Internet platform-based cardiac radiofrequency ablation surgery simulation training and evaluation system

Abstract

The present application relates to the technical field of medical virtual simulation and surgical simulation training, in particular to a heart radiofrequency ablation surgical simulation training and evaluation system based on an Internet platform, comprising: a multi-modal data management module: storing static high-resolution structure data and receiving real-time dynamic image data; a non-rigid motion mapping engine: extracting sparse motion feature vectors to drive static data to generate non-rigid deformation and generate a time-varying high-precision three-dimensional model; an interventional simulation calculation unit: solving the pose of a virtual surgical instrument and calculating the range of physical field effects; a dynamic risk assessment module: monitoring the relative position relationship between the range of physical field effects and the key protective structure, and generating an operation feedback signal; the present application solves the problem that a static model cannot simulate the influence of heart beating on the interventional path, and effectively trains the risk avoidance ability of doctors in a dynamic anatomical environment.

Description

Technical Field

[0001] This invention relates to the field of medical virtual simulation and surgical simulation training technology, specifically a cardiac radiofrequency ablation surgical simulation training and evaluation system based on an Internet platform. Background Technology

[0002] In the existing field of medical simulation training, cardiac radiofrequency ablation surgery simulation systems are widely used in the skills training of doctors. The system uses a three-dimensional anatomical model to construct a virtual cardiac environment, enabling doctors to practice virtual surgical instrument navigation and ablation operations in a virtual scene. Existing simulation systems typically reconstruct static 3D models based on preoperative medical images such as computed tomography or magnetic resonance imaging, and simulate the operation of virtual surgical instruments within these fixed geometries, or drive the overall displacement of the model through simple rigid body motion algorithms to simulate heart rhythm, thereby providing operators with basic visual references. However, relying solely on static or rigid body motion models cannot accurately reflect the complex non-rigid deformations that occur in the heart during its beating cycle. This results in the inability to update the relative positional relationship between the myocardial surface and critical protective structures such as the coronary arteries in real time. Consequently, there are deviations in the distance calculations between virtual surgical instruments and critical structures in the simulation environment. This makes it difficult for the system to accurately assess the potential surgical injury risks in a dynamic environment. These risks include two types: the first is the risk of needle insertion, which manifests as puncture damage to the coronary arteries, left ventricle, and right ventricular angle; the second is the risk during the ablation process, which manifests as damage to the cardiac conduction system, including the left bundle branch, right bundle branch, and atrioventricular node. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention provides a cardiac radiofrequency ablation surgery simulation training and evaluation system based on an internet platform. Specifically, the technical solution of this invention includes: The multimodal data management module is configured to store static high-resolution structural data describing the anatomical topology of the target deformed object in the cloud via the Internet platform, and to receive real-time dynamic image data reflecting the current motion state of the target deformed object in real time via the Internet platform. The non-rigid motion mapping engine is configured to extract sparse motion feature vectors based on the real-time dynamic image data, and use the sparse motion feature vectors to drive the static high-resolution structural data to generate non-rigid deformation, so as to generate a time-varying high-precision three-dimensional model that is spatiotemporally synchronized with the real-time dynamic image data. The simulation calculation unit is configured to calculate the spatial pose of the virtual surgical instrument in real time within the coordinate system of the time-varying high-precision three-dimensional model, and to calculate the range of the physical field effect of the virtual surgical instrument acting on the time-varying high-precision three-dimensional model. The dynamic risk assessment module is configured to monitor the relative positional relationship between the physical field effect range and the key protective structure in the time-varying high-precision three-dimensional model, and generate an operation feedback signal based on the relative positional relationship. The static high-resolution structural data is derived from a three-dimensional mesh model constructed by computed tomography (CT) or magnetic resonance MRI; the real-time dynamic imaging data is derived from intraoperative ultrasound (US) sequences or simulated hydrodynamic signals.

[0004] Preferably, the non-rigid motion mapping engine includes: The feature tracking unit is configured to perform optical flow field analysis or speckle tracking on consecutive frames of the real-time dynamic image data, and calculate the sparse motion feature vector that reflects the tissue motion trend. The elastic registration unit is configured to define the grid nodes in the static high-resolution structural data as subordinate nodes, define the endpoints of the sparse motion feature vector as driving control points, and calculate the deformation trajectory of the subordinate nodes as the driving control points are displaced by the interpolation algorithm. The calculation of the deformation trajectory follows volume preservation constraints to ensure that the target deformable object does not collapse during simulated contraction or relaxation.

[0005] Preferably, the intervention simulation calculation unit includes: The thermodynamic evolution submodule is configured to calculate the diffusion boundary of the physical field in the dynamic deformable mesh based on the output power parameters and contact time of the virtual surgical instrument, combined with the local tissue thermal conductivity coefficient of the time-varying high-precision three-dimensional model. The boundary dynamic update submodule is configured to respond to the mesh deformation of the time-varying high-precision 3D model and remap the spatial distribution of the physical field in real time, ensuring that the range of the physical field effect shifts synchronously with the movement of the tissue.

[0006] Preferably, the dynamic risk assessment module includes: The topological distance monitoring unit is configured to calculate the minimum Euclidean distance between the boundary of the physical field effect range and the surface mesh of the critical protective structure within each simulation time step. The safety threshold determination unit is configured to: output a safety status signal when the minimum Euclidean distance is greater than a preset safety buffer threshold; output a warning signal when the minimum Euclidean distance is less than or equal to the safety buffer threshold and greater than zero; and output a damage determination signal when the minimum Euclidean distance is equal to zero or mesh interpenetration is detected.

[0007] Preferably, the system further includes: A temporal consistency filter is configured to smooth the sparse motion feature vector in the time dimension to filter out deformation jitter caused by the signal-to-noise ratio fluctuation of the real-time dynamic image data. The key protective structure is preset as a coronary artery vascular model, and the non-rigid motion mapping engine is configured to apply topological constraints in deformation calculations to maintain the attachment topology of the coronary artery vascular model relative to the myocardial surface, thereby limiting structural separation or deep penetration under extreme deformation conditions.

[0008] Preferably, the system further includes: The force-tactile rendering interface is configured to receive contact depth data between the virtual surgical instrument and the time-varying high-precision three-dimensional model, and generate a damping control signal for the driving force feedback device based on the contact depth data and the dynamic elastic modulus of the target deformable object. The dynamic elastic modulus is configured to change in real time with the contraction state of the target deformable object, and its value is dynamically mapped between a first stiffness value corresponding to the contraction period and a second stiffness value corresponding to the diastolic period according to the cardiac cycle phase.

[0009] Preferably, the system further includes: The client-side augmented reality visualization terminal is configured to receive data from the Internet platform, use the real-time dynamic image data as the background layer, and use the time-varying high-precision 3D model, the virtual surgical instruments and the physical field effect range as the enhancement layer for fusion rendering. The client-side augmented reality visualization terminal is also configured to: render a yellow warning halo around the critical protective structure in response to the warning signal; and render a red damage marker in the intersection area of ​​the physical field effect range and the critical protective structure in response to the damage determination signal.

[0010] Preferably, the system further includes: The cloud-based training and evaluation server is configured to record the historical sequence of the operation feedback signals via an internet platform, and generate and send an operation skill score report to the platform user based on the number of times the damage judgment signal is triggered and its duration. The operational skill scoring report includes a multimodal motion feature coupling index, which is configured to quantitatively assess the user's ability to predict the motion patterns of anatomical structures in a dynamic environment by calculating the temporal correlation or phase synchronization deviation between the motion velocity vector sequence of the virtual surgical instrument and the motion velocity vector sequence of the corresponding contact area on the surface of the target deformable object.

[0011] Compared with the prior art, the present invention has the following beneficial effects: 1. This system achieves spatiotemporal synchronization of high-resolution static anatomical structures and real-time dynamic ultrasound images by combining a non-rigid motion mapping engine with volume-preserving constraints. It achieves the effect of generating time-varying high-precision three-dimensional models without volume collapse during the simulation of cardiac contraction or relaxation. Compared with existing simulation methods that rely solely on static models or simple rigid body motion, this invention can use sparse motion feature vectors to drive the mesh to generate realistic non-rigid deformation, solving the problem that traditional models cannot accurately reflect the impact of heartbeat on the intervention path, allowing doctors to intuitively observe the real state of coronary arteries as they undulate with the heart wall. 2. This system, through the collaborative work of the simulation calculation unit and the dynamic risk assessment module, achieves real-time calculation and collision monitoring of the physical field effect range within the dynamic deformable mesh coordinate system; it captures instantaneous risks that only occur during cardiac contraction and generates precise operational feedback signals; compared with the shortcomings of existing technologies, such as misjudgment of ablation line breakage due to model motion errors or neglect of dynamic distance changes, this invention ensures that the physical field effect range shifts synchronously with tissue movement, effectively training doctors' risk avoidance capabilities in dynamic anatomical environments; 3. This system introduces a dynamic elastic modulus that changes with the cardiac cycle through a force-tactile rendering interface, realizing pulsatile force feedback that conforms to physiological characteristics; it achieves the effect of mapping different stiffness values ​​in real time according to the heart's contraction and relaxation states, and generating vector damping control signals through a viscoelastic model; compared with existing systems that lack dynamic tactile perception, this invention can simulate the compression and rebound of instruments in the normal direction, allowing the operator to clearly perceive the real-time changes in myocardial stiffness, and solving the problem of mismatch between tactile feedback and visual dynamics; 4. This system combines a client-side augmented reality visualization terminal with a cloud-based training and evaluation server to achieve the visualization of implicit physical field risks and the quantitative assessment of operational skills; it achieves the effect of rendering warning halos around key structures and quantifying hand-eye coordination ability through multimodal motion feature coupling degree indicators; compared with existing technologies that lack immediate visual penalties and dynamic follow-up evaluation indicators, this invention helps trainees establish an intuitive understanding of dangerous areas through intuitive color marking and phase synchronization deviation analysis, and improves their ability to predict the movement patterns of anatomical structures in dynamic environments. Attached Figure Description

[0012] The present invention will be further explained below with reference to the accompanying drawings and embodiments: Figure 1 This is a structural diagram of the system of the present invention. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0014] Example 1: Please see Figure 1 A simulation training and evaluation system for cardiac radiofrequency ablation surgery based on an internet platform, the system includes: The multimodal data management module is configured to store static high-resolution structural data describing the anatomical topology of the target deformed object in the cloud via the Internet platform, and to receive real-time dynamic image data reflecting the current motion state of the target deformed object in real time via the Internet platform. The non-rigid motion mapping engine is configured to extract sparse motion feature vectors based on real-time dynamic image data, and use the sparse motion feature vectors to drive static high-resolution structural data to generate non-rigid deformation, so as to generate a time-varying high-precision 3D model that is spatiotemporally synchronized with real-time dynamic image data. The simulation calculation unit is configured to solve the spatial pose of the virtual surgical instruments in real time in the coordinate system of the time-varying high-precision three-dimensional model, and to calculate the range of physical field effects of the virtual surgical instruments on the time-varying high-precision three-dimensional model. The dynamic risk assessment module is configured to monitor the relative positional relationship between the physical field effect range and the key protective structures in the time-varying high-precision three-dimensional model, and generate operation feedback signals based on the relative positional relationship. Among them, static high-resolution structural data are derived from three-dimensional mesh models constructed by computed tomography (CT) or magnetic resonance MRI; real-time dynamic imaging data are derived from intraoperative ultrasound (US) sequences or simulated hydrodynamic signals.

[0015] This embodiment details the core architecture and data flow logic of the system, aiming to construct a dynamic simulation environment with both high spatial and temporal resolution. The multimodal data management module executes the initialization loading program, retrieves the patient's preoperative CT or MRI data from the hospital PACS system, and reconstructs a high-precision static tetrahedral mesh containing the atria, ventricles, and coronary arteries through image segmentation algorithms. This data defines the baseline topology of the target deformable object at time zero. The system connects to the video stream interface of the ultrasound simulator or intraoperative ultrasound equipment to capture two-dimensional ultrasound image sequences with a frame rate greater than 30fps in real time. Although this data has a low spatial signal-to-noise ratio, it carries information about ventricular wall motion reflecting the cardiac cycle. The non-rigid motion mapping engine is activated, using motion information from ultrasound images as the driving source. It constructs a deformation field by extracting sparse feature points, forcing a high-precision static mesh model to deform and generating a time-varying high-precision 3D model. Based on this, the interventional simulation calculation unit calculates the tip pose of the virtual surgical instruments in real time within the continuously deforming mesh coordinate system, and calculates the physical field effect range of the ablation energy acting on the tissue, i.e., the physical field effect range. The dynamic risk assessment module monitors the Euclidean distance between this dynamic physical field effect range and key protective structures such as the coronary arteries in real time. Once a potential collision risk is detected, a feedback signal is immediately generated. This embodiment successfully solves the technical problem that traditional static models cannot simulate the impact of heartbeat on the intervention path through the above-mentioned non-rigid motion mapping mechanism. In the scenario of cardiac radiofrequency ablation surgery simulation, the system enables trainees to intuitively observe the real state of the coronary artery undulating on the heart wall with the heartbeat, and predict whether the physical field effect range of the virtual surgical instrument tip will accidentally touch the coronary artery due to heart contraction, thereby effectively training their risk avoidance ability in a dynamic anatomical environment.

[0016] Example 2: Non-rigid motion mapping engines include: The feature tracking unit is configured to perform optical flow field analysis or speckle tracking on consecutive frames of real-time dynamic image data, and calculate sparse motion feature vectors that reflect the movement trend of tissues. The elastic registration unit is configured to define the grid nodes in the static high-resolution structural data as subordinate nodes, define the endpoints of the sparse motion feature vectors as driving control points, and calculate the deformation trajectory of the subordinate nodes as the displacement of the driving control points through an interpolation algorithm. The calculation of the deformation trajectory follows volume preservation constraints to ensure that the target deformed object does not collapse during the simulated contraction or relaxation process.

[0017] This embodiment provides a detailed description of the specific implementation of the non-rigid motion mapping engine. The feature tracking unit executes adaptive processing logic based on the modal type of the input data: if the input is real-time ultrasound image, the unit preprocesses the continuous frame images, uses optical flow to perform pixel-by-pixel analysis of the image sequence, assumes that the brightness is constant for a very short time, and calculates the velocity vector of sparse feature points by minimizing the optical flow constraint equation; if the input is a simulated fluid dynamics signal, i.e., CFD velocity field data, the unit directly performs spatial sampling on the flow field mesh and extracts the velocity vector at the corresponding position as the sparse motion feature vector; these feature points constitute the skeleton that drives the mesh deformation. The elastic registration element introduces an interpolation algorithm based on radial basis functions to densify the sparse motion information and transmit it to the entire 3D mesh. To calculate the displacement of any subordinate node in the static mesh at the current moment, the system uses an elastic mapping function for solution, as shown in the following formula:

[0018] in, The initial coordinates of subordinate nodes in a static mesh are derived from static high-resolution structure data; The total number of driving control points is derived from the number of feature points extracted by the feature tracking unit. :No. The initial coordinate positions of each drive control point are derived from the initial registration results; :No. The weight coefficient vector of each control point has a dimension of . ; Radial basis kernel function, specifically in Gaussian radial basis form:

[0019] in, It is a natural constant. The kernel width parameter is preset; in this embodiment, in order to balance the locality of deformation and smoothness, The value is set to 2.5 times the average side length of all tetrahedral elements in the static high-resolution structural data, specifically: ; The affine transformation matrix, physically representing the global rotation and scaling components, is initialized at the initial iteration. identity matrix ; : Global translation vector, physically representing the displacement components of a rigid body, initialized to a zero vector at the initial moment of the iteration; : No. The weight coefficient vector of each control point has the physical meaning of the non-rigid displacement amplitude parameter of the control point in the local space. Its value is determined by iterative solution of the subsequent energy minimization equation. During this process, the system enforces volume preservation constraints and constructs a composite energy objective function. To ensure that the optimization algorithm can solve for the core deformation parameters and global parameters and The system has constructed a complete gradient calculation framework; in addition to the aforementioned... and Apart from the partial derivatives, for the weight coefficients The gradient calculation formula is as follows:

[0020] in, For the first The total number of tetrahedral mesh elements affected by each control point; The first one calculated based on the current deformation parameters Real-time location coordinates of the affected grid nodes; The coordinates of the target position of the node are calculated based on the sparse motion feature vector output by the feature tracking unit. For the first The volume penalty coefficient for a local area of ​​a control point is defined by the following formula:

[0021] in, This is the global adjustment weight, with a value of 1.0. The average volume of the entire grid. For the first The average grid volume within the influence domain of each control point is used to achieve adaptive stiffness constraints. For the corresponding to the first The Jacobian determinant of the grid cells, here and Refers to the same traversal index relationship; For gradient operators; in, It is explicitly defined as the sum of squares of the penalty Jacobian determinant deviating from the unit value. ; For the first The set of tetrahedral mesh elements affected by each control point, i.e., the summation index in the aforementioned gradient calculation formula. The range of the collection being traversed; This is the determinant of the deformation gradient tensor of this element; based on the complete gradient information described above. , , The system uses L-BFGS or gradient descent to synchronously update the weight coefficients, affine matrix, and translation vector until... convergence; In particular, to ensure the solvability of the volume preservation constraint at the code level, the system explicitly calculates the Jacobian matrix based on the deformation mapping function. The analytical solution is obtained, and the Jacobian determinant of each tetrahedral element is assembled. This provides the gradient descent optimizer with accurate feedback on the rate of volume change.

[0022] Example 3: The simulation calculation unit includes: The thermodynamic evolution submodule is configured to calculate the diffusion boundary of the physical field in the dynamic deformable mesh based on the output power parameters and contact time of the virtual surgical instrument, combined with the local tissue thermal conductivity coefficient of the time-varying high-precision three-dimensional model. The boundary dynamic update submodule is configured to respond to the mesh deformation of the time-varying high-precision 3D model, and remap the spatial distribution of the physical field in real time to ensure that the range of physical field effects shifts synchronously with the movement of the tissue.

[0023] This embodiment specifies the thermodynamic calculations in the interventional simulation calculation unit; the thermodynamic evolution submodule calculates the damage radius in a local coordinate system, i.e., assuming the tissue is relatively stationary, based on a simplified form of the Pennes biothermal equation; this calculation process involves the following formulas:

[0024] in, Damage radius, in physical terms, is the spherical range within which the cell necrosis temperature is reached; Natural constant; The output power parameters of the virtual surgical instruments are derived from the user's settings on the simulator console; here, we use... To distinguish it from position coordinate symbols; Effective ablation time, which is the duration from the start of energy output to the current calculation moment; Tissue density, in this embodiment, is preset to standard myocardial density. ; Specific heat capacity, which is preset in this embodiment as follows: ; Local tissue thermal conductivity coefficient, which physically represents the rate at which biological tissue conducts heat; its value is obtained from the system's pre-set database of anatomical tissue thermal properties, based on the anatomical labels of the current contact point of the virtual surgical instrument in the three-dimensional model, such as the myocardium, epicardial fat, or fibrotic area, through real-time indexing. Specifically, to ensure the executability of the computation, the pre-built database stores... The values ​​are set as follows: For normal myocardial tissue, corresponding to the myocardial layer label, The value is set to ;Targeting epicardial adipose tissue, The value is set to For fibrotic areas, i.e. scar tissue, The value is set to If the contact point is identified as a high-flow-scouring area, then the equivalent thermal conductivity is used. To simulate the cooling effect caused by thermal convection; The temperature rise threshold that leads to cell necrosis is set in this embodiment as follows: Corresponding target temperature Subtract body temperature ; : Empirical correction factor; where, The thermal diffusivity correction factor is configured as a dimensionless value, and its value depends on the local tissue thermal conductivity. To ensure dimensional balance in the formula, this coefficient is specifically defined as follows:

[0025] in, Normalized thermal conductivity is , This is a dimensionless reference constant that depends on the tip geometry and perfusion pattern of the virtual surgical instrument; for a standard 4mm tip perfusion surgical instrument, this value is calibrated to 0.68; to meet the requirement of sufficient disclosure of parameter sources, the specific calibration experimental procedure of this reference constant is described here: a standard agar phantom was selected, and its thermophysical parameter was measured as thermal conductivity. ,density Specific heat capacity Insert the surgical instrument into the phantom and set the output power. Duration of ablation At this point, it is considered to have reached a quasi-steady state, and the actual damage radius is measured. Substituting the known experimental data above into the formula, we can deduce the result in reverse:

[0026] Thus determine The value; The fitting data is derived from in vitro animal experimental data, and its value is specifically determined in this embodiment as follows: The method for obtaining this value is as follows: under the experimental conditions described above, different ablation time points were recorded. Corresponding damage radius Data sequence Using nonlinear least squares method to analyze exponentially growing terms Regression analysis was performed to calculate the best-fit value. goodness of fit The dimension of this parameter is the reciprocal of time, ensuring that the exponent term is a dimensionless value. The boundary dynamic update submodule performs coordinate mapping operations. In response to the displacement of mesh nodes driven by the non-rigid motion mapping engine, this module binds the physics center to the centroid of the mesh element corresponding to the contact point and uses the aforementioned elastic mapping function to remap the vertex coordinates of the physics boundary in real time. Specifically, it reconstructs the coordinates corresponding to... The isothermal surface mesh is used to update the position of the physical field effect range in the world coordinate system; This embodiment achieves a simulation effect of stillness within motion, meaning that although the heart is beating violently, the damaged area formed by ablation can adhere to the myocardium and move with it like real scar tissue, rather than being suspended in a fixed spatial position. This ensures the accuracy of the assessment of the cumulative damage range, enabling doctors to accurately determine whether continuous transmural damage has been achieved between points after multiple ablations, avoiding misjudgments of ablation line breakage caused by model motion errors.

[0027] Example 4: The dynamic risk assessment module includes: The topological distance monitoring unit is configured to calculate the minimum Euclidean distance between the boundary of the physical field effect range and the surface mesh of the critical protective structure within each simulation time step. The safety threshold determination unit is configured to: output a safety status signal when the minimum Euclidean distance is greater than the preset safety buffer threshold; output a warning signal when the minimum Euclidean distance is less than or equal to the safety buffer threshold and greater than zero; and output a damage determination signal when the minimum Euclidean distance is equal to zero or mesh interpenetration is detected.

[0028] This embodiment details the decision logic of the dynamic risk assessment module; the topology distance monitoring unit uses the bounding box hierarchy tree (BVH) to coarsely screen the geometry in the scene at each simulation time step, for example, 0.01 seconds, and calculates the precise minimum Euclidean distance for overlapping areas; the distance calculation formula is as follows:

[0029] in, The minimum Euclidean distance, in physical terms, is the shortest path from the boundary of the physical field to the surface of the blood vessel. The surface mesh point set within the physical field effect range is derived from the intervention simulation calculation unit; The key protective structures, namely the surface grid point set of the coronary arteries, are derived from the multimodal data management module. To enable multiple risk assessments, the key protective structures also include the left ventricle, the right ventricular angle, and the cardiac conduction system. The safety threshold determination unit executes hierarchical logic based on the calculation results. In response to the minimum Euclidean distance being greater than a preset safety buffer threshold (set to 5mm in this embodiment), which is derived by superimposing the physical diameter of the virtual surgical instrument (1.7mm) and the statistical standard deviation of the physical field edge (2.5mm), or by directly referencing the sum of the 2mm energy field diffusion error limit and the 3mm instrument positioning uncertainty specified in clinical ablation guidelines, the system outputs a safety status signal. In response to the distance being less than or equal to the safety buffer threshold and greater than zero, the system determines a high risk and outputs a warning signal. In response to the distance being equal to zero or grid interpenetration being detected (i.e., the distance value is negative or the normal vector is reversed), the system determines that the ablation energy has invaded the coronary artery and outputs a damage determination signal. This embodiment calculates the minimum Euclidean distance in real time on a dynamic deformation model, enabling the system to capture instantaneous risks that only occur during cardiac systole. For example, a coronary artery may be far from the ablation point during diastole, but it may momentarily approach the tip of a virtual surgical instrument during systole. This is a clinical scenario that traditional static models cannot train on. This mechanism forces trained physicians to consider the impact of the cardiac cycle on anatomical distance, thereby cultivating the habit of performing safe operations in a dynamic environment.

[0030] Example 5: The system also includes: The temporal consistency filter is configured to smooth the sparse motion feature vector in the time dimension to filter out deformation jitter caused by the signal-to-noise ratio fluctuation of real-time dynamic image data. The key protective structure is preset as a coronary artery model, and the non-rigid motion mapping engine is configured to apply topological constraints in deformation calculations to maintain the attachment topology of the coronary artery model relative to the myocardial surface, thereby limiting structural separation or deep penetration under extreme deformation conditions.

[0031] This embodiment introduces a temporal consistency filter and specific topological constraints; the temporal consistency filter applies an exponentially weighted moving average to the original sparse motion feature vector extracted from the ultrasound image to calculate the corrected feature vector, as shown in the following formula:

[0032] in, This is the smoothed feature vector at the current time step; This is the original sparse motion feature vector output by the feature tracking unit at the current moment, i.e., the unfiltered instantaneous observation; The smoothing factor is set to 0.3; this value is chosen to balance system response delay and noise immunity. Experimental results show that when... At that time, the system's time lag is controlled within Within approximately one frame, it can effectively filter out non-physical high-frequency jitter caused by ultrasonic spot noise; The non-rigid motion mapping engine applies topological constraints in deformation calculation. To address the problem that traditional ray projection methods are prone to failure under large non-rigid deformations, i.e., normal rays may not hit severely distorted surfaces, this system uses a centroid coordinate binding mechanism to implement topological constraints. The specific steps are as follows: Initialize binding: In the static mesh at time zero, for each coronary artery node... Search for its nearest neighbor triangular facet on the myocardial surface mesh. And calculate the coordinates of the node relative to the centroid of the triangle. ,satisfy ,in, For the normal vector of the surface, This is the initial normal height offset; tuple Store in the topology constraint cache; Dynamic constraint update: After each frame of deformation calculation is completed, the system does not directly apply optical flow to the blood vessel nodes, but instead updates the optical flow based on the new coordinates of the vertices of the myocardial triangle at the current moment. and the updated normal vector Through formula Forcefully reconstruct the spatial location of vascular nodes; This method mathematically ensures that no matter how the myocardium is stretched or twisted, the coronary arteries are always anchored on the corresponding myocardial texture coordinates, thereby physically eliminating the possibility of structural separation or deep penetration and ensuring anatomical topological consistency under extreme deformation conditions.

[0033] Example 6: The system also includes: The force-tactile rendering interface is configured to receive contact depth data between virtual surgical instruments and time-varying high-precision 3D models, and generate damping control signals for driving force feedback devices based on the contact depth data and the dynamic elastic modulus of the target deformable object. The dynamic elastic modulus is configured to change in real time with the contraction state of the target deformable object, and its value is dynamically mapped between a first stiffness value corresponding to the contraction period and a second stiffness value corresponding to the diastolic period according to the phase of the cardiac cycle.

[0034] This embodiment adds a force-haptic rendering interface to provide tactile feedback that conforms to physiological characteristics; in order to obtain accurate cardiac cycle phase, the system executes a real-time phase-locked algorithm based on R-wave detection to calculate the phase. ; The force-haptic rendering interface calculates the real-time Young's modulus of the target deformed object. ;Regarding the time-varying activation function defined in the specification To avoid black-box description, this embodiment concretizes it as a sinusoidal squared pulse function corrected based on the statistical law of cardiac cycle:

[0035] in, The duration of the contraction phase is calculated using the following formula: Second, The RR interval of the current cardiac cycle; in order to generate a damping control signal in vector form and solve the dimension matching problem of scalar and vector operations, the system introduces the Kelvin-Woyt viscoelastic model and explicitly defines the contact normal vector; The specific calculation logic is as follows: The system acquires the unit normal vector of the contact point between the tip of the virtual surgical instrument and the surface of the heart in real time. and the velocity vector of the instrument tip ; Dynamic elastic modulus Convert to equivalent contact stiffness ,in, Assuming a contact characteristic length, the diameter of the virtual surgical instrument is taken as 2.5 mm, and the damping coefficient is calculated. ; Here Defined as the equivalent inertial mass of the tip of a virtual surgical instrument, used to simulate the inertial drag of the virtual surgical instrument tip moving in a fluid environment, with a preset value of [value missing]. ; This is a dimensionless damping ratio, preset to 0.8; this value corresponds to an underdamped oscillating system. The design aims to simulate the slight rebound effect of myocardial tissue as a viscoelastic body when subjected to rapid impact, making the tactile feedback more biologically realistic. Force feedback output vector The calculation formula is:

[0036] in, This represents the dot product of the virtual surgical instrument tip velocity vector and the surface normal unit vector. This term extracts the instrument's compressive velocity component in the normal direction, ensuring that the damping force is generated only in the squeezing or rebound direction and does not hinder the tangential sliding operation, in order to simulate the low-friction characteristics of the wetted myocardial surface. Let be the penetration depth scalar along the normal direction; the force vector generated by this formula The servo controller, which sends the force feedback to the handle, allows the operator to clearly perceive the pulsatile reaction force that changes with the heartbeat rhythm.

[0037] Example 7: The system also includes: The client-side augmented reality visualization terminal is configured to receive data from the Internet platform, use the real-time dynamic image data as the background layer, and use the time-varying high-precision 3D model, the virtual surgical instruments and the physical field effect range as the enhancement layer for fusion rendering. The client-side augmented reality visualization terminal is also configured to: render a yellow warning halo around the critical protective structure in response to an early warning signal; and render a red damage marker in the intersection area of ​​the physical field effect range and the critical protective structure in response to a damage determination signal.

[0038] This embodiment describes the rendering logic of the client-side augmented reality visualization terminal. The visualization terminal executes a layered rendering process, placing the grayscale real-time ultrasound image data in the bottom background and the semi-transparent time-varying high-precision 3D model and virtual surgical instruments in the upper enhancement layer. Based on this, the system performs conditional rendering according to the output signal of the dynamic risk assessment module. In response to receiving a warning signal, the system generates a particle system around the tubular mesh of the coronary artery. The particle color is set to yellow, and the particle emission rate is inversely proportional to the distance, forming a warning halo. In response to receiving a damage determination signal, the system uses real-time fragment shader technology or depth buffer contrast technology to calculate the visual intersection area between the physical field effect range and the key protective structure to avoid high-latency geometric Boolean operations on the dynamic mesh, and renders the pixels in this area in real time as a bright red material with a pulsating glow effect. This embodiment visualizes the invisible physical field risks and anatomical location relationships through AR overlay and dynamic color marking; in particular, the red markings in the intersection areas intuitively show where and how much has been ablated, providing trainees with immediate visual punishment feedback; this visualized risk warning mechanism helps doctors establish an intuitive understanding of dangerous areas in the early stages of training, shortening the conversion cycle from theoretical knowledge to muscle memory.

[0039] Example 8: The system also includes: The cloud-based training and evaluation server is configured to record the historical sequence of the operation feedback signals via an internet platform, and generate and send an operation skill score report to the platform user based on the number of times the damage judgment signal is triggered and its duration. The operational skills assessment report includes a multimodal motion feature coupling index, which is configured to quantitatively assess the user's ability to predict the motion patterns of anatomical structures in a dynamic environment by calculating the temporal correlation or phase synchronization deviation between the motion velocity vector sequence of the virtual surgical instrument and the motion velocity vector sequence of the corresponding contact area on the surface of the target deformable object.

[0040] This embodiment details the calculation logic of the training and evaluation server and its core metrics; the system records the instrument and tissue movement trajectories throughout the surgical simulation; after the simulation ends, the server calculates the multimodal motion feature coupling degree index; to strictly conform to the definition of motion velocity vector sequence in this embodiment and accurately reflect the directional consistency of the operation, i.e., to distinguish between following in the same direction and opposing in the opposite direction, this index uses the vector cosine similarity integral formula instead of the traditional scalar correlation calculation:

[0041] Furthermore, the system quantifies the phase synchronization deviation in the embodiment through peak hysteresis analysis of the cross-correlation function:

[0042] in, The operator represents the independent variable that maximizes the cross-correlation function within the parentheses. The value; to prevent mismatches across cardiac cycles, the time delay parameter The search optimization range is limited to Within the range; in, Coupling degree index, physically meaning operational follow-up score; Phase synchronization deviation, in physical terms, is the operator's reaction delay phase angle; : The total number of sampling points involved in the calculation, in this embodiment, refers to the total number of frames in a complete simulation training task or a sliding time window of a set duration; The tip of the virtual surgical instrument is at the first The three-dimensional velocity vectors of each sampling point constitute a velocity vector sequence. ; The contact point on the surface of the target deformed object is at the 1st The three-dimensional velocity vectors of each sampling point constitute a velocity vector sequence. ; Numerical stability constant, for example This is derived from system presets and is used to prevent calculation errors caused by the operator remaining absolutely still or the heart being in a simulated state of cardiac arrest, i.e., the modulus being zero, resulting in a zero denominator. Finally, the system integrates this indicator into the operational skill scoring report and provides feedback to the user. The coupling degree indicator proposed in this embodiment quantifies the hand-eye coordination and dynamic follow-up ability, which are extremely difficult to describe in advanced surgical skills. The closer the value of this indicator is to 1, the more synchronized the operator's instrument movement is with the movement of the heart wall, that is, it can be stably attached to the surface of the beating heart. The lower the value, the more the operator is resisting the movement of the heart.

[0043] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A cardiac radiofrequency ablation surgery simulation training and evaluation system based on an internet platform, characterized in that, The system includes: The multimodal data management module is configured to store static high-resolution structural data describing the anatomical topology of the target deformed object in the cloud via the Internet platform, and to receive real-time dynamic image data reflecting the current motion state of the target deformed object in real time via the Internet platform. The non-rigid motion mapping engine is configured to extract sparse motion feature vectors based on the real-time dynamic image data, and use the sparse motion feature vectors to drive the static high-resolution structural data to generate non-rigid deformation, so as to generate a time-varying high-precision three-dimensional model that is spatiotemporally synchronized with the real-time dynamic image data. The simulation calculation unit is configured to calculate the spatial pose of the virtual surgical instrument in real time within the coordinate system of the time-varying high-precision three-dimensional model, and to calculate the range of the physical field effect of the virtual surgical instrument acting on the time-varying high-precision three-dimensional model. The dynamic risk assessment module is configured to monitor the relative positional relationship between the physical field effect range and the key protective structure in the time-varying high-precision three-dimensional model, and generate an operation feedback signal based on the relative positional relationship. The static high-resolution structural data is derived from a three-dimensional mesh model constructed by computed tomography (CT) or magnetic resonance MRI; the real-time dynamic imaging data is derived from intraoperative ultrasound (US) sequences or simulated hydrodynamic signals.

2. The cardiac radiofrequency ablation surgery simulation training and evaluation system based on an internet platform according to claim 1, characterized in that, The non-rigid motion mapping engine includes: The feature tracking unit is configured to perform optical flow field analysis or speckle tracking on consecutive frames of the real-time dynamic image data, and calculate the sparse motion feature vector that reflects the tissue motion trend. The elastic registration unit is configured to define the grid nodes in the static high-resolution structural data as subordinate nodes, define the endpoints of the sparse motion feature vector as driving control points, and calculate the deformation trajectory of the subordinate nodes as the driving control points are displaced by the interpolation algorithm. The calculation of the deformation trajectory follows volume preservation constraints to ensure that the target deformable object does not collapse during simulated contraction or relaxation.

3. The cardiac radiofrequency ablation surgery simulation training and evaluation system based on an internet platform according to claim 2, characterized in that, The intervention simulation calculation unit includes: The thermodynamic evolution submodule is configured to calculate the diffusion boundary of the physical field in the dynamic deformable mesh based on the output power parameters and contact time of the virtual surgical instrument, combined with the local tissue thermal conductivity coefficient of the time-varying high-precision three-dimensional model. The boundary dynamic update submodule is configured to respond to the mesh deformation of the time-varying high-precision 3D model and remap the spatial distribution of the physical field in real time, ensuring that the range of the physical field effect shifts synchronously with the movement of the tissue.

4. The cardiac radiofrequency ablation surgery simulation training and evaluation system based on an internet platform according to claim 3, characterized in that, The dynamic risk assessment module includes: The topological distance monitoring unit is configured to calculate the minimum Euclidean distance between the boundary of the physical field effect range and the surface mesh of the critical protective structure within each simulation time step. The safety threshold determination unit is configured to: output a safety status signal when the minimum Euclidean distance is greater than a preset safety buffer threshold; output a warning signal when the minimum Euclidean distance is less than or equal to the safety buffer threshold and greater than zero; and output a damage determination signal when the minimum Euclidean distance is equal to zero or mesh interpenetration is detected.

5. The cardiac radiofrequency ablation surgery simulation training and evaluation system based on an internet platform according to claim 4, characterized in that, The system also includes: A temporal consistency filter is configured to smooth the sparse motion feature vector in the time dimension to filter out deformation jitter caused by the signal-to-noise ratio fluctuation of the real-time dynamic image data. The key protective structure is preset as a coronary artery vascular model, and the non-rigid motion mapping engine is configured to apply topological constraints in deformation calculations to maintain the attachment topology of the coronary artery vascular model relative to the myocardial surface, thereby limiting structural separation or deep penetration under extreme deformation conditions.

6. The cardiac radiofrequency ablation surgery simulation training and evaluation system based on an internet platform according to any one of claims 1-5, characterized in that, The system also includes: The force-tactile rendering interface is configured to receive contact depth data between the virtual surgical instrument and the time-varying high-precision three-dimensional model, and generate a damping control signal for the driving force feedback device based on the contact depth data and the dynamic elastic modulus of the target deformable object. The dynamic elastic modulus is configured to change in real time with the contraction state of the target deformable object, and its value is dynamically mapped between a first stiffness value corresponding to the contraction period and a second stiffness value corresponding to the diastolic period according to the cardiac cycle phase.

7. The cardiac radiofrequency ablation surgery simulation training and evaluation system based on an internet platform according to claim 6, characterized in that, The system also includes: The client-side augmented reality visualization terminal is configured to receive data from the Internet platform, use the real-time dynamic image data as the background layer, and use the time-varying high-precision 3D model, the virtual surgical instruments and the physical field effect range as the enhancement layer for fusion rendering. The client-side augmented reality visualization terminal is also configured to: render a yellow warning halo around the critical protective structure in response to the warning signal; and render a red damage marker in the intersection area of ​​the physical field effect range and the critical protective structure in response to the damage determination signal.

8. The cardiac radiofrequency ablation surgery simulation training and evaluation system based on an internet platform according to claim 7, characterized in that, The system also includes: The cloud-based training and evaluation server is configured to record the historical sequence of the operation feedback signals via an internet platform, and generate and send an operation skill score report to the platform user based on the number of times the damage judgment signal is triggered and its duration. The operational skill scoring report includes a multimodal motion feature coupling index, which is configured to quantitatively assess the user's ability to predict the motion patterns of anatomical structures in a dynamic environment by calculating the temporal correlation or phase synchronization deviation between the motion velocity vector sequence of the virtual surgical instrument and the motion velocity vector sequence of the corresponding contact area on the surface of the target deformable object.

Images