First-aid effect analysis method and system based on simulated person
Through medical imaging processing and three-dimensional reconstruction technology, combined with collision detection and dynamic methods, the quality of first aid operation is evaluated and feedback, and the problems of high cost and poor interactivity of traditional first aid training are solved, and efficient and intelligent first aid training evaluation and guidance are achieved.
Patent Information
- Application Number
- CN202411103100.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-13
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2044-08-13
AI Technical Summary
Traditional first aid training relies on theoretical learning and simulated training, which has problems such as high training costs, poor interaction and difficulty in objective evaluation. Simulators lack intelligent interaction and feedback capabilities, and cannot effectively evaluate and guide students' performance.
The tomography data is collected by medical imaging equipment, and a triangular mesh model of soft tissue is generated through medical image processing and three-dimensional reconstruction technology. Combined with collision detection algorithms and position-based dynamics methods, multi-tact tactile feedback force is calculated in real time and the user is extracted. Operational characteristics, use machine learning algorithms to evaluate the quality of first aid operations, generate personalized operation prompts and improvement strategies.
It improves the authenticity and effectiveness of first aid training, provides objective and accurate feedback and guidance, helps students improve their skills, reduces training costs, and realizes intelligent interaction and evaluation of simulators.
Smart Images

Figure CN118917976A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to simulation analysis technology, and in particular to a method and system for first aid effectiveness analysis based on simulated humans. Background Art
[0002] Traditional first aid training mainly relies on theoretical learning and simulator training, which has problems such as high training costs, poor interactivity, and difficulty in objective evaluation.
[0003] Currently, virtual reality technology has found some application in medical education, such as surgical simulation training systems and anatomical visualization platforms. These systems create realistic virtual scenes and interactive methods, providing an immersive learning experience and improving learning efficiency and enthusiasm. However, existing virtual reality medical education systems primarily focus on training specific medical procedures and lack comprehensive assessment and feedback mechanisms for students.
[0004] In the field of first aid training, manikins are a commonly used teaching tool. By simulating the physiological characteristics and pathological reactions of the human body, they provide trainees with a near-realistic training environment. However, traditional manikins lack intelligent interaction and feedback capabilities, making it impossible to provide objective evaluation and guidance based on trainee performance. Furthermore, manikins are expensive to use and difficult to maintain and update, limiting their widespread application in first aid training. Summary of the Invention
[0005] The embodiments of the present invention provide a method and system for analyzing first aid effectiveness based on a simulated person, which can solve the problems in the prior art.
[0006] According to a first aspect of the embodiments of the present invention, Provides a method for analyzing the effectiveness of first aid based on a simulator, including: Using medical imaging equipment to collect tomographic data of key parts of the human body, extracting soft tissue contour information from the tomographic data through medical image processing technology, and using a three-dimensional reconstruction algorithm to generate a triangular mesh model of the soft tissue. The triangular mesh model of the soft tissue is tetrahedronized, and an adaptive mesh encryption strategy and a mesh smoothing algorithm are used to generate a discretized model suitable for first aid simulation; A collision detection algorithm is used to detect the contact state between the user's hand and the first aid simulation discretization model in real time. The virtual finger is modeled as a mass-spring system with stiffness and damping. Based on the contact state between the virtual finger and the soft tissue surface, a position-based dynamics method is used to calculate multi-point tactile feedback force in real time. When a collision event is detected, the collision force is applied to the first aid simulation discretization model as a boundary condition to drive the soft tissue to deform. The operation features corresponding to the user's hand when a collision event occurs are extracted, and the operation features include operation position, operation speed, operation force and operation trajectory. Combined with a preset operation evaluation rule library, the first aid operation quality score is determined through a preset quality evaluation model; personalized operation prompts and improvement strategies are adaptively generated based on the first aid operation quality score, wherein the preset quality evaluation model is constructed based on a machine learning algorithm.
[0007] In an optional embodiment, The method further comprises: According to the nonlinear, anisotropic and incompressible mechanical characteristics of human soft tissue, a high-order hyperelastic constitutive model is used to set the material parameters of the first aid simulation discretization model, and a nonlinear control equation of soft tissue deformation is established. The nonlinear control equation is spatially discretized using the finite element method to obtain the deformation response of the first aid simulation discretization model under the action of external force.
[0008] In an optional embodiment, Medical imaging equipment is used to collect tomographic data of key parts of the human body. Medical image processing technology is used to extract the contour information of soft tissue from the tomographic data. A triangular mesh model of the soft tissue is generated using a three-dimensional reconstruction algorithm. The triangular mesh model of the soft tissue is tetrahedronized. An adaptive mesh encryption strategy and a mesh smoothing algorithm are used to generate a discretized model suitable for first aid simulation, including: Perform a tomographic scan of the target organ using medical imaging equipment to obtain a series of two-dimensional tomographic images; Preprocessing and segmenting the two-dimensional tomographic image, using image enhancement technology to improve the image contrast and signal-to-noise ratio, and combining a threshold segmentation algorithm to extract two-dimensional contour information of the soft tissue region of interest; According to the two-dimensional contour information, the contours in adjacent tomographic images are triangulated using the Marching Cubes algorithm to generate a triangular mesh model of the soft tissue; The triangular mesh model is optimized and subdivided, and the irregular boundaries of the surface of the triangular mesh model are eliminated using the Laplace smoothing algorithm; the optimized triangular mesh model is converted into a volume mesh model, and a tetrahedral mesh inside the soft tissue is generated according to the shape and topology of the volume mesh model using a tetrahedral subdivision algorithm as a discretized model for first aid simulation.
[0009] In an optional embodiment, A collision detection algorithm is used to detect the contact state between the user's hand and the first aid simulation discretization model in real time. The virtual finger is modeled as a mass-spring system with stiffness and damping. According to the contact state between the virtual finger and the soft tissue surface, a position-based dynamic method is used to calculate the multi-point tactile feedback force in real time, including: The motion capture device collects the user's hand's three-dimensional position and posture information in real time to obtain motion data of the virtual finger. The virtual finger is divided into a hierarchical bounding box, and a bounding box tree is constructed. By traversing the bounding box tree and excluding areas where collision is unlikely, the collision detection calculation range is narrowed to local leaf nodes. Precalculate and store the distance field of the soft tissue surface to obtain a distance field voxel grid. During collision detection, the vertices of the virtual finger are mapped to the distance field voxel grid. The directed distance from the vertex to the distance field voxel grid is obtained through trilinear interpolation. If the distance is less than a given distance threshold, a collision is determined. The virtual finger is abstracted into a discretized model consisting of mass points and springs. By adjusting the mass of the mass points, the stiffness and damping coefficient of the springs, virtual fingers of different materials and shapes can be simulated. After a collision is detected, a position-based dynamics method is used to iteratively optimize the position of the virtual finger's mass point to satisfy the collision and spring constraints, and calculate the tactile feedback force at each contact point. The magnitude and direction of the friction force at each contact point are calculated according to Coulomb's friction law, simulating the sliding and viscosity effects between the finger and the soft tissue surface, and incorporating the changes in surface normal and curvature caused by soft tissue deformation into the calculation of the feedback force.
[0010] In an optional embodiment, Using a position-based dynamics method, the position of the virtual finger's mass point is iteratively optimized to satisfy the collision and spring constraints. The tactile feedback force at each contact point is calculated, including: The virtual finger is modeled as a discretized model consisting of mass points and springs. The mass points represent the joints and surface feature points of the finger, and the springs represent the elastic connections between the mass points. The soft tissue is voxelized to obtain a uniform hexahedral mesh, where each voxel contains material properties and stress-strain relationships; In the collision detection phase, it is determined whether the virtual finger's mass point contacts the soft tissue surface, and the local deformation of the soft tissue at the collision point is calculated; According to the deformation at the collision point and the motion state of the virtual finger mass, a position-based dynamic equation group is established to solve the displacement and velocity of the mass point; Based on the estimated position of the virtual finger mass, the penetration depth between the virtual finger mass and the soft tissue surface is calculated, and the penetration depth is introduced into the dynamic equations as a position constraint. According to the initial length and stiffness coefficient of the virtual finger spring, the elastic displacement of the spring is calculated and introduced into the dynamic equations as an internal force constraint; By solving the dynamic equations through iterative optimization, the position and velocity of the virtual finger mass are updated while satisfying the collision and spring constraints; According to the displacement and velocity of the virtual finger mass point, the interaction force exerted on the soft tissue is calculated. The interaction force includes the normal pressure along the surface normal direction and the friction force along the tangential direction. The calculated interaction force is applied to the corresponding voxel nodes on the soft tissue surface, causing deformation and stress redistribution of the soft tissue; The interaction force between the virtual finger and soft tissue is calculated and updated in real time until the user stops the operation or the preset simulation time is reached.
[0011] In an optional embodiment, Extract the operation features corresponding to the user's hand when a collision occurs. The operation features include operation position, operation speed, operation force and operation trajectory. Combined with the preset operation evaluation rule library, the first aid operation quality score is determined by the preset quality evaluation model, including: During virtual first aid training, collision events between the user's hand and virtual soft tissue are detected in real time, and the timestamp and duration of the collision are recorded; Extract the three-dimensional coordinates of the virtual finger mass point in the collision event and map it to the local coordinate system of the virtual soft tissue to obtain the operation position characteristics relative to the soft tissue surface; Calculate the operation velocity characteristics, including velocity magnitude and direction, based on the position change and time difference of the virtual finger mass point at the beginning and end of the collision event; The collision penalty force is calculated based on the penetration depth and contact speed of the virtual finger mass and the soft tissue. At the same time, the spring force is calculated based on the elastic displacement and damping coefficient of the virtual finger mass. The two are superimposed to obtain the operation force characteristics. During the collision event, the three-dimensional coordinate sequence of the virtual finger mass point is recorded at a fixed sampling frequency, and then smoothed and interpolated to obtain a continuous operation trajectory characteristic curve; The extracted operation position features, operation speed features, operation force features, and operation trajectory features are input into a pre-trained quality assessment model. This model uses a supervised learning algorithm to train expert demonstration operations and operation sample data from users of different levels to establish a mapping relationship between operation features and operation quality scores. The quality assessment model calculates the quality score of the current operation in real time based on the input information.
[0012] According to a second aspect of the embodiments of the present invention, Provides a system for analyzing the effectiveness of first aid based on simulators, including: The first unit is used to collect tomographic data of key parts of the human body using medical imaging equipment, extract soft tissue contour information from the tomographic data using medical image processing technology, generate a triangular mesh model of the soft tissue using a three-dimensional reconstruction algorithm, tetrahedralize the triangular mesh model of the soft tissue, and use an adaptive mesh encryption strategy and a mesh smoothing algorithm to generate a discretized model suitable for first aid simulation; The second unit is configured to use a collision detection algorithm to detect the contact state between the user's hand and the first aid simulation discretization model in real time, model the virtual finger as a mass-spring system with stiffness and damping, and calculate multi-point tactile feedback force in real time using a position-based dynamics method based on the contact state between the virtual finger and the soft tissue surface. When a collision event is detected, the collision force is applied to the first aid simulation discretization model as a boundary condition to drive the soft tissue to deform; The third unit is used to extract the operation features corresponding to the user's hand when a collision event occurs. The operation features include operation position, operation speed, operation force and operation trajectory, and determine the first aid operation quality score through a preset quality assessment model in combination with a preset operation evaluation rule library; based on the first aid operation quality score, personalized operation prompts and improvement strategies are adaptively generated, wherein the preset quality assessment model is constructed based on a machine learning algorithm.
[0013] According to a third aspect of the embodiments of the present invention, An electronic device is provided, comprising: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the aforementioned method.
[0014] According to a fourth aspect of the embodiments of the present invention, A computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method described above is implemented.
[0015] By collecting human tomographic data using medical imaging equipment and applying medical image processing and 3D reconstruction techniques, a high-precision, high-fidelity discretized model of human soft tissue is generated. This provides an anatomically accurate virtual environment for emergency simulation training, enhancing both the authenticity and effectiveness of the training. Adaptive mesh encryption strategies and mesh smoothing algorithms are used to optimize the soft tissue model, generating a discretized model suitable for real-time simulation. This balances computational efficiency and model quality, ensuring the smoothness and stability of simulation training.
[0016] A position-based dynamics method is introduced to calculate multi-point tactile feedback forces between virtual fingers and soft tissue in real time. Combined with a collision detection algorithm, it drives soft tissue deformation, achieving realistic force-tactile interaction and tissue response, enhancing the immersion and sense of substitution in simulation training. The system extracts user hand operation characteristics, including position, speed, force, and trajectory. Combined with a preset operation evaluation rule library and a machine learning-based quality assessment model, it evaluates the quality of the user's first aid operation in real time, providing users with objective and accurate feedback and guidance, helping them to promptly identify deficiencies and make improvements.
[0017] Based on the user's first aid operation quality score, personalized operation prompts and improvement strategies are adaptively generated to provide targeted learning guidance for users of different levels, improving the efficiency and pertinence of first aid skills training. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] Figure 1 Schematic diagram of a flow chart of a method for analyzing first aid effectiveness based on a simulated person according to an embodiment of the present invention; Figure 2 Schematic diagram of the structure of a system for analyzing first aid effectiveness based on a simulated person according to an embodiment of the present invention. DETAILED DESCRIPTION
[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.
[0020] The following specific embodiments are used to describe the technical solution of the present invention in detail. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.
[0021] Figure 1 FIG. 1 is a flow chart of a method for analyzing first aid effectiveness based on a simulated person according to an embodiment of the present invention. Figure 1 As shown, the method includes: S101. Using medical imaging equipment to collect tomographic data of key human body parts, extracting soft tissue contour information from the tomographic data using medical image processing techniques, and generating a triangular mesh model of the soft tissue using a 3D reconstruction algorithm. The triangular mesh model of the soft tissue is then tetrahedralized, and an adaptive mesh refinement strategy and mesh smoothing algorithm are employed to generate a discretized model suitable for emergency medical simulation. S102. Using a collision detection algorithm, detect the contact state between the user's hand and the first aid simulation discretization model in real time. Model the virtual finger as a mass-spring system with stiffness and damping. Based on the contact state between the virtual finger and the soft tissue surface, calculate multi-point tactile feedback force in real time using a position-based dynamics method. When a collision event is detected, apply the collision force as a boundary condition to the first aid simulation discretization model to drive the soft tissue to deform. S103. Extract the operation features corresponding to the user's hand when a collision event occurs, wherein the operation features include operation position, operation speed, operation force and operation trajectory, and determine the first aid operation quality score through a preset quality assessment model in combination with a preset operation evaluation rule library; adaptively generate personalized operation prompts and improvement strategies based on the first aid operation quality score, wherein the preset quality assessment model is constructed based on a machine learning algorithm.
[0022] By collecting human tomographic data using medical imaging equipment and applying medical image processing and 3D reconstruction techniques, a high-precision, high-fidelity discretized model of human soft tissue is generated. This provides an anatomically accurate virtual environment for emergency simulation training, enhancing both the authenticity and effectiveness of the training. Adaptive mesh encryption strategies and mesh smoothing algorithms are used to optimize the soft tissue model, generating a discretized model suitable for real-time simulation. This balances computational efficiency and model quality, ensuring the smoothness and stability of simulation training.
[0023] A position-based dynamics method is introduced to calculate multi-point tactile feedback forces between virtual fingers and soft tissue in real time. Combined with a collision detection algorithm, it drives soft tissue deformation, achieving realistic force-tactile interaction and tissue response, enhancing the immersion and sense of substitution in simulation training. The system extracts user hand operation characteristics, including position, speed, force, and trajectory. Combined with a preset operation evaluation rule library and a machine learning-based quality assessment model, it evaluates the quality of the user's first aid operation in real time, providing users with objective and accurate feedback and guidance, helping them to promptly identify deficiencies and make improvements.
[0024] Based on the user's first aid operation quality score, personalized operation prompts and improvement strategies are adaptively generated to provide targeted learning guidance for users of different levels, improving the efficiency and pertinence of first aid skills training.
[0025] In an optional embodiment, The method further comprises: According to the nonlinear, anisotropic and incompressible mechanical characteristics of human soft tissue, a high-order hyperelastic constitutive model is used to set the material parameters of the first aid simulation discretization model, and a nonlinear control equation of soft tissue deformation is established. The nonlinear control equation is spatially discretized using the finite element method to obtain the deformation response of the first aid simulation discretization model under the action of external force.
[0026] For example, a high-order hyperelastic constitutive model is used to describe the stress-strain relationship of human soft tissue, targeting its nonlinear mechanical properties. Commonly used high-order hyperelastic constitutive models include the Ogden model, the Mooney-Rivlin model, and the Yeoh model. Taking the Ogden model as an example, its strain energy function consists of the principal stretch ratio, volume ratio, and material parameters, and is used to characterize the nonlinear and incompressible properties of soft tissue.
[0027] Determine the material parameters of the selected high-order hyperelastic constitutive model based on literature or experimental test data. Use an optimization algorithm (such as a genetic algorithm or particle swarm optimization) to fit the material parameters so that the constitutive model's predictions are consistent with the experimental data.
[0028] Based on a high-order hyperelastic constitutive model, the nonlinear governing equations for soft tissue deformation are established. Using the virtual work principle or the minimum potential energy principle, the equilibrium and constitutive equations for soft tissue deformation are derived, forming a system of nonlinear partial differential equations. Using the virtual work principle as an example, the governing equations for soft tissue deformation take into account factors such as strain, stress, displacement, body force, and surface force, describing the mechanical equilibrium relationships during soft tissue deformation.
[0029] The nonlinear governing equations are spatially discretized using the finite element method. The soft tissue region is divided into a mesh of tetrahedral elements, and appropriate interpolation functions (such as linear interpolation and quadratic interpolation) are selected to approximate the displacement field within the elements. Using the Galerkin method, the nonlinear governing equations are transformed into a system of nonlinear algebraic equations, where the stiffness matrix and the displacement field are nonlinearly related, and the external force load vector reflects the stress conditions on the soft tissue.
[0030] Solving a system of nonlinear algebraic equations yields the displacement field and stress-strain distributions of soft tissue deformation. Commonly used nonlinear solution methods include the Newton-Raphson iteration method, the modified Newton-Raphson iteration method, and the arc-length method. Considering the large deformation and rotational characteristics of soft tissue deformation, the total Lagrangian method or the updated Lagrangian method is used to account for nonlinear geometric effects.
[0031] The resulting displacement field is applied to the nodes of the discretized emergency simulation model, updating the coordinates of the soft tissue surface and volume mesh to reflect the soft tissue's deformation response under external forces. Simultaneously, based on the stress-strain distribution, physical quantities such as strain energy and stress-strain during the soft tissue deformation process are calculated for subsequent collision detection, tactile feedback, and quality assessment.
[0032] To improve the computational efficiency of soft tissue deformation responses, numerical techniques such as reduced integration, hourglass control, and mass reduction can be employed to reduce the number of element integration points, suppress zero-energy modes, and simplify the mass matrix. Furthermore, GPU parallel computing and preprocessing techniques can be utilized to accelerate the nonlinear finite element solution process and achieve real-time interactive simulation of soft tissue deformation.
[0033] This technical solution accurately and efficiently simulates the nonlinear deformation behavior of human soft tissue under external forces, providing realistic visual and force feedback for first aid simulation training, enhancing the authenticity and effectiveness of training. Furthermore, the physical quantities of soft tissue deformation responses provide an important basis for evaluating operational quality, facilitating intelligent skill assessment and personalized guidance.
[0034] In an optional embodiment, Medical imaging equipment is used to collect tomographic data of key parts of the human body. Medical image processing technology is used to extract the contour information of soft tissue from the tomographic data. A triangular mesh model of the soft tissue is generated using a three-dimensional reconstruction algorithm. The triangular mesh model of the soft tissue is tetrahedronized. An adaptive mesh encryption strategy and a mesh smoothing algorithm are used to generate a discretized model suitable for first aid simulation, including: Perform a tomographic scan of the target organ using medical imaging equipment to obtain a series of two-dimensional tomographic images; Preprocessing and segmenting the two-dimensional tomographic image, using image enhancement technology to improve the image contrast and signal-to-noise ratio, and combining a threshold segmentation algorithm to extract two-dimensional contour information of the soft tissue region of interest; According to the two-dimensional contour information, the contours in adjacent tomographic images are triangulated using the Marching Cubes algorithm to generate a triangular mesh model of the soft tissue; The triangular mesh model is optimized and subdivided, and the irregular boundaries of the surface of the triangular mesh model are eliminated using the Laplace smoothing algorithm; the optimized triangular mesh model is converted into a volume mesh model, and a tetrahedral mesh inside the soft tissue is generated according to the shape and topology of the volume mesh model using a tetrahedral subdivision algorithm as a discretized model for first aid simulation.
[0035] For example, a medical imaging device (e.g., CT, MRI, etc.) is used to perform a tomographic scan of a target organ (e.g., the heart, liver, etc.) to obtain a series of two-dimensional tomographic images. During the scanning process, appropriate scanning parameters (e.g., slice thickness, slice spacing, resolution, etc.) are selected to ensure image quality and data integrity.
[0036] The acquired 2D tomographic images are preprocessed and segmented. First, image enhancement techniques (such as histogram equalization and median filtering) are used to improve image contrast and signal-to-noise ratio, highlighting soft tissue edges and details. Then, threshold segmentation algorithms (such as Otsu thresholding and region growing) are combined to extract 2D contours of the soft tissue region of interest. Based on the grayscale value range and morphological characteristics of the soft tissue, the optimal threshold or seed point is determined to achieve automatic segmentation of the soft tissue from the background.
[0037] Based on the extracted 2D contour information, the Marching Cubes algorithm is used to triangulate the contours in adjacent slice images to generate a triangular mesh model of the soft tissue. The Marching Cubes algorithm traverses the voxels in adjacent slice images, determines the topological relationship of the isosurface based on the grayscale values of the voxel vertices, generates local triangular patches, and ultimately constructs a complete 3D surface mesh model.
[0038] The generated triangular mesh model is optimized and subdivided. Laplace smoothing is used to eliminate irregular boundaries and jagged noise on the surface of the triangular mesh model. Laplace smoothing iteratively shifts mesh vertices to the average position of adjacent vertices, making the mesh surface smooth and uniform. Furthermore, based on the curvature and detail characteristics of the mesh model, the triangular facets are adaptively subdivided to increase mesh density and improve model accuracy and realism.
[0039] The optimized triangular mesh model is converted into a volumetric mesh model. Based on the shape and topological relationships of the triangular mesh model, a tetrahedral mesh is generated within the soft tissue using a tetrahedral meshing algorithm. Commonly used tetrahedral meshing algorithms include Delaunay tetrahedralization, the advancing front method, and the eight-tree method. During the tetrahedral meshing process, the mesh density and element size are adaptively adjusted based on mesh quality criteria (such as minimum dihedral angle and volume ratio) to ensure that the generated tetrahedral mesh has a good shape and numerical stability.
[0040] Perform quality assessment and optimization on the generated tetrahedral mesh model. Calculate tetrahedral element quality metrics (such as condition number and distortion) and identify and delete low-quality elements. Use mesh smoothing algorithms (such as Laplacian smoothing and optimized smoothing) to improve the tetrahedral mesh quality and enhance element regularity and orthogonality. If necessary, re-mesh or refine the mesh in local areas to meet the accuracy requirements of the numerical simulation.
[0041] The optimized tetrahedral mesh model is used as the discretization model for emergency simulation and subsequent physical modeling and numerical simulation. Based on the node coordinates and element topology of the tetrahedral mesh, the material properties, boundary conditions, and load conditions of the soft tissue are defined, and a finite element system of equations is constructed. Using explicit or implicit time integration methods, the dynamic response of the soft tissue under external forces is solved to obtain physical field quantities such as deformation, stress, and strain.
[0042] The above technical solution allows the construction of a high-precision, high-fidelity discretized model of human soft tissue from medical imaging data through a series of steps, including image processing, 3D reconstruction, and mesh generation. This model not only accurately describes the anatomical structure and morphological characteristics of soft tissue, but also reflects its mechanical properties and deformation behavior, providing a reliable physical foundation and data support for emergency simulation training.
[0043] In an optional embodiment, A collision detection algorithm is used to detect the contact state between the user's hand and the first aid simulation discretization model in real time. The virtual finger is modeled as a mass-spring system with stiffness and damping. According to the contact state between the virtual finger and the soft tissue surface, a position-based dynamic method is used to calculate the multi-point tactile feedback force in real time, including: The motion capture device collects the user's hand's three-dimensional position and posture information in real time to obtain motion data of the virtual finger. The virtual finger is divided into a hierarchical bounding box, and a bounding box tree is constructed. By traversing the bounding box tree and excluding areas where collision is unlikely, the collision detection calculation range is narrowed to local leaf nodes. Precalculate and store the distance field of the soft tissue surface to obtain a distance field voxel grid. During collision detection, the vertices of the virtual finger are mapped to the distance field voxel grid. The directed distance from the vertex to the distance field voxel grid is obtained through trilinear interpolation. If the distance is less than a given distance threshold, a collision is determined. The virtual finger is abstracted into a discretized model consisting of mass points and springs. By adjusting the mass of the mass points, the stiffness and damping coefficient of the springs, virtual fingers of different materials and shapes can be simulated. After a collision is detected, a position-based dynamics method is used to iteratively optimize the position of the virtual finger's mass point to satisfy the collision and spring constraints, and calculate the tactile feedback force at each contact point. The magnitude and direction of the friction force at each contact point are calculated according to Coulomb's friction law, simulating the sliding and viscosity effects between the finger and the soft tissue surface, and incorporating the changes in surface normal and curvature caused by soft tissue deformation into the calculation of the feedback force.
[0044] For example, a motion capture device (such as a data glove or optical tracking system) collects the user's hand's three-dimensional position and posture information in real time to obtain motion data of a virtual finger. The virtual finger is then mapped into the virtual environment, keeping it synchronized with the user's hand movements.
[0045] The virtual finger is hierarchically divided into bounding boxes, constructing a bounding box tree. Using methods such as AABB (Axis-Aligned Bounding Box) or OBB (Oriented Bounding Box), the virtual finger is recursively divided into multiple levels of bounding boxes based on its geometric shape and spatial distribution. During collision detection, the bounding box tree is traversed and intersection tests are performed between bounding boxes to quickly eliminate areas where collision is unlikely, narrowing the collision detection calculation scope to local leaf nodes and improving detection efficiency.
[0046] Precompute and store the distance field for the soft tissue surface. Using algorithms such as the Fast Marching Method or the Distance Transform, calculate the shortest distance from each vertex to every other vertex on the soft tissue surface, generating a distance field voxel grid. During collision detection, the vertices of the virtual finger are mapped to the distance field voxel grid, and trilinear interpolation is used to obtain the directed distance from the vertex to the soft tissue surface. If the distance is less than a given distance threshold, the vertex is considered to have collided with the soft tissue.
[0047] The virtual finger is abstracted as a discretized model consisting of mass points and springs. Based on the virtual finger's geometry and topology, it is discretized into multiple mass points connected by springs. By adjusting the mass of the mass points, the stiffness of the springs, and the damping coefficient, virtual fingers of different materials and shapes can be simulated. The motion of the mass points is affected by the spring force, damping force, and external forces. The position and velocity of the mass points are updated using numerical integration methods (such as the explicit and implicit Euler methods).
[0048] After detecting a collision between a virtual finger and soft tissue, a position-based dynamics approach is used to calculate the tactile feedback force. The virtual finger's mass position is iteratively optimized to satisfy both collision and spring constraints. The collision constraint ensures that the virtual finger does not penetrate the soft tissue surface, while the spring constraint maintains the virtual finger's shape and mechanical properties. At each contact point, the normal feedback force and tangential friction force are calculated based on the relative position and velocity of the virtual finger and soft tissue. The normal feedback force is related to the penetration depth and contact stiffness, while the tangential friction force is calculated according to Coulomb's friction law.
[0049] Changes in the surface normal and curvature caused by soft tissue deformation are incorporated into the feedback force calculation. When a virtual finger contacts soft tissue, the soft tissue deforms, causing changes in the surface normal and curvature at the contact point. Using the finite element method or mass-spring method to simulate soft tissue deformation, information about the deformed surface normal and curvature is obtained. This information is used to modify the normal feedback force at the contact point to adapt it to the deformed soft tissue surface, enhancing the realism of the tactile feedback.
[0050] Based on the calculated tactile feedback force, a tactile feedback device (such as a force feedback glove or tactile vibrator) is driven to provide tactile stimulation to the user. The feedback force is mapped to the degrees of freedom of the tactile feedback device, controlling the motor or vibrator to produce the corresponding force or vibration effect, simulating the tactile sensation of a virtual finger contacting soft tissue. Additionally, audio feedback, such as contact sound and friction sound, can be added based on the contact state and material properties to enhance multi-sensory immersion.
[0051] The above technical solutions enable real-time collision detection and tactile feedback between virtual fingers and soft tissue. Hierarchical bounding boxes and distance fields accelerate collision detection, improving computational efficiency. A mass-spring model and position-based dynamics methods simulate the deformation and force of the virtual finger. By considering the impact of soft tissue deformation on contact force, the realism of tactile feedback is enhanced. The combined application of these technologies provides users with a realistic and natural tactile interaction experience, enhancing the immersion and teaching effectiveness of first aid simulation training.
[0052] In an optional embodiment, Using a position-based dynamics method, the position of the virtual finger's mass point is iteratively optimized to satisfy the collision and spring constraints. The tactile feedback force at each contact point is calculated, including: The virtual finger is modeled as a discretized model consisting of mass points and springs. The mass points represent the joints and surface feature points of the finger, and the springs represent the elastic connections between the mass points. The soft tissue is voxelized to obtain a uniform hexahedral mesh, where each voxel contains material properties and stress-strain relationships; In the collision detection phase, it is determined whether the virtual finger's mass point contacts the soft tissue surface, and the local deformation of the soft tissue at the collision point is calculated; According to the deformation at the collision point and the motion state of the virtual finger mass, a position-based dynamic equation group is established to solve the displacement and velocity of the mass point; Based on the estimated position of the virtual finger mass, the penetration depth between the virtual finger mass and the soft tissue surface is calculated, and the penetration depth is introduced into the dynamic equations as a position constraint. According to the initial length and stiffness coefficient of the virtual finger spring, the elastic displacement of the spring is calculated and introduced into the dynamic equations as an internal force constraint; By solving the dynamic equations through iterative optimization, the position and velocity of the virtual finger mass are updated while satisfying the collision and spring constraints; According to the displacement and velocity of the virtual finger mass point, the interaction force exerted on the soft tissue is calculated. The interaction force includes the normal pressure along the surface normal direction and the friction force along the tangential direction. The calculated interaction force is applied to the corresponding voxel nodes on the soft tissue surface, causing deformation and stress redistribution of the soft tissue; The interaction force between the virtual finger and soft tissue is calculated and updated in real time until the user stops the operation or the preset simulation time is reached.
[0053] For example, a virtual finger is modeled as a discretized model consisting of mass points and springs. The mass points represent the joints and surface features of the finger, describing its shape and motion. The mass points are connected by springs, whose stiffness and damping coefficient determine the elasticity and damping properties of the finger. By adjusting the mass and position of the mass points and the parameters of the springs, virtual fingers of different shapes and materials can be simulated.
[0054] Soft tissue is segmented and discretized into a uniform hexahedral mesh. Each voxel contains material properties (such as density and Young's modulus) and stress-strain relationships (such as hyperelastic and viscoelastic models). Using the finite element method or mass-spring method, a physical model of the soft tissue is established to calculate its deformation and stress distribution under external forces.
[0055] During the collision detection phase, we determine whether the virtual finger's mass contacts the soft tissue surface. Acceleration algorithms such as hierarchical bounding boxes and distance fields are used to quickly identify mass points and voxels that may collide. For any mass points that do contact, we calculate the depth and direction of their penetration into the soft tissue surface, and estimate the local deformation of the soft tissue at the point of contact based on the penetration depth.
[0056] Based on the deformation at the collision point and the motion state of the virtual finger mass, a set of position-based dynamic equations is established. These equations include Newton's equations of motion for the mass and Hooke's law for springs, describing the motion of the mass under the influence of external forces and constraints. Numerical integration methods (such as explicit and implicit Euler methods) are used to solve the dynamic equations and obtain the displacement and velocity of the mass.
[0057] Based on the estimated position of the virtual finger mass, the penetration depth into the soft tissue surface is calculated. This penetration depth is incorporated into the dynamic equations as a position constraint to ensure that the virtual finger does not over-penetrate the soft tissue. The product of the penetration depth and the contact stiffness is used to calculate the normal contact force, which serves as an external constraint on the mass's motion.
[0058] Based on the initial length and stiffness coefficient of the virtual finger spring, the spring's elastic displacement is calculated. This elastic displacement is introduced as an internal force constraint into the dynamic equations to maintain the shape and mechanical properties of the virtual finger. The product of the spring's elastic force and elastic displacement can be used to calculate the internal constraint force between the two end masses of the spring.
[0059] The dynamic equations are solved through iterative optimization, updating the position and velocity of the virtual finger mass while satisfying collision and spring constraints. Common optimization algorithms include sequential quadratic programming (SQP) and the Lagrange multiplier method. Through multiple iterations, the virtual finger reaches a state of equilibrium under the influence of external forces and constraints, achieving a stable and realistic interaction effect.
[0060] The interaction force exerted on the soft tissue is calculated based on the displacement and velocity of the virtual finger mass. This interaction force consists of a normal pressure along the surface normal direction and a friction force along the tangential direction. The normal pressure is related to the penetration depth and contact stiffness, while the friction force is calculated according to Coulomb's friction law and is proportional to the normal pressure and the friction coefficient. The calculated interaction force is applied to the corresponding voxel node on the soft tissue surface, causing deformation and stress redistribution in the soft tissue.
[0061] The interaction force is applied as an external load to the soft tissue physical model, and the soft tissue deformation response is calculated using the finite element method or mass-spring method. The soft tissue deformation causes changes in the surface normal and curvature, which in turn affects collision detection and interaction force calculation in the next time step. By updating the soft tissue's shape and mechanical state in real time, bidirectional coupled interaction between the virtual finger and the soft tissue is achieved.
[0062] The interaction force between the virtual finger and soft tissue is calculated and updated in real time until the user stops or the preset simulation time expires. During the interaction process, collision detection, constraint solving, and physics simulation parameters are dynamically adjusted based on the movement of the virtual finger and the deformation of the soft tissue, balancing computational efficiency and realism. Simultaneously, the calculated interaction force is fed back to a force feedback device, such as a haptic glove or stylus, providing the user with a realistic tactile experience.
[0063] The above technical solution enables realistic and stable interactive simulation between a virtual finger and soft tissue. A position-based dynamics approach achieves bidirectional coupling between the finger and soft tissue by introducing collision and spring constraints, while simultaneously considering the deformation and motion of the virtual finger, as well as the physical properties and deformation response of the soft tissue. Iterative optimization solves the dynamics equations to ensure natural finger movement and realistic touch within physical constraints. This technical solution is of great significance for improving simulation quality and user experience in applications such as medical surgical training and product design evaluation.
[0064] In an optional embodiment, Extract the operation features corresponding to the user's hand when a collision occurs. The operation features include operation position, operation speed, operation force and operation trajectory. Combined with the preset operation evaluation rule library, the first aid operation quality score is determined by the preset quality evaluation model, including: During virtual first aid training, collision events between the user's hand and virtual soft tissue are detected in real time, and the timestamp and duration of the collision are recorded; Extract the three-dimensional coordinates of the virtual finger mass point in the collision event and map it to the local coordinate system of the virtual soft tissue to obtain the operation position characteristics relative to the soft tissue surface; Calculate the operation velocity characteristics, including velocity magnitude and direction, based on the position change and time difference of the virtual finger mass point at the beginning and end of the collision event; The collision penalty force is calculated based on the penetration depth and contact speed of the virtual finger mass and the soft tissue. At the same time, the spring force is calculated based on the elastic displacement and damping coefficient of the virtual finger mass. The two are superimposed to obtain the operation force characteristics. During the collision event, the three-dimensional coordinate sequence of the virtual finger mass point is recorded at a fixed sampling frequency, and then smoothed and interpolated to obtain a continuous operation trajectory characteristic curve; The extracted operation position features, operation speed features, operation force features, and operation trajectory features are input into a pre-trained quality assessment model. This model uses a supervised learning algorithm to train expert demonstration operations and operation sample data from users of different levels to establish a mapping relationship between operation features and operation quality scores. The quality assessment model calculates the quality score of the current operation in real time based on the input information.
[0065] For example, during virtual first aid training, a collision detection algorithm detects collisions between the user's hand (virtual finger) and virtual soft tissue in real time. When a collision is detected, the collision timestamp and duration are recorded for subsequent feature extraction and quality assessment.
[0066] The 3D coordinates of the virtual finger mass point during the collision event are extracted and converted from the global coordinate system to the local coordinate system of the virtual soft tissue. By calculating the position of the virtual finger mass point relative to the soft tissue surface, the operation position feature is obtained, which represents the contact position and depth of the finger on the soft tissue.
[0067] The operation velocity signature is calculated based on the position change and time difference between the virtual finger mass point at the start and end of the collision event. By dividing the change in mass point position along the three coordinate axes by the time difference, the velocity components in each direction are obtained, and the magnitude and direction of the velocity are then calculated. The operation velocity signature reflects the speed and direction of the user's hand movement.
[0068] The collision penalty force is calculated based on the penetration depth and contact velocity of the virtual finger mass point with the soft tissue. The greater the penetration depth and the faster the contact velocity, the greater the penalty force, which is used to prevent the finger from over-penetrating the soft tissue. Simultaneously, the spring force is calculated based on the elastic displacement (displacement relative to the initial position) and the damping coefficient of the virtual finger mass point. The greater the elastic displacement, the greater the spring force, which is used to maintain the shape and mechanical properties of the finger. The collision penalty force and the spring force are superimposed to obtain the operation force characteristic, which reflects the force applied by the user on the soft tissue.
[0069] During the collision event, the 3D coordinate sequence of the virtual finger mass point is recorded at a fixed sampling frequency (e.g., 30 times per second). This acquired coordinate sequence is smoothed to remove high-frequency noise and jitter, and an interpolation algorithm (such as spline interpolation) is used to generate a continuous trajectory curve between the sampling points. The smoothed and interpolated coordinate sequence constitutes the manipulation trajectory feature, reflecting the path and continuity of the user's hand movement.
[0070] The extracted operation position, speed, force, and trajectory features are fed into a pre-trained quality assessment model. This model uses supervised learning algorithms, such as support vector machines (SVMs) and random forests, to train on sample data from expert demonstrations and user operations at varying skill levels. By mapping operation features to expert-assigned operation quality scores, the model learns the correlation between different combinations of operation features and operation quality.
[0071] The quality assessment model calculates the quality score of the current operation in real time based on the input operation characteristics. The score can be a continuous value (e.g., 0-100) or a discrete level (e.g., excellent, good, qualified, unqualified). The score reflects the similarity and quality level between the user operation and the standard operation, and can be used for timely feedback and guidance on user improvement.
[0072] To improve the accuracy and robustness of quality assessment, multiple evaluation metrics can be introduced, such as operation completion time, operation accuracy, and operation smoothness. The scores of different metrics can be weighted and combined to obtain a comprehensive quality score. At the same time, through cross-validation and data augmentation techniques, the generalization and adaptability of the quality assessment model can be improved, enabling it to meet the evaluation needs of different users and different operation scenarios.
[0073] During virtual first aid training, quality assessment results are displayed in real time, and visual and auditory feedback is combined to guide users in optimizing their operations. For example, visual elements like color coding and progress bars can be used to directly demonstrate the quality of operations. Voice prompts and warning sounds can also be used to remind users of key steps and common errors. Multimodal feedback helps enhance user awareness and learning outcomes.
[0074] After training, a detailed quality assessment report is generated, including scores for various operational indicators, comparative analysis with standard operations, and a summary of strengths and weaknesses. This report helps users comprehensively review their operational performance, identify areas for improvement, and provide reference and goals for the next training session. Through continuous training and assessment, users can gradually improve operational quality to a standardized and proficient level.
[0075] The above technical solution achieves automatic evaluation and feedback of virtual first aid operations by extracting user operation features in real time and combining them with a pre-trained quality assessment model. This solution integrates technologies such as collision detection, feature extraction, and machine learning to comprehensively evaluate key factors such as the position, speed, force, and trajectory of the operation, providing users with objective and accurate quality evaluations. At the same time, through multimodal feedback and detailed reports, it helps users intuitively understand their operation performance, improve training in a targeted manner, and enhance learning efficiency and effectiveness. This technical solution is of great value in improving the intelligence level and teaching quality of virtual first aid training.
[0076] Figure 2 FIG. 1 is a structural diagram of a system for analyzing first aid effectiveness based on a simulated person according to an embodiment of the present invention. Figure 2 As shown, the system includes: The first unit is used to collect tomographic data of key parts of the human body using medical imaging equipment, extract soft tissue contour information from the tomographic data using medical image processing technology, generate a triangular mesh model of the soft tissue using a three-dimensional reconstruction algorithm, tetrahedralize the triangular mesh model of the soft tissue, and use an adaptive mesh encryption strategy and a mesh smoothing algorithm to generate a discretized model suitable for first aid simulation; The second unit is configured to use a collision detection algorithm to detect the contact state between the user's hand and the first aid simulation discretization model in real time, model the virtual finger as a mass-spring system with stiffness and damping, and calculate multi-point tactile feedback force in real time using a position-based dynamics method based on the contact state between the virtual finger and the soft tissue surface. When a collision event is detected, the collision force is applied to the first aid simulation discretization model as a boundary condition to drive the soft tissue to deform; The third unit is used to extract the operation features corresponding to the user's hand when a collision event occurs. The operation features include operation position, operation speed, operation force and operation trajectory, and determine the first aid operation quality score through a preset quality assessment model in combination with a preset operation evaluation rule library; based on the first aid operation quality score, personalized operation prompts and improvement strategies are adaptively generated, wherein the preset quality assessment model is constructed based on a machine learning algorithm.
[0077] According to a third aspect of the embodiments of the present invention, An electronic device is provided, comprising: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the aforementioned method.
[0078] According to a fourth aspect of the embodiments of the present invention, A computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method described above is implemented.
[0079] The present invention may be a method, an apparatus, a system and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present invention.
[0080] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for analyzing the effectiveness of first aid based on a simulated person, characterized in that: include: Using medical imaging equipment to collect tomographic data of key parts of the human body, extracting contour information of soft tissue from the tomographic data through medical image processing technology, and using a three-dimensional reconstruction algorithm to generate a triangular patch mesh model of the soft tissue, tetrahedralizing the triangular patch mesh model of the soft tissue, and using an adaptive mesh encryption strategy and a mesh smoothing algorithm to generate a discretized model suitable for emergency simulation; A collision detection algorithm is used to detect the contact state between the user's hand and the first aid simulation discretization model in real time, and a virtual finger is modeled as a mass-spring system with stiffness and damping. According to the contact state between the virtual finger and the soft tissue surface, a position-based dynamic method is used to calculate the multi-point tactile feedback force in real time. When a collision event is detected, the collision force is applied to the first aid simulation discretization model as a boundary condition to drive the soft tissue to deform. The operation features corresponding to the user's hand when a collision event occurs are extracted, and the operation features include operation position, operation speed, operation force and operation trajectory. Combined with a preset operation evaluation rule library, the first aid operation quality score is determined through a preset quality evaluation model; personalized operation prompts and improvement strategies are adaptively generated according to the first aid operation quality score, wherein the preset quality evaluation model is constructed based on a machine learning algorithm.
2. The method according to claim 1, characterized in that: The method further comprises: According to the nonlinear, anisotropic and incompressible mechanical characteristics of human soft tissue, a high-order hyperelastic constitutive model is used to set the material parameters of the first aid simulation discretization model, and the nonlinear control equation of soft tissue deformation is established. The finite element method is used to spatially discretize the nonlinear control equation to obtain the deformation response of the first aid simulation discretization model under the action of external force.
3. The method according to claim 1, characterized in that The tomographic data of key parts of the human body are collected by medical imaging equipment, the contour information of soft tissue is extracted from the tomographic data by medical image processing technology, and a triangular patch mesh model of the soft tissue is generated by a three-dimensional reconstruction algorithm. The triangular patch mesh model of the soft tissue is tetrahedronized, and an adaptive mesh encryption strategy and a mesh smoothing algorithm are used to generate a discretized model suitable for emergency simulation, including: Use medical imaging equipment to perform tomography on the target organ to obtain a series of two-dimensional tomographic images; Preprocessing and segmenting the two-dimensional tomographic image, using image enhancement technology to improve the image contrast and signal-to-noise ratio, and combining a threshold segmentation algorithm to extract two-dimensional contour information of the soft tissue region of interest; According to the two-dimensional contour information, the contours in adjacent tomographic images are triangulated using the Marching Cubes algorithm to generate a triangulated mesh model of the soft tissue; The triangular patch mesh model is optimized and subdivided, and the irregular boundaries of the surface of the triangular patch mesh model are eliminated by using a Laplace smoothing algorithm; the optimized triangular patch mesh model is converted into a volume mesh model, and a tetrahedral mesh inside the soft tissue is generated according to the shape and topology of the volume mesh model by using a tetrahedral partitioning algorithm as a discretization model for first aid simulation.
4. The method according to claim 1, characterized in that: A collision detection algorithm is used to detect the contact state between the user's hand and the first aid simulation discretization model in real time, and the virtual finger is modeled as a mass-spring system with stiffness and damping. According to the contact state between the virtual finger and the soft tissue surface, a position-based dynamic method is used to calculate the multi-point tactile feedback force in real time, including: The motion capture device collects the 3D position and posture information of the user's hand in real time to obtain the motion data of the virtual finger; the virtual finger is divided into bounding boxes, and a bounding box tree is constructed. By traversing the bounding box tree, areas where collision is unlikely to occur are excluded, and the calculation range of collision detection is narrowed to local leaf nodes; Precalculate and store the distance field of the soft tissue surface to obtain the distance field voxel grid. During collision detection, map the vertices of the virtual finger to the distance field voxel grid. Use trilinear interpolation to obtain the directed distance from the vertex to the distance field voxel grid. If the distance is less than a given distance threshold, it is determined that a collision occurs. The virtual finger is abstracted into a discretized model consisting of mass points and springs. By adjusting the mass of the mass points, the stiffness and damping coefficient of the springs, virtual fingers of different materials and shapes can be simulated. After a collision is detected, a position-based dynamics method is used to iteratively optimize the position of the virtual finger mass to satisfy the collision constraint and spring constraint, and calculate the tactile feedback force at each contact point; The magnitude and direction of the friction force at each contact point are calculated according to Coulomb's friction law, the sliding and viscosity effects between the finger and the soft tissue surface are simulated, and the changes in the surface normal and curvature caused by the deformation of the soft tissue are incorporated into the calculation of the feedback force.
5. The method according to claim 4, characterized in that The position-based dynamics method is used to iteratively optimize the position of the virtual finger mass point to satisfy the collision constraint and spring constraint. The tactile feedback force at each contact point is calculated, including: The virtual finger is modeled as a discretized model consisting of mass points and springs. The mass points represent the joints and surface feature points of the finger, and the springs represent the elastic connections between the mass points. The soft tissue is voxelized and segmented to obtain a uniform hexahedral mesh, where each voxel contains material properties and stress-strain relationships; In the collision detection phase, it is determined whether the mass point of the virtual finger is in contact with the soft tissue surface, and the local deformation of the soft tissue at the collision point is calculated; According to the deformation at the collision point and the motion state of the virtual finger mass point, a group of dynamic equations based on position is established to solve the displacement and velocity of the mass point; According to the estimated position of the virtual finger mass point, the penetration depth between the virtual finger mass point and the soft tissue surface is calculated, and the penetration depth is introduced into the dynamic equations as a position constraint; According to the initial length and stiffness coefficient of the virtual finger spring, the elastic displacement of the spring is calculated, and the elastic displacement is introduced into the dynamic equations as an internal force constraint; By solving the dynamic equations through iterative optimization, the position and velocity of the virtual finger mass are updated while satisfying the collision constraint and spring constraint; According to the displacement and velocity of the virtual finger mass point, the interaction force exerted on the soft tissue is calculated. The interaction force includes the normal pressure along the surface normal direction and the friction force along the tangential direction. The calculated interaction force is reacted to the voxel nodes corresponding to the soft tissue surface, causing deformation and stress redistribution of the soft tissue; The interaction force between the virtual finger and soft tissue is calculated and updated in real time until the user stops the operation or the preset simulation time is reached.
6. The method according to claim 1, characterized in that The operation features corresponding to the user's hand when a collision event occurs are extracted. The operation features include the operation position, operation speed, operation force and operation trajectory. Combined with the preset operation evaluation rule library, the first aid operation quality score is determined through the preset quality evaluation model, including: During virtual first aid training, the collision event between the user's hand and the virtual soft tissue is detected in real time, and the timestamp and duration of the collision are recorded; Extract the three-dimensional coordinates of the virtual finger mass point in the collision event, map it to the local coordinate system of the virtual soft tissue, and obtain the operation position characteristics relative to the soft tissue surface; According to the position change and time difference of the virtual finger mass point at the beginning and end of the collision event, the operation speed characteristics, including the speed magnitude and direction, are calculated; The collision penalty force is calculated by the penetration depth and contact speed of the virtual finger mass point and the soft tissue. At the same time, the spring force is calculated according to the elastic displacement and damping coefficient of the virtual finger mass point. The two are superimposed to obtain the operation force characteristics. During the collision event, the three-dimensional coordinate sequence of the virtual finger mass point is recorded at a fixed sampling frequency, and smoothed and interpolated to obtain a continuous operation trajectory characteristic curve; The extracted operation position features, operation speed features, operation force features and operation trajectory features are input into the pre-trained quality assessment model. The model is trained on expert demonstration operations and operation sample data of users of different levels using a supervised learning algorithm to establish a mapping relationship between operation features and operation quality scores. The quality assessment model calculates the quality score of the current operation in real time based on the input information.
7. A system for analyzing the effectiveness of first aid based on simulated human beings, used to implement the method according to any one of claims 1 to 6, characterized in that: include: The first unit is used to collect tomographic data of key parts of the human body using medical imaging equipment, extract contour information of soft tissue from the tomographic data using medical image processing technology, generate a triangular patch mesh model of the soft tissue using a three-dimensional reconstruction algorithm, tetrahedralize the triangular patch mesh model of the soft tissue, and use an adaptive mesh encryption strategy and a mesh smoothing algorithm to generate a discretized model suitable for emergency simulation; The second unit is used to use a collision detection algorithm to detect the contact state between the user's hand and the first aid simulation discretization model in real time, model the virtual finger as a mass-spring system with stiffness and damping, and calculate the multi-point tactile feedback force in real time according to the contact state between the virtual finger and the soft tissue surface using a position-based dynamics method. When a collision event is detected, the collision force is applied to the first aid simulation discretization model as a boundary condition to drive the soft tissue to deform; The third unit is used to extract the operation features corresponding to the user's hand when a collision event occurs, wherein the operation features include operation position, operation speed, operation force and operation trajectory, and determine the first aid operation quality score through a preset quality assessment model in combination with a preset operation evaluation rule library; and adaptively generate personalized operation prompts and improvement strategies based on the first aid operation quality score, wherein the preset quality assessment model is constructed based on a machine learning algorithm.
8. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method according to any one of claims 1 to 6.
9. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 6 is implemented.
Citation Information
Patent Citations
Method for simulating tension / deformation on soft tissue epidermis of any shape
CN103745058A
Orbital surgery training method based on virtual reality, system and storage medium
CN109979600A
Vascular intervention operation training construction method and system based on deep reinforcement learning
CN114118771A
Visual and tactile fusion feedback method and system for interaction of virtual hand and deformable body
CN117873321A
Method for generating sense of reality of virtual object in teaching scene
US11282404B1
Cited By
Analogue simulation system based on plastic surgery
CN119170287A
Biological soft tissue virtual simulation method and system and medium
CN120337692A
