A method for estimating guidewire morphology and stress in vascular interventional surgery robots

Abstract

The application discloses a method for estimating the stress of a guide wire in the shape of a vascular interventional surgery robot, and the method comprises the following steps: S1, designing a state expression of the guide wire and a mapping relationship from a state space to an observation space; S2, designing a likelihood function for evaluating state estimation results and introducing a regularization term for processing multiple solutions; S3, applying a particle filter method to track and estimate the state of the guide wire in real time; and S4, visualizing the estimation results. The guide wire shape and stress information estimated according to the image can be enhanced and displayed on the original image, thereby assisting a doctor in judging the surgery state.

Description

Technical Field

[0001] This invention relates to the field of surgical robot operation, and in particular to a method for estimating the force on guidewire morphology in vascular interventional surgical robots. Background Technology

[0002] In vascular interventional surgery, medical catheters and guidewires serve as surgical instruments, crucial for surgeons to successfully complete the procedure. The vascular environment within the human body is complex, and some vessel walls are extremely fragile. To prevent malpractice that could endanger the patient's life, we need to monitor the guidewire's morphology and stress within the vessel in real time. Due to space constraints, compared to instruments used in interventional surgeries of natural cavities such as the trachea, it is difficult to embed sensors in the guidewires used in vascular interventional surgery; therefore, surgical information is typically obtained only with the aid of external medical imaging equipment.

[0003] The following is a comparison with existing technologies:

[0004] Technical comparison with patent CN110729051A "A method, system and electronic device for guidewire mechanical analysis in interventional surgery"

[0005] The estimation of forces acting on the guidewire in patent CN110729051A involves first extracting the centerline shape of the guidewire from a 3D image, and then solving for the volume force on the unit element based on the constitutive relationship between stress and strain in continuum mechanics. In contrast, this invention first searches and samples the possible force state space based on the previous moment's force state of the guidewire using particle filtering. It then calculates the 3D deformation shape of the guidewire under the corresponding external force using forward dynamic equations. Finally, it renders the corresponding 2D shape of the guidewire on the image plane based on the scene's camera parameters and compares it with the actual image to calculate the likelihood, thereby obtaining the optimal estimate of the guidewire's stress state.

[0006] Second, the guidewire force estimation method in patent CN110729051A is based on single-frame image calculation, which does not consider the state at previous moments. In contrast, the state estimation method of this invention considers temporal state information, resulting in better stability and continuity of the estimation results. Furthermore, the particle filtering method used in this invention is based on Monte Carlo sampling, mitigating the effects of system nonlinearity and exhibiting better robustness to observation noise.

[0007] Comparison with patent CN115227394A "Risk Warning Method for Robotic Minimally Invasive Vascular Interventional Surgery Based on Digital Twin"

[0008] First, the estimation of guidewire force in patent CN115227394A requires the pre-establishment of a digital environment for the surgical scene, including the vascular network structure, material properties of objects such as blood vessels and guidewires, etc. It involves executing the same control commands in both the simulation and real environments, and reading force information from the simulation environment as a reference for the real environment. In contrast, this invention only requires pre-setting the guidewire material parameters and image rendering-related parameters, making preoperative preparation much simpler.

[0009] Second, patent CN115227394A uses the Kirchhoff rod model to simulate the guidewire, while this invention uses the Cosserat rod model. The Kirchhoff rod model simplifies the strain of the elastic rod, assuming that the strain of a slender elastic rod only includes bending and torsional strain. In contrast, the Cosserat rod model also models the tensile, compressive, and shear strains of the slender elastic rod, making it adaptable to more scenarios.

[0010] This invention proposes a method for estimating and visualizing guidewire morphology and stress based on medical image information. It can enhance the display of guidewire morphology and stress information estimated from images on the original images, assisting doctors in judging the surgical status. Summary of the Invention

[0011] The purpose of this invention is to overcome the above-mentioned defects of the prior art and propose a method for estimating the guidewire morphology and force for vascular interventional surgery robots. This method can estimate the three-dimensional morphology of the guidewire and the contact force between the guidewire and the blood vessel wall based on real-time medical image information, and enhance the display on the original image to assist doctors in judging the surgical status.

[0012] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0013] A method for estimating guidewire morphology and force in a vascular interventional surgical robot according to an embodiment of the present invention is characterized in that;

[0014] S1. Design the state representation of the guidewire and the mapping relationship from the state space to the observation space:

[0015] S2. Design a likelihood function to evaluate the state estimation results, and introduce a regularization term to handle the case of multiple solutions:

[0016] S3. Apply particle filtering method to perform real-time tracking and estimation of guidewire status:

[0017] S4. Visualize the estimation results.

[0018] Furthermore, step S1 is specifically as follows;

[0019] The guidewire's state is encoded by the force information, i.e., the force applied to the middle section of the guidewire is N.F The state of the guidewire is represented by the forces acting on it and their relative positions, as well as the force and torque information at the fixed end. This is denoted as... Based on the Cosserat rod model, a forward dynamic equation is established to map the state space to the observation space φ. The three-dimensional strain shape of the guidewire under the action of external force is calculated from the state variables. It is approximately represented by N pose matrices taken at equal intervals along the centerline, denoted as C∈SE(3). N .

[0020] Furthermore, step S2 is specifically as follows;

[0021] Based on the imaging model of the corresponding imaging equipment, the estimated three-dimensional shapes of each guidewire are projected onto the imaging plane, denoted as... With the observed images Compare the likelihood of the difference calculation. The likelihood function L is calculated based on the nearest point distance between two pixel sets in the observed guidewire image and the estimated guidewire image, denoted as L. For the possibility of multiple solutions in state estimation.

[0022] Furthermore, in step S2, a regularization term designed for the application scenario was added to the likelihood function for suppression, including;

[0023] 1) Regularization of the direction of external force applied to the middle section of the guidewire;

[0024] Because the medical guidewires used in vascular interventional procedures are relatively smooth, and blood flow further reduces the friction between the guidewire and the vessel wall, the force between the guidewire and the vessel wall is a supporting force. The force on the middle section of the guidewire should be approximately orthogonal to the centerline direction of the guidewire at the contact point. This regularization term is expressed as follows: Where F i Let t represent the i-th force. i This represents the centerline tangent vector of the guidewire at the location of the applied force;

[0025] 2) Based on imaging information, the location where the guidewire may come into contact with the blood vessel wall can be determined in advance;

[0026] It is assumed that the estimated location of the external force acting on the guidewire should be close to the possible contact location determined based on the image. This regularization term is expressed as follows: in This indicates the location of the force determined in advance based on the image, s i Indicates the estimated location of the force;

[0027] In the above formula, N F Let represent the total number of external forces, α1 and α2 be the weighting coefficients of the regularization term, and the final likelihood function be...

[0028] Furthermore, step S3 is specifically as follows;

[0029] The possible states are iteratively searched using particle filtering, where each particle actually represents a possible state. The i-th particle at time t is represented as... Before the first iteration, for all N p Each particle is randomly initialized within a specified state space. Before each subsequent iteration, the particle sampling position at the current moment is determined based on the state values ​​from previous moments.

[0030]

[0031] Where k and ∑ are adjustable parameters that control the sampling range of the particles, the guidewire morphology corresponding to each state variable is calculated through the aforementioned mapping φ in a forward manner. Render the guide wire image corresponding to the estimated state based on the camera parameters of the scene. With the observed images Substituting the values ​​into the likelihood function, we can calculate the estimated weights representing the state of the particle. The optimal stress state estimate is obtained based on the weights.

[0032] Furthermore, step S4 is specifically as follows;

[0033] After obtaining the estimated values ​​of the guidewire's shape and stress information, the three-dimensional shape of the guidewire and the external forces it experiences are rendered in the virtual scene according to the scene's camera parameters. These are represented by curves and arrows, and the information is enhanced and displayed on the original image to help doctors judge the surgical status.

[0034] The beneficial effects of this invention are:

[0035] (1) The present invention models the dynamics of guidewires based on the Cosserat rod model, and can adapt to medical guidewires of different shapes and materials by pre-setting physical parameters;

[0036] (2) The computational process of this invention is based on the particle filter framework, which has good robustness in dealing with the state estimation problem of nonlinear systems.

[0037] (3) The present invention uses GPU hardware acceleration in the solution process, which can achieve near real-time visualization of morphological force estimation. Attached Figure Description

[0038] Figure 1 This is a schematic diagram illustrating the visual effects in an embodiment of the present invention;

[0039] Figure 2This is a schematic diagram of the encoded information of the guidewire status in an embodiment of the present invention;

[0040] Figure 3 This is a flowchart illustrating the calculation process for estimating the stress state of the guidewire in an embodiment of the present invention.

[0041] Figure 4 This is a schematic diagram of the force estimation direction regularization term in an embodiment of the present invention;

[0042] Figure 5 This is a schematic diagram of the force estimation position regularization term in an embodiment of the present invention. Detailed Implementation

[0043] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments:

[0044] Figure 1 This is a visualization diagram illustrating the effect in an embodiment of the present invention. The left side shows the original input medical image, and the right side shows the visualization of the guidewire shape and force estimation results. The shape of the guidewire is represented by curves of different colors, and the contact force between the guidewire and the blood vessel wall is indicated by arrows of different colors at corresponding positions.

[0045] Figure 2 This is a schematic diagram illustrating the encoded information of the guidewire state in an embodiment of the present invention. Based on the application scenario, it is assumed that external torques only act on the guidewire at both ends, with the middle portion subjected to a few external forces. The relative position of the i-th force on the guidewire is determined by s. i This indicates that the magnitude and direction of the force are determined by the three-dimensional vector F. i Indicated. The external torques acting at both ends are M0 and M, respectively. L Representation. Ultimately encoded as a vector in the state space.

[0046] Figure 3This is a flowchart illustrating the calculation process for estimating the guidewire's stress state in this embodiment of the invention. Throughout the calculation, a set of particles continuously samples a specific region in the state space for state estimation, where each particle is essentially a sampled state space vector. The ordinary differential equations of the Cosserat rod model are solved numerically, and the spatial geometry of the guidewire after deformation under stress is reconstructed based on the previously defined state variables, thus completing the mapping from the state space to the observation space. In each iteration, the state transition of each particle in the current step is predicted based on its historical information in the state space. Random sampling is then performed in the neighborhood of the predicted value, and the possible shape of the guidewire is calculated forward. The estimated shape is projected onto the image plane according to the imaging model of the imaging device, and its similarity is calculated by comparing it with the observed data. Weights are assigned to the predicted states of each particle based on the degree of morphological matching to fit the true probability distribution of the guidewire state in the current step. The expected value of this distribution is used as the final predicted state for the current step. For the degradation problem that may be encountered in the particle filtering algorithm, resampling can be used to mitigate it and improve the stability of the state estimation.

[0047] Figure 4 , Figure 5 This is a schematic diagram illustrating the design of the regularization term for force estimation of the guidewire in an embodiment of the present invention. The regularization term for the direction of external force applied to the middle section of the guidewire is expressed as follows: Where F i Let t represent the i-th force. i This represents the tangent vector along the centerline of the guidewire at the location of the applied force. The regularization term for the location of the external force acting on the guidewire is expressed as... in This indicates the location of the force determined in advance based on the image, s i This indicates the estimated location of the force. In the above formula, N... F α1 and α2 represent the total number of external forces, respectively, and are the weight coefficients of the regularization term.

[0048] The technical approach of the present invention has been described in detail above with reference to examples and accompanying drawings. However, those skilled in the art will readily understand that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions resulting from these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A guidewire shape force estimation method for a vascular intervention surgery robot, characterized by ； S1. Design the state representation of the guidewire, the mapping relationship from state space to observation space; The step S1 is specifically as follows; The state of the guide wire is encoded by the force information, i.e. the force acting on the middle segment of the guide wire and the relative position of the force, the force and moment information of the fixed end, denoted as , represent the state of the guide wire, denoted as , the forward dynamics equation is established according to the Cosserat rod model to map the state space to the observation space , the three-dimensional strain shape of the guide wire under the action of the external force is calculated from the state quantity, which is approximately represented by N pose matrices taken equidistantly along the center line, denoted as ; S2. Design the likelihood function for evaluating the state estimation result, and introduce a regularization term to deal with the multiple solution situation: The step S2 is specifically as follows; Projecting each estimated guidewire 3D shape to the imaging plane according to the imaging model of the corresponding imaging device, denoted as , and comparing with the observed image , the likelihood is computed as the difference , between the two pixel sets based on the nearest point distance, denoted as ;​ S3. Apply the particle filter method to track and estimate the guidewire state in real time: The step S3 is specifically as follows; The possible states are iteratively searched using a particle filtering method, where each particle actually represents a possible state. Individual particles The time is represented as Before the first iteration, all Each particle is randomly initialized within a specified state space. Before each subsequent iteration, the particle sampling position at the current moment is determined based on the state values ​​from previous moments. ； wherein are all adjustable parameters, controlling the sampling range of particles, through the aforementioned mapping Forward solving the corresponding guidewire shape of each state variable , according to the camera parameters of the scene, rendering the guidewire image corresponding to the estimated state , and the observed image into the likelihood function, and the estimated weight of the particle representing the state is calculated , and the optimal force state estimation is obtained according to the weight ; S4. The estimation result is visualized.

2. The method of claim 1, wherein, In the step S2, the regularization term designed for the application scenario is added in the likelihood function for suppression, including; 1) The regularization of the direction of the external force acting on the middle segment of the guidewire; Since the medical guide wire used in the vascular intervention surgery scene is relatively smooth, and the blood flow will further reduce the friction between the guide wire and the blood vessel wall, the force between the guide wire and the blood vessel wall is the supporting force, and the middle segment of the guide wire is subjected to the force which should be basically orthogonal to the center line direction of the guide wire at the contact position. The regularization term is expressed as wherein represents the first force, represents the tangent vector of the center line of the guide wire at the force position. 2) The position where the guidewire is likely to contact the blood vessel wall is judged based on the image information in advance; It is considered that the estimated position of the force acting on the guidewire should be close to the possible contact position judged from the image, and the regularization term is expressed as wherein represents the position of the force acting on the guidewire estimated from the image, represents the position of the force acting on the guidewire estimated from the image, In the above formula, represents the total number of external forces, are the weight coefficients of the regularization terms, and the final likelihood function is .

3. The method of claim 1, wherein the method is a method of estimating a force applied to a guide wire of a vascular intervention robot. The step S4 is specifically as follows; After the shape and force information estimation value of the guidewire are obtained, the three-dimensional shape of the guidewire and the external force acting on it are rendered in the virtual scene according to the camera parameters of the scene, are represented by a curve and an arrow respectively, and the information is enhanced and displayed on the original image to assist the doctor in judging the operation state.

Images