gradient mapping algorithm to convert the collected temperature data into thermal damage distribution characteristics. These characteristics are presented intuitively in the form of two-dimensional or three-dimensional thermal maps, clearly showing the range and degree of heat diffusion. For example, when the electrosurgical power is too high or the action time is too long, the heat diffusion range will significantly increase, and the temperature gradient in the corresponding area of ​​the thermal damage distribution characteristics will rise significantly, indicating that the biomimetic material in that area has suffered severe thermal damage. These thermal damage distribution characteristics accurately quantify the thermal impact of the energy device on the biomimetic material, helping medical personnel to intuitively understand the distribution of thermal damage, thereby optimizing the electrosurgical unit's operating parameters and reducing the impact of thermal damage on surrounding tissues.

[0104] In one embodiment, S3 of the gynecological pelvic and abdominal lymph node dissection prediction model training method provided by the present invention specifically includes the following steps:

[0105] S31: The mechanical feature vectors, fluid dynamic parameters and thermal damage distribution characteristics of multi-source surgery features are divided into time windows. The continuous data is divided into independent time periods by sliding windows to generate a multimodal time series dataset.

[0106] Specifically, the system synchronously aligns multi-source surgical features to ensure consistency of mechanical feature vectors, fluid dynamic parameters, and thermal damage distribution characteristics along the time axis. Simultaneously, a sliding window technique is employed to segment continuous data, dividing it into multiple independent time periods to form a multimodal time-series dataset. The length and step size of the sliding window are optimized based on the specific characteristics of the surgical procedure to achieve appropriate overlap between data segments, ensuring data continuity and integrity. For example, during delicate lymph node dissection, the window length can be set to 1 second with a step size of 0.5 seconds, allowing for precise capture of short-term changes during the procedure; while for longer procedures, such as lymphatic vessel ligation, the window length can be extended to 5 seconds, with the step size adjusted accordingly to 2.5 seconds.

[0107] This sliding window segmentation method transforms the original continuous data into multiple fixed-length data segments. Each segment contains mechanical feature vectors, fluid dynamic parameters, and thermal damage distribution characteristics for that time period, thus generating a multimodal time-series dataset. This provides a structured data foundation for subsequent feature extraction and analysis. This dataset not only preserves the temporal relationships of the time-series data but also integrates multimodal feature information, providing high-quality input data for training deep learning models.

[0108] S32: Based on temporal convolutional networks, perform correlation processing on multimodal temporal datasets, extract local and global features through multi-layer dilated convolutional kernels, and generate high-dimensional spatiotemporal features.

[0109] Specifically, a Temporal Convolutional Network (TCN) is a deep learning architecture specifically designed for processing time-series data. It automatically extracts local and global features from the data by sliding one-dimensional convolutional operations along the time axis. The TCN system comprises multiple dilated convolutional layers, each using a kernel with a different dilation rate to expand the network's receptive field and capture dependencies over longer timeframes. For example, the first dilated kernel might have a dilation rate of 1, primarily for extracting short-term local features; the second layer might have a dilation rate of 2, capturing features over slightly longer timeframes; subsequent layers would progressively increase the dilation rate until the critical time window of the entire surgical procedure is covered. This multi-layered dilated kernel design enables the network to extract features at different time scales, capturing both instantaneous changes in operational force and the gradual changes in fluid flow rate, as well as the cumulative effects of thermal damage.

[0110] Each layer of the network contains multiple convolutional kernels, each responsible for extracting a specific type of feature. For example, some kernels focus on extracting force variation patterns in mechanical feature vectors, while others emphasize flow velocity fluctuations in fluid dynamic parameters. Still others are used to identify temperature gradient changes in thermal damage distribution features. Through layer-by-layer convolutional operations and nonlinear activation functions, the network transforms multimodal time-series datasets into high-dimensional spatiotemporal features. These features not only contain information from the original data but also incorporate the correlations between different modalities, providing richer feature representations for subsequent time-series modeling.

[0111] S33: Based on long short-term memory networks, the temporal dependencies of high-dimensional spatiotemporal features are modeled and processed. Key temporal information is filtered through a gating mechanism, and the probability of lymphatic vessel rupture is output.

[0112] Specifically, Long Short-Term Memory (LSTM) networks are a special type of Recurrent Neural Network (RNN) architecture capable of effectively processing and predicting long-term dependencies in time-series data. In practice, the system inputs a high-dimensional spatiotemporal feature sequence into the LSTM network. LSTM uses its unique gating mechanism—input gate, forget gate, and output gate—to filter and update information in the feature sequence. The input gate controls the degree to which feature information from the current moment enters the cell state; the forget gate determines which information in the cell state needs to be forgotten; and the output gate determines which parts of the cell state will be output as the hidden state at the current moment. For example, when processing surgical operation data, the input gate might selectively allow information about changes in the intensity of the current operation to enter the cell state. If the current operation is weakly correlated with a previous key operation, the forget gate might discard some information from that previous operation. The output gate then outputs the feature vector that best represents the surgical state at the current moment, based on the cell state. In this way, LSTM can automatically learn and filter out the most critical temporal information for predicting the probability of lymphatic vessel rupture, ignoring relatively unimportant information, thereby improving the model's prediction accuracy. After processing by multiple layers of LSTM, the final output layer of the system maps the network's output to the probability of lymphatic vessel rupture. This probability value is represented by a number between 0 and 1, intuitively reflecting the risk level of lymphatic vessel rupture caused by the current surgical procedure. This provides medical staff with important real-time risk warnings, helping them to adjust surgical strategies in a timely manner, reduce surgical risks, and improve the safety and accuracy of the surgery.

[0113] In one embodiment, such as Figure 4 As shown, S4 of the gynecological pelvic and abdominal lymph node dissection prediction model training method provided by the present invention specifically includes the following steps:

[0114] S41: Based on the probability of lymphatic vessel rupture and a preset clinical safety threshold, the risk of injury is dynamically assessed and processed. By comparing with the threshold in real time, a fluorescent warning signal is generated for high-risk operations.

[0115] Specifically, the system first stores a preset clinical safety threshold in memory. This threshold, determined based on extensive clinical data and expert experience, distinguishes safe and high-risk areas during surgical procedures. During surgical simulation, the system calculates the probability of lymphatic vessel rupture in real time and compares it to this clinical safety threshold. When the probability exceeds the preset threshold, the system immediately triggers an alarm mechanism, generating a fluorescent warning signal. The fluorescent warning signal is generated based on fluorescent markers and sensors installed around the lymph node model. When a high-risk operation occurs, the system controls the fluorescent sensors to generate ultraviolet light, flashing at a specific frequency to excite the fluorescent markers, thus visually displaying the high-risk area in the surgical field. For example, during a cutting operation near a lymphatic vessel, if the force applied is too great or the heat spread is too wide, the system detects a sharp increase in the probability of lymphatic vessel rupture. Once this probability exceeds the threshold, a fluorescent warning is immediately activated to alert medical staff to the operational risks. In this way, the system can provide real-time and intuitive feedback to medical staff on the risk status of the surgical procedure, helping them adjust their operational strategies in a timely manner and avoid potential surgical complications.

[0116] S42: Based on the characteristics of multi-source surgery and the probability of lymphatic vessel rupture, the intensity of operation by medical staff is optimized by gradient and boundary constraints to generate a corrected intensity of operation.

[0117] Specifically, the system analyzes multi-source surgical features such as mechanical feature vectors, fluid dynamic parameters, and thermal damage distribution characteristics, combined with the probability of lymphatic vessel rupture, to establish a loss function. This loss function measures the difference between the current operating force and the ideal operating force. Preferably, the system can employ a gradient descent algorithm to continuously adjust the operating force parameters to minimize the value of the loss function. Simultaneously, the system sets reasonable boundary constraints to ensure that the optimized operating force remains within a safe range. The boundary constraints are set with reference to a large amount of clinical surgical data, determining the maximum safe operating force corresponding to different surgical stages and tissue types. For example, when performing fine lymph node dissection, the system sets a lower upper limit for the operating force, while appropriately increasing the upper limit when retracting larger tissue blocks. After gradient optimization and boundary constraint processing, the system generates a corrected operating force and provides real-time feedback to medical staff via a force feedback device, guiding them to adjust the operating force, reducing surgical risks, and improving the safety and precision of the surgery.

[0118] S43: Based on the fluorescent warning signal and the strength of the correction operation, perform visualization rendering on the risk area to generate training correction instructions and risk heat map.

[0119] Specifically, the system combines fluorescent warning signals with a three-dimensional anatomical framework, using computer graphics technology to render a risk heatmap in real time within the surgical simulation interface. The risk heatmap uses different colors to represent different levels of risk, typically red for high-risk areas, yellow for medium-risk areas, and green for low-risk areas. Simultaneously, the system converts corrective manipulation force into visual prompts, overlaid on the risk heatmap, providing intuitive operational guidance for medical staff. For example, when medical staff apply excessive force in a certain area, the system displays a red warning for that area and suggests reducing the direction and magnitude of the force through arrows or text prompts. Furthermore, the system generates detailed training correction instructions, provided to medical staff in text or voice format, including specific operational adjustment suggestions such as changing the angle, direction, or force of instruments. In this way, the system not only helps medical staff understand the risk distribution during surgical procedures in real time but also provides them with specific improvement measures, thereby effectively improving the safety and accuracy of surgical procedures and ultimately optimizing clinical outcomes.

[0120] The above-mentioned training method for predictive models of gynecological pelvic and abdominal lymph node dissection, through the synergistic integration of dynamic feedback control mechanism and visual risk mapping technology, can effectively overcome the blindness of surgical operation and the risk of complications caused by the lack of real-time risk warning and operation correction guidance during surgery in traditional training methods. This method, based on dynamic threshold discrimination of lymphatic vessel rupture probability and fluorescent warning signal triggering mechanism, achieves real-time perception and visual prompts for high-risk operation nodes, thus overcoming the technical limitations of existing static models that cannot simulate the dynamic evolution of intraoperative tissue damage. Furthermore, through gradient optimization of multi-source surgical features and boundary constraint algorithms for operational force, it constructs an adaptive mapping relationship between surgical instrument motion parameters and safe operating ranges, precisely guiding medical staff to adjust key operational parameters such as traction force and energy release within biomechanical safety thresholds, significantly reducing the risk of lymphatic vessel mechanical rupture and thermal damage spread caused by exceeding operational limits. Combined with real-time rendering of risk heatmaps and dynamic generation of correction instructions, it forms an immersive training feedback system that combines spatial positioning accuracy with timely operational guidance, enabling medical staff to intuitively identify anatomically sensitive areas and simultaneously optimize surgical operation paths, thereby comprehensively improving the standardized operation level and clinical safety of lymph node dissection surgery in complex pelvic and abdominal environments.

[0121] Preferably, the present invention also provides a lymph node model for a gynecological pelvic and abdominal lymph node dissection prediction model, comprising:

[0122] The biomimetic lymph node unit is a replica of the anatomical structure of pelvic lymph nodes and para-aortic lymph nodes using 3D printing technology. The biomimetic lymph node unit is made of magnetorheological elastomer, and the stiffness of the magnetorheological elastomer is adjusted in real time by an external magnetic field to simulate the tissue deformation caused by traction during surgery.

[0123] The biomimetic lymphatic network is a microchannel structure made of temperature-sensitive hydrogel. The biomimetic lymphatic network is filled with fluorescently labeled fluid that simulates lymph fluid. The permeability of the biomimetic lymphatic network is dynamically adjusted according to the operating pressure of medical staff.

[0124] Bionic vascular stents are vascular skeletons formed by 3D printing of flexible polymer materials. The bionic vascular stents and bionic lymphatic vessels are distributed in a network-like interlacing pattern.

[0125] The dynamic response module includes a mechanical sensor and an internal flow sensor integrated into the biomimetic lymph node unit. The dynamic response module is used to collect operational force and fluid dynamic data in real time.

[0126] A fluorescent labeling feedback module, comprising a fluorescent labeling fluid and a fluorescent sensor filled within a biomimetic vascular stent;

[0127] Among them, the fluorescence sensor is used to trigger a lymphatic vessel rupture alarm signal when the operating force of medical staff exceeds a preset safety threshold. The lymphatic vessel rupture alarm signal is used to control the local fluorescence intensity enhancement of the fluorescently labeled fluid.

[0128] During the initial surgical training, medical staff donned surgical equipment and positioned themselves, installing and fixing the 3D-printed bionic lymph node units and bionic lymphatic network according to the preset pelvic and abdominal anatomical locations to ensure successful initialization. After system initialization, medical staff began simulating surgical procedures by operating surgical instruments. The pulling and cutting actions of the instruments caused deformation of the bionic lymph node units, and the system collected operational force data in real time through mechanical sensors. Simultaneously, changes in operational pressure dynamically adjusted the permeability of the bionic lymphatic network, and flow rate sensors monitored the fluid flow status in real time. The dynamic response module fed back the collected force and fluid data to the system for analysis and processing in real time. If the operational force of the medical staff approached or exceeded the preset safety threshold, the system would immediately trigger an alarm signal by activating the fluorescent labeling feedback module. At this time, the lymphatic vessel rupture alarm signal would control the local fluorescence intensity of the fluorescently labeled fluid in the corresponding area to increase, allowing medical staff to visually see the fluorescent warning in the risk area and adjust the operational force in time. The system continuously monitored and fed back operational data, and medical staff continuously adjusted and optimized their operations based on real-time feedback until the simulated surgical training was completed. After the procedure, the system automatically generates a training report, which medical staff can view to summarize and analyze in order to improve their surgical skills.

[0129] In summary, the lymph node model of the gynecological pelvic and abdominal lymph node dissection prediction model provided by this invention solves the technical defects of traditional static models in simulating tissue deformation, fluid interaction, and dynamic feedback linkage during pelvic and abdominal lymph node dissection by constructing a high-fidelity surgical training model with multi-level biomimetic structures working synergistically. The biomimetic lymph node unit, based on the magnetic field response characteristics of magnetorheological elastomers, accurately reproduces the dynamic changes in tissue stiffness caused by intraoperative traction and separation operations, overcoming the limitations of traditional materials with constant stiffness. The biomimetic lymphatic network utilizes the self-regulating permeability characteristics of temperature-sensitive hydrogels, combined with a visual tracking mechanism of fluorescently labeled fluids, to achieve real-time mapping between operational pressure and lymphatic fluid seepage state, thereby overcoming the distortion problem of existing models in simulating dynamic fluid resistance. The topologically interwoven structure of the biomimetic vascular stent and lymphatic network highly replicates the real anatomical spatial relationship, and the deformation compatibility design of flexible polymer materials ensures structural stability under simultaneous deformation of multiple tissues. The dynamic response module integrates dual-modal sensing technology of mechanics and flow velocity to achieve full-dimensional data acquisition and synchronous analysis of operational force and fluid state, constructing a quantitative correlation model between intraoperative biomechanical behavior and tissue damage. The fluorescent labeling feedback module, through a threshold-triggered local fluorescence intensity enhancement mechanism, intuitively transforms the risk of lymphatic vessel rupture into a spatially visualized signal, forming a multi-sensory collaborative real-time warning system. These technologies synergistically solve the technical gaps in anatomical structure fidelity, multi-physics dynamic coupling, and real-time risk feedback in traditional training devices, providing medical staff with a highly realistic and complex surgical interactive environment. This significantly improves their ability to recognize and control the combined effects of intraoperative mechanical boundaries, fluid dynamics, and energy interactions, ultimately achieving the core goal of reducing operational error rates and avoiding clinical complications.

[0130] In one embodiment, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described training method for the gynecological pelvic and abdominal lymph node dissection prediction model.

[0131] In one embodiment, this application also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the above-described training method for the gynecological pelvic and abdominal lymph node dissection prediction model.

[0132] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of those different embodiments or examples.

[0133] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0134] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various variations or substitutions within the technical scope disclosed in this application, and these should all be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for training a predictive model for gynecological pelvic and abdominal lymph node dissection, characterized in that, Includes the following steps: S1: Dynamic modeling and processing of preoperative imaging data and intraoperative instrument operation parameters of historical patients are performed. By integrating the anatomical structure and operation parameters of historical cases, a dynamic mechanical-fluid coupling lymph node model based on biomimetic materials is constructed. The lymph node model is made of magnetorheological elastomer and thermosensitive hydrogel composite and is used to simulate the mechanical deformation and fluid flow of the lymphatic system under various clinical scenarios. S2: Perform multimodal data acquisition and processing on the lymph node model used by medical staff, record the operation force and fluid state through force feedback device and optical sensor, and extract multi-source surgical features containing mechanical features, fluid flow features and energy device thermal diffusion features. The multi-source surgical features are used to quantify the dynamic impact of surgical operation on the lymphatic system. S3: Perform multimodal fusion analysis on the multi-source surgical features, use a deep learning model to learn the correlation between historical operations and damage risks, and output the probability of lymphatic vessel rupture. The probability of lymphatic vessel rupture is used to assess the operational risks of medical staff. S4: Real-time feedback control processing is performed based on the probability of lymphatic vessel rupture. By dynamically adjusting the operating parameters and generating visual signals, training correction instructions and risk heat maps are generated. The training correction instructions are used to guide medical staff to optimize surgical skills.

2. The method according to claim 1, characterized in that, S1 includes: S11: Perform pelvic lymph node and blood vessel segmentation on the preoperative imaging data of historical patients, and generate a three-dimensional anatomical structure framework of biomimetic materials through 3D printing technology. The three-dimensional anatomical structure framework includes a magnetorheological elastomer skeleton and hydrogel-filled lymphatic microchannels. S12: Based on the intraoperative instrument operation parameters of historical patients, dynamic modeling is performed. Combined with magnetic field control and operation force changes, a dynamic stiffness function and a fluid resistance function are generated. The dynamic stiffness function is used to quantify the real-time changes in material stiffness with magnetic field and operation force, and the fluid resistance function is used to simulate the dynamic adjustment of lymph flow resistance. S13: Perform parameter fusion processing on the three-dimensional anatomical structure data, dynamic stiffness function and fluid resistance function, and map the mechanical parameters to the anatomical structure through finite element mesh to construct a lymph node model based on biomimetic materials with dynamic mechanical-fluid coupling.

3. The method according to claim 2, characterized in that, S12 includes: S121: The magnetic field strength and operating force in the intraoperative instrument operation parameters of historical patients are processed, and combined with the basic elastic modulus of the magnetorheological elastomer, a dynamic stiffness function is generated. The expression of the dynamic stiffness function is as follows: Where E(t) is the dynamic stiffness function, representing the value of the dynamic stiffness function at time t, E0 is the basic elastic modulus of the magnetorheological elastomer, k is the response coefficient of the magnetorheological elastomer, α and β are weighting factors calibrated using historical surgical data, H(τ) is the magnetic field strength, and F is the operating force. S122: The operating force and initial fluid resistance of the hydrogel microchannel in the intraoperative instrument operation parameters of the historical patients are processed, and combined with the dynamic stiffness function, a fluid resistance function is generated. The expression of the fluid resistance function is: Where R(t) is the fluid resistance function, representing the fluid resistance function value at time t, R0 is the initial fluid resistance of the hydrogel microchannel, and γ is the hydrogel permeability correction coefficient.

4. The method according to claim 2, characterized in that, S13 includes: S131: Based on the spatial parameters of the lymphatic microchannels in the three-dimensional anatomical framework, fluorescent dye is injected into the hydrogel-filled microchannels. The dye concentration and flow rate are controlled by a microfluidic pump to generate a visual fluid flow path. The visual fluid flow path is used to observe the lymph flow status in real time. S132: Based on the dynamic stiffness function and magnetic field control parameters, the magnetorheological elastomer skeleton of the three-dimensional anatomical structure frame is subjected to dynamic magnetic field loading treatment. The local magnetic field strength is adjusted in real time through an electromagnetic coil array to generate a biomimetic material hardness change synchronized with the surgical operation. S133: Based on the fluid resistance function and hydrogel permeability parameters, the fluid resistance of the lymphatic microchannels of the three-dimensional anatomical framework is dynamically adjusted. The cross-sectional area of ​​the channel is adjusted by a variable aperture valve to generate lymph flow velocity feedback that matches the operating force. S134: Based on the visualized fluid flow path, the change in hardness of the biomimetic material, and the flow velocity feedback, the three-dimensional anatomical structure framework is subjected to multi-physics field coupling verification processing. By comparing historical surgical data to calibrate the model response, a high-fidelity biomimetic material dynamic mechanical-fluid coupling lymph node model is generated.

5. The method according to claim 1, characterized in that, The multi-source surgical features include mechanical feature vectors, fluid dynamic parameters, and thermal damage distribution features, and S2 includes: S21: Based on the force feedback data of the medical staff training obtained from the lymph node model, the clamping force output by the force feedback device is sampled with high precision, and the time sequence signal is divided by an equal interval time window to generate a mechanical feature vector. The mechanical feature vector is used to quantify the time sequence change of the force of the medical staff operation. S22: Based on the lymph node model, the fluid flow path of the lymphatic microchannel is visualized to perform optical tracking of the fluorescent dye displacement. The flow velocity and flow resistance change rate are calculated by high-speed imaging and particle image velocimetry technology to generate fluid dynamic parameters. The fluid dynamic parameters are used to evaluate the integrity of the lymphatic vessels. S23: Based on the thermal response characteristics of biomimetic materials, infrared thermal imaging processing is performed on the thermal diffusion range of the electrosurgical unit used by medical personnel. Thermal damage distribution characteristics are generated by temperature gradient mapping. These thermal damage distribution characteristics are used to quantify the thermal impact of energy devices on biomimetic materials.

6. The method according to claim 1, characterized in that, S3 includes: S31: Perform time window division processing on the mechanical feature vector, fluid dynamic parameters and thermal damage distribution characteristics of the multi-source surgical features, and divide the continuous data into independent time periods by sliding window to generate a multimodal time series dataset; S32: Based on a temporal convolutional network, perform correlation processing on the multimodal temporal dataset, extract local and global features through multi-layer dilated convolutional kernels, and generate high-dimensional spatiotemporal features; S33: The temporal dependencies of the high-dimensional spatiotemporal features are modeled and processed based on a long short-term memory network. Key temporal information is filtered through a gating mechanism, and the probability of lymphatic vessel rupture is output.

7. The method according to any one of claims 1-6, characterized in that, S4 includes: S41: Based on the lymphatic vessel rupture probability and the preset clinical safety threshold, the risk of injury is dynamically assessed and processed. By comparing with the threshold in real time, a fluorescent warning signal is generated during high-risk operations. S42: Based on the multi-source surgical characteristics and the lymphatic vessel rupture probability, perform gradient optimization and boundary constraint processing on the operating force of medical staff to generate a corrected operating force; S43: Based on the fluorescent warning signal and the correction operation intensity, perform visualization rendering processing on the risk area to generate training correction instructions and risk heat map.

8. A lymph node model for a gynecological pelvic and abdominal lymph node dissection prediction model, characterized in that, include: The biomimetic lymph node unit is a replica of the anatomical structure of pelvic lymph nodes and para-aortic lymph nodes using 3D printing technology. The biomimetic lymph node unit is made of magnetorheological elastomer, and the stiffness of the magnetorheological elastomer is adjusted in real time by an external magnetic field to simulate tissue deformation caused by traction during surgery. A biomimetic lymphatic network, wherein the biomimetic lymphatic network is a microchannel structure made of temperature-sensitive hydrogel, and the biomimetic lymphatic network is filled with fluorescently labeled fluid that simulates lymph fluid. The permeability of the biomimetic lymphatic network is dynamically adjusted according to the operating pressure of medical staff. A biomimetic vascular stent, wherein the biomimetic vascular stent is a vascular skeleton formed by 3D printing of flexible polymer material, and the biomimetic vascular stent and the biomimetic lymphatic vessels are distributed in a network-like interlacing pattern. The dynamic response module includes a flow velocity sensor integrated into the mechanical sensor inside the bionic lymph node unit. The dynamic response module is used to collect operational force and fluid dynamic data in real time. A fluorescent labeling feedback module, comprising a fluorescent labeling fluid and a fluorescent sensor filled within the bionic vascular stent; The fluorescent sensor is used to trigger a lymphatic vessel rupture alarm signal when the force exerted by medical personnel exceeds a preset safety threshold. The lymphatic vessel rupture alarm signal is used to control the local fluorescence intensity enhancement of the fluorescently labeled fluid.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 7.

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

Images