Percutaneous puncture precision dynamic matching and navigation method and system based on AI surgical robot

Abstract

The application provides a percutaneous puncture accurate dynamic matching and navigation method and system based on an AI surgery robot, which generates a puncture scheme for puncture positioning and navigation at different stages through collection and analysis of preoperative and intraoperative patient medical image data; realizes real-time capture of tissue deformation in percutaneous puncture, further realizes dynamic optimization of a puncture path based on the real-time captured tissue deformation state, realizes real-time generation of a dynamic puncture path, and makes real-time adjustment of a motion trajectory of the robot, so as to improve the accurate navigation and dynamic matching degree of percutaneous puncture surgery and realize the purpose of accurate puncture.

Description

Technical Field

[0001] This invention relates to the field of percutaneous puncture guidance and positioning technology based on AI surgical robots, and particularly to a method and system for precise dynamic matching and navigation of percutaneous puncture based on AI surgical robots. Background Technology

[0002] Percutaneous puncture surgery, as a key technology in minimally invasive medicine, plays an important role in treatments such as tumor ablation due to its precision and low invasiveness. Precise navigation and dynamic adjustment capabilities directly determine the safety and efficacy of the procedure. However, existing methods often exhibit limitations when facing complex tissue environments. Traditional navigation systems rely on preoperative static images and struggle to adapt to intraoperative tissue deformation and displacement, leading to deviations in the puncture path and incomplete coverage of the ablation area.

[0003] In actual clinical practice, the contradiction between tissue deformation and real-time intraoperative navigation has become a core challenge. During the operation, tissues deform due to respiration, changes in patient position, or the action of instruments, causing the pre-planned puncture path to lose its precision. This deformation not only affects the spatial parameters of the path, but also places higher demands on the real-time updating of the robot's motion trajectory. If the deformation cannot be accurately captured and the path cannot be dynamically adjusted, the ablation needle may deviate from the target area, resulting in poor treatment effects.

[0004] Therefore, how to capture tissue deformation in real time during the operation and dynamically optimize the puncture path and robot movement trajectory has become a key issue for precise navigation and dynamic matching in percutaneous puncture surgery. Summary of the Invention

[0005] This invention provides a method and system for precise dynamic matching and navigation of percutaneous puncture based on an AI surgical robot. It enables real-time capture of tissue deformation during percutaneous puncture, and further optimizes the puncture path based on the real-time captured tissue deformation state. Based on the real-time generation of the dynamic puncture path, the robot's motion trajectory is adjusted in real time, thereby improving the precision navigation and dynamic matching of percutaneous puncture surgery and achieving the goal of precise puncture.

[0006] This invention provides a method for precise dynamic matching and navigation of percutaneous puncture based on an AI surgical robot, comprising:

[0007] S1, acquire preoperative medical images and generate surgical planning data based on the medical images;

[0008] S2, based on surgical planning data, determine the ablation path and ablation area;

[0009] S3, the puncture operation is performed by the robotic system to complete the ablation treatment of the target area;

[0010] Step S3 further includes: acquiring real-time intraoperative image data multiple times, and dynamically adjusting the ablation path and ablation parameters in real time based on each set of real-time intraoperative image data.

[0011] Preferably, the patient's preoperative imaging data is acquired, and a three-dimensional model of the lesion area is constructed based on the preoperative imaging data. The three-dimensional model of the lesion area is then used to generate initial surgical planning data.

[0012] The spatial parameters of the puncture path are calculated using the initial surgical planning data, and the coverage effect of the target area is verified by simulation. The robot joint motion trajectory corresponding to the puncture path is calculated and the puncture operation is controlled. Real-time tissue images are acquired during the puncture process and registered with the three-dimensional model to determine the tissue deformation state. If tissue deformation exists, the spatial parameters of the puncture path are recalculated based on the registration results and the robot motion trajectory is updated.

[0013] The three-dimensional model is simulated and analyzed. Based on the simulation analysis results, the expansion range of the ablation area is predicted and the initial surgical planning data is adjusted. When multiple needles are punctured, the motion control system is used to synchronously monitor the positioning accuracy of each puncture trajectory and dynamically correct the puncture parameters. After each adjustment of the path parameters, the finite element simulation verification and robot motion trajectory calculation are re-executed until the preset ablation coverage conditions are met.

[0014] Preferably, the patient's preoperative medical image data is acquired, and the lesion location and tissue characteristics are determined based on the medical image data; a three-dimensional model is constructed based on the lesion location and tissue characteristics; and surgical planning data is generated through the three-dimensional model, wherein the surgical planning data is the surgical planning route.

[0015] Preferably, surgical planning parameters are obtained based on the surgical planning route, and the insertion position and angle of the ablation needle are calculated based on these parameters; the insertion position and angle are optimized so that the puncture path can cover the target area;

[0016] It also includes: verifying the coverage effect of the puncture path on the target area.

[0017] Preferably, the spatial coordinates of the robot joints are calculated based on the surgical planning data; the motion trajectory of the robot joints is determined using the spatial coordinates; and the robot is controlled to precisely locate the ablation needle and perform the puncture operation based on the motion trajectory.

[0018] Preferably, real-time tissue images are acquired using an intraoperative image acquisition device, and tissue displacement or deformation information is obtained based on the real-time tissue images; the ablation path and ablation parameters are adjusted according to the tissue displacement or deformation information.

[0019] Preferably, simulation analysis is performed on medical image data, and the ablation effect is predicted based on the simulation results; the predicted ablation effect is used to assist in generating surgical planning data and optimizing the surgical plan.

[0020] Preferably, the trajectory of the ablation needle is controlled by a robot; a single-point multi-needle puncture operation is performed according to the trajectory of the ablation needle; and the puncture accuracy and stability are monitored in real time for the single-point multi-needle puncture operation.

[0021] Preferably, the intraoperative real-time images are registered with the preoperative model and then compared; based on the comparison results, the tissue changes are determined, and the ablation path and ablation parameters are dynamically adjusted according to the tissue change status.

[0022] The present invention also provides a percutaneous puncture precision dynamic matching and navigation system based on an AI surgical robot. The system uses a percutaneous puncture precision dynamic matching and navigation method based on an AI surgical robot to perform dynamic matching and navigation during the percutaneous puncture procedure, and provides intelligent guidance for assisting the percutaneous puncture procedure based on dynamic matching and navigation.

[0023] The working principle and beneficial effects of this invention are as follows:

[0024] This invention provides a method for precise dynamic matching and navigation of percutaneous puncture based on an AI surgical robot, comprising: acquiring preoperative medical images and generating surgical planning data based on the medical images; determining the ablation path and ablation area according to the surgical planning data; performing the puncture operation through the robot system to complete the ablation treatment of the target area; step S3 further includes: acquiring real-time intraoperative image data multiple times, and dynamically adjusting the ablation path and ablation parameters in real time based on each real-time intraoperative image data.

[0025] This invention provides a method for precise dynamic matching and navigation of percutaneous puncture based on an AI surgical robot. It generates puncture plans for different stages of puncture positioning and navigation by collecting and analyzing preoperative and intraoperative patient medical image data. It achieves real-time capture of tissue deformation during percutaneous puncture and further optimizes the puncture path based on the real-time captured tissue deformation state. The real-time generation of the dynamic puncture path allows for real-time adjustment of the robot's trajectory, thereby improving the precision navigation and dynamic matching of percutaneous puncture surgery and achieving precise puncture.

[0026] The present invention also provides a percutaneous puncture precision dynamic matching and navigation system based on an AI surgical robot. The system uses a percutaneous puncture precision dynamic matching and navigation method based on an AI surgical robot to perform dynamic matching and navigation during the percutaneous puncture procedure, and provides intelligent guidance for assisting the percutaneous puncture procedure based on dynamic matching and navigation.

[0027] More specifically, this invention uses preoperative medical images of the patient to construct a three-dimensional model of the lesion area, thereby generating initial surgical planning data, and then using the initial surgical planning data to calculate the puncture path and robot movement trajectory.

[0028] During the procedure, this invention can acquire intraoperative images in real time and register them with a 3D model to determine the state of tissue deformation and dynamically adjust the ablation path and parameters. It utilizes simulation analysis to predict the expansion range of the ablation area, thereby achieving real-time optimization of the surgical plan; and it achieves higher puncture accuracy through precise execution of the puncture operation via a robotic system and real-time monitoring.

[0029] This invention enables intelligent surgical planning and precise surgical execution, improving the safety and effectiveness of tumor ablation surgery and providing technical support for personalized treatment of complex cases.

[0030] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0031] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0032] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0033] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0034] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0035] according to Figure 1 As shown, this embodiment of the invention provides a method for precise dynamic matching and navigation of percutaneous puncture based on an AI surgical robot, including:

[0036] S1, acquire preoperative medical images and generate surgical planning data based on the medical images;

[0037] S2, based on surgical planning data, determine the ablation path and ablation area;

[0038] S3, the puncture operation is performed by the robotic system to complete the ablation treatment of the target area;

[0039] Step S3 further includes: acquiring real-time intraoperative image data multiple times, and dynamically adjusting the ablation path and ablation parameters in real time based on each set of real-time intraoperative image data.

[0040] Step S1 specifically involves: acquiring the patient's preoperative imaging data, constructing a three-dimensional model of the lesion area based on the preoperative imaging data, and generating initial surgical planning data using the three-dimensional model of the lesion area;

[0041] Step S2 specifically involves: calculating the spatial parameters of the puncture path using the initial surgical planning data, and verifying the coverage effect of the target area through simulation; calculating the robot joint motion trajectory corresponding to the puncture path and controlling the execution of the puncture operation; acquiring real-time tissue images during the puncture process and registering them with the 3D model to determine the tissue deformation state; if tissue deformation exists, recalculating the spatial parameters of the puncture path and updating the robot motion trajectory based on the registration results.

[0042] The three-dimensional model is simulated and analyzed. Based on the simulation analysis results, the expansion range of the ablation area is predicted and the initial surgical planning data is adjusted. When multiple needles are used for puncture, the motion control system is used to synchronously monitor the positioning accuracy of each puncture trajectory and dynamically correct the puncture parameters. After each adjustment of the path parameters, the finite element simulation verification and robot motion trajectory calculation are re-executed until the preset ablation coverage conditions are met.

[0043] This invention provides a method for precise dynamic matching and navigation of percutaneous puncture based on an AI surgical robot. It generates puncture plans for different stages of puncture positioning and navigation by collecting and analyzing preoperative and intraoperative patient medical image data. It achieves real-time capture of tissue deformation during percutaneous puncture and further optimizes the puncture path based on the real-time captured tissue deformation state. The real-time generation of the dynamic puncture path allows for real-time adjustment of the robot's trajectory, thereby improving the precision navigation and dynamic matching of percutaneous puncture surgery and achieving precise puncture.

[0044] More specifically, this invention uses preoperative medical images of the patient to construct a three-dimensional model of the lesion area, thereby generating initial surgical planning data, and then using the initial surgical planning data to calculate the puncture path and robot movement trajectory.

[0045] During the procedure, this invention can acquire intraoperative images in real time and register them with a 3D model to determine the state of tissue deformation and dynamically adjust the ablation path and parameters. It utilizes simulation analysis to predict the expansion range of the ablation area, thereby achieving real-time optimization of the surgical plan; and it achieves higher puncture accuracy through precise execution of the puncture operation via a robotic system and real-time monitoring.

[0046] This invention enables intelligent surgical planning and precise surgical execution, improving the safety and effectiveness of tumor ablation surgery and providing technical support for personalized treatment of complex cases.

[0047] The present invention also provides a percutaneous puncture precision dynamic matching and navigation system based on an AI surgical robot. The system uses a percutaneous puncture precision dynamic matching and navigation method based on an AI surgical robot to perform dynamic matching and navigation during the percutaneous puncture procedure, and provides intelligent guidance for assisting the percutaneous puncture procedure based on dynamic matching and navigation.

[0048] More specifically, this invention uses preoperative medical images of the patient to construct a three-dimensional model of the lesion area, thereby generating initial surgical planning data, and then using the initial surgical planning data to calculate the puncture path and robot movement trajectory.

[0049] During the procedure, this invention can acquire intraoperative images in real time and register them with a 3D model to determine the state of tissue deformation and dynamically adjust the ablation path and parameters. It utilizes simulation analysis to predict the expansion range of the ablation area, thereby achieving real-time optimization of the surgical plan; and it achieves higher puncture accuracy through precise execution of the puncture operation via a robotic system and real-time monitoring.

[0050] This invention enables intelligent surgical planning and precise surgical execution, improving the safety and effectiveness of tumor ablation surgery and providing technical support for personalized treatment of complex cases.

[0051] In one embodiment, such as Figure 1 As shown, this invention provides a method for precise dynamic matching and navigation of percutaneous puncture based on an AI surgical robot, comprising:

[0052] A 3D model of the lesion area is constructed using the patient's preoperative imaging data, and initial surgical planning data is generated. Spatial parameters of the puncture path are calculated based on the initial surgical planning data to verify the coverage effect of the target area. The robot joint motion trajectory corresponding to the puncture path is calculated using an inverse kinematics algorithm to control the execution of the puncture operation. Real-time tissue images are acquired and registered with the 3D model to determine the tissue deformation state. If tissue deformation exists, the spatial parameters of the puncture path are recalculated based on the registration results, and the robot motion trajectory is updated. Simulation analysis is performed on the 3D model to predict the ablation area expansion range and adjust the initial surgical planning data. The puncture parameters are dynamically corrected by monitoring the puncture trajectory positioning accuracy. Verification and trajectory calculation are repeated until the preset ablation coverage conditions are met.

[0053] In this scheme, preoperative patient imaging data is acquired, and a 3D model of the lesion area is constructed using an image segmentation algorithm to obtain initial surgical planning data. A geometric optimization algorithm is then used to calculate the spatial parameters of the puncture path based on the initial surgical planning data. Next, the coverage effect on the target area is verified to obtain a path parameter set. The robot joint motion trajectory is generated using the path parameter set, and this trajectory is used to perform the puncture operation, resulting in the actual puncture trajectory.

[0054] Real-time tissue images are acquired and registered with the 3D model of the lesion. If the registration deviation exceeds a preset threshold, tissue deformation is determined, and deformation state data is obtained. The spatial parameters of the puncture path are recalculated using the deformation state data, and then the robot's motion trajectory is updated to obtain optimized trajectory data.

[0055] Finite element simulation analysis was used to predict the expansion range of the ablation area in the three-dimensional model of the lesion. Based on the expansion range, the initial surgical planning data was further optimized to obtain the adjusted planning data.

[0056] The positioning accuracy of the monitored and optimized trajectory data is determined. If the deviation exceeds the preset threshold, the puncture parameters are dynamically adjusted and the verification is repeated until the ablation coverage conditions are met, and the final puncture plan is determined.

[0057] This embodiment provides the following examples to further illustrate the above technical solution: In the stage of acquiring the patient's preoperative image data, a 1mm slice-thickness CT scan is used to acquire a 512×512 pixel DICOM format image. The liver lesion area is extracted using a threshold segmentation algorithm to create a three-dimensional model. Based on the established three-dimensional model, the system automatically generates an initial surgical plan containing 5 ablation target points, with each target point having a diameter of 2-2.5mm and a number of 6-8 targets. The overlapping area is optimized using a spherical stacking algorithm, or the spatial arrangement of the target points is optimized using a greedy algorithm, so that the ablation area covers more than 90% of the lesion area.

[0058] The puncture path was calculated by detecting collisions with blood vessels and bones using ray projection, with a safe path distance set at ≥5 mm. A respiratory motion model was introduced for Monte Carlo simulation, for example, using a sine function to simulate diaphragmatic movement. After 500 Monte Carlo simulations, the target area coverage was achieved by path sampling.

[0059] In this scheme, inverse kinematics calculations are performed using the DH parameter model. Taking a 6-DOF robotic arm as an example, the joint angles are calculated using the Jacobian matrix pseudo-inverse method. The trajectory is calculated by interpolating the motion angle using a 5th-order B-spline curve, thereby obtaining the puncture motion trajectory. The positional error of the puncture motion trajectory can be controlled within 0.1 mm.

[0060] During the operation, the lesion site data is acquired in real time through imaging. Based on this data, a lesion area model is built. The lesion area model is registered with the preoperative model. When liver displacement is detected to exceed 3 mm, a path replanning command is triggered.

[0061] Finite element analysis based on a biomechanical model was used to correct the tensile modulus, distinguish normal liver tissue from lesion areas, and predict tissue deformation based on this, so as to achieve real-time updates of the puncture path inclination and depth. When registering the lesion area model and the preoperative model, the lesion coverage rate was predicted. When the predicted coverage rate was less than 85%, an auxiliary target point was automatically added. When the cumulative error exceeded the preset threshold, closed-loop control correction was triggered. During the process, iterative optimization was continuously carried out until the overlap rate between the ablation prediction volume of all targets and the lesion reached ≥95% covering the core area of ​​the lesion and ≥90% of the 5mm safety margin, thus meeting the ablation coverage conditions.

[0062] In one embodiment, such as Figure 1 As shown, this invention provides a method for precise dynamic matching and navigation of percutaneous puncture based on an AI surgical robot, comprising:

[0063] Preoperative imaging data of patients is acquired, segmented, and lesion area features are extracted. A three-dimensional model is constructed based on the lesion area features. The boundary and volume of the lesion area are determined according to the three-dimensional model. Initial surgical planning data is generated using the boundary and volume. The initial surgical planning data includes the puncture point location and puncture depth.

[0064] In this scheme, preoperative image data is acquired from medical imaging equipment, parsed using DICOM format, and a first image dataset is obtained; the first image dataset is preprocessed by using anisotropic diffusion filtering or nonlocal mean filtering to obtain a second image dataset.

[0065] Mean filtering denoising reduces noise by averaging the neighboring pixel values, but it can blur image edges, especially for low-contrast soft tissue lesions (such as liver tumors), which may lead to a decrease in segmentation accuracy. Therefore, anisotropic diffusion filtering is preferred for filtering the second image dataset.

[0066] If the resolution of the second image dataset meets the preset threshold, the second image dataset is segmented using the U-Net algorithm. Lesion region features are extracted from the segmented second image dataset to obtain a lesion feature set. If the resolution of the second image dataset does not meet the preset threshold, the resolution is adjusted and the dataset is re-segmented to obtain the lesion feature set.

[0067] A three-dimensional model of the lesion feature set is constructed using stereomicroscopy to generate the first three-dimensional model. The boundary of the lesion region is determined using a boundary tracing algorithm, and the boundary dataset is extracted. The volume data is obtained by accumulating voxels in the first three-dimensional model using a threshold-weighted integral method or based on a segmentation probability map. Based on the boundary dataset and volume data, the puncture point location is determined using a geometric optimization algorithm, and the puncture depth data is calculated. Initial surgical planning data is generated based on the puncture point location and puncture depth data.

[0068] Based on coordinate mapping, a puncture path is generated from the initial surgical planning data, resulting in a surgical path dataset. In actual clinical applications, the mapping from the image coordinate system to the surgical robot coordinate system relies on external marker points. If the positioning error of the marker points is 0.5mm, the cumulative error after multiple coordinate transformations may exceed 1mm. To avoid the above problems, this embodiment utilizes optical navigation and real-time intraoperative image registration to dynamically correct coordinate system offsets, thereby avoiding the problem of spatial registration error accumulation in coordinate mapping technology.

[0069] More specifically, preoperative image data of patients is acquired from medical imaging equipment. The scan layer thickness of the preoperative image data is set to 1 mm, the resolution is 512×512 pixels, and a DICOM format image file is generated.

[0070] When segmenting preoperative image data, deep learning algorithms such as U-Net networks are used. The training set contains 5,000 labeled lesion regions, and the model achieves an accuracy of 95%. The lesion regions are segmented and their features, such as shape, density, and edge information, are extracted. At the same time, the Dice coefficient or Hausdorff distance is used to evaluate the quality of boundary segmentation; for example, the boundary error is ≤0.3mm.

[0071] When constructing a 3D model based on the features of the lesion area, the MarchingCubes algorithm is used to reconstruct the 2D slice data into 3D volume data, with an isotropic accuracy of 0.5 mm. Alternatively, super-resolution reconstruction can be used to improve the Z-axis resolution, ensuring that the geometric shape of the lesion area is accurately represented.

[0072] When determining the boundaries and volume of the lesion region based on the 3D model, a region growing algorithm is used to tighten the threshold according to the lesion type, or to combine it with multi-sequence MRI-assisted segmentation; for example, the threshold range is set to 30-80 HU, the calculated lesion volume is 15.6 cubic centimeters, and the boundary error is less than 0.5 millimeters.

[0073] When generating initial surgical planning data based on boundaries and volume, a path planning algorithm is used to determine the puncture point location, for example: 2.2 cm from the skin surface, with a puncture depth of 5.8 cm; the planned path avoids important blood vessels and nerves; for example: setting dynamic safety boundaries, such as ≥5 mm around blood vessels and ≥8 mm of elastic buffer zone during the respiratory cycle.

[0074] In one embodiment, such as Figure 1 As shown, this invention provides a method for precise dynamic matching and navigation of percutaneous puncture based on an AI surgical robot, comprising:

[0075] The starting and ending points of the puncture path are determined based on the initial surgical planning data; the spatial parameters of the puncture path, including the path angle and length, are calculated; the puncture process is simulated using finite element simulation; the coverage of the target area is determined based on the simulation results; if the coverage is lower than a preset threshold, the spatial parameters of the puncture path are adjusted.

[0076] In this scheme, the coordinates of the start and end points of the puncture path are obtained from the initial surgical planning data to determine the initial geometric description of the puncture path; then, based on the coordinates of the start and end points of the puncture path, the path angle parameter and the path length parameter are calculated to obtain the spatial parameter description of the puncture path, where the path angle parameter is the angle between the path and the reference plane, and the path length parameter is the Euclidean distance from the start point to the end point.

[0077] Based on spatial parameter description, the finite element method (FEM) is used to simulate the puncture process and obtain the distribution of the puncture path's effect within the target area. Specifically, the FEM simulation discretizes the target area into a mesh and calculates the impact of the puncture path on each mesh element. Based on the simulation results, the coverage rate of the target area is calculated, providing a quantitative description of the target area's coverage effect.

[0078] The formula for calculating coverage is:

[0079]

[0080] in This indicates the area effectively covered by the puncture path. This indicates the total area of ​​the target region.

[0081] If the coverage rate C is lower than the preset threshold, the path angle parameters and path length parameters are optimized by the gradient descent algorithm, the spatial parameters of the puncture path are adjusted, and a new spatial parameter description is obtained. At the same time, the parameters are updated iteratively using the gradient descent algorithm.

[0082] Based on the adjusted spatial parameters, finite element simulation and coverage calculation are repeated to obtain the updated target area coverage. If the updated coverage is still lower than the preset threshold, the spatial parameter adjustment and simulation process is iteratively executed until the coverage meets the preset threshold, thereby obtaining the optimized puncture path spatial parameters.

[0083] This embodiment provides the following example to further illustrate the above technical solution: Based on the initial surgical planning data, the starting point of the puncture path is first determined as the patient's body surface coordinates. The endpoint is the coordinates of the tumor center. .

[0084] The angle of the puncture path is calculated using spatial geometry algorithms. 45°, length

[0085]

[0086] .

[0087] In this example, to ensure accurate positioning in three-dimensional space, multiple geometric parameters are introduced for the puncture angle parameter, such as the pitch angle. and yaw angle By jointly defining the three-dimensional path direction, the reference plane, such as the transverse plane or sagittal plane, can be better identified in actual clinical applications, thereby improving the accuracy of the puncture position.

[0088] For multi-angle calculations, please refer to the following:

[0089]

[0090] Next, finite element simulation was used to simulate the puncture process; first, the diameter of the puncture needle was set to 2 mm, the puncture speed to 5 mm / s, and the simulation time to 15 seconds.

[0091] During the simulation, the interaction forces between the puncture needle and surrounding tissues, as well as the stress distribution along the puncture path, were recorded. Simultaneously, a biomechanical multi-field coupling model was used to simulate the ablation process, and the parameters were calibrated.

[0092] Based on the simulation results, the coverage rate of the target area is calculated. Assuming the tumor volume is 1000 cubic millimeters and the tumor volume covered by the puncture path is 800 cubic millimeters, the coverage rate is 80%. If the preset threshold is 85%, the coverage rate is lower than the threshold. In this case, the spatial parameters of the puncture path need to be adjusted to meet the coverage rate.

[0093] Specifically, define spatial coverage weights (e.g., lesion core weight > edge weight), and use weighted coverage rate:

[0094]

[0095] in, Weighted coverage ratio; For the first The spatial coverage weight of each region, such as the weight of the lesion core being greater than the weight of the edge, reflects the importance of different regions. For the first The region, the first The effective coverage volume of each region, i.e. the actual tumor volume covered during the ablation process (dynamically calculated using the biothermal equation). For the first The region, the first The total volume of the tumor in each region, such as the total tumor volume. .

[0096] Among them, the ablation zone expansion is simulated by combining biothermal equations (such as the Pennes equation) to simulate the actual tumor volume covered during the ablation process, thereby achieving the purpose of dynamically calculating the effective coverage volume.

[0097] Adjust the puncture path angle to The length was adjusted to Then, a new finite element simulation was performed, and the new coverage rate was calculated to be 85.5%. At this point, the new coverage rate meets the preset threshold requirement.

[0098] In this example, the puncture path angle is directly optimized. and puncture path length At this time, the location of the puncture endpoint may be changed (e.g., the length of the puncture path may be adjusted). This could lead to the needle tip either not reaching the center of the lesion or exceeding the center of the lesion, resulting in deviation from the treatment target. Based on this, this embodiment also proposes the following solutions for further optimization:

[0099] The endpoint of the puncture is fixed at the center of the lesion, and the location of the needle entry point or the direction of the path are optimized. Design multi-objective optimization functions, such as optimizing the needle entry point position or path direction by calculating the comprehensive optimization objective function E:

[0100] ;

[0101] in, To comprehensively optimize the objective function, by minimizing Achieve the optimal puncture path; All are weighting coefficients; C is the coverage rate; (1−C) represents the penalty term for non-coverage; safety risk is the quantification of the risk of damage to surrounding healthy tissues by the puncture path, such as the quantification of the risk of damage to critical structures such as blood vessels and nerves. Robotic arm motion cost is the quantification of the motion complexity of the robotic arm executing the puncture path, such as the quantification of the motion complexity corresponding to equipment energy consumption, joint rotation angle, or path smoothness.

[0102] In one embodiment, such as Figure 1 As shown, the present invention provides a method for precise dynamic matching and navigation of percutaneous puncture based on an AI surgical robot, comprising: determining the pose of the robot's end effector according to the spatial parameters of the puncture path; calculating the angles of each joint corresponding to the pose; generating the robot's motion trajectory based on the joint angles; driving the robot to execute the motion trajectory through a motion control system; and monitoring the robot's execution status in real time and recording the puncture operation data.

[0103] In this embodiment, the specific scheme is as follows: Spatial parameters of the puncture path are obtained, the target pose of the end effector is calculated, and pose parameters are obtained. Based on the pose parameters, the angles of each joint are calculated using an inverse kinematics algorithm, and then the joint angles are obtained. The robot's motion trajectory is generated based on the joint angles, and trajectory data is generated, where the trajectory data is the trajectory execution command. The robot is driven to execute the trajectory data through a motion control system; the robot's execution status is monitored in real time based on the executed trajectory data, and sensor feedback data is acquired to obtain corresponding status information; if the status information deviates from a preset threshold, the joint angles are adjusted and the trajectory data is updated.

[0104] This embodiment provides the following example to further illustrate the above technical solution: Based on the puncture path spatial parameters, the pose of the robot's end effector is first determined. For example, in three-dimensional space, the position coordinates of the end effector are set as follows: Millimeters, attitude angle is a quaternion rotation matrix or rotation matrix.

[0105] The inverse kinematics algorithm is used to calculate the joint angles corresponding to this pose. Assuming the robot has 6 joints, the joint angles are calculated using inverse kinematics. The robot's motion trajectory is generated based on the joint angles mentioned above. To ensure the trajectory is smooth and continuous, a cubic spline interpolation algorithm is used to further optimize the smoothness of the robot's motion trajectory. In this scheme, the inverse kinematics algorithm can be calculated using damped least squares (DLS), which can suppress the influence of singularities during the calculation. In addition, to ensure puncture safety, joint limit checks and self-collision detection are verified in real time after the inverse kinematics solution to prevent the robot from exceeding its physical limits during the movement.

[0106] In one embodiment, such as Figure 1 As shown, this invention provides a method for precise dynamic matching and navigation of percutaneous puncture based on an AI surgical robot, comprising:

[0107] Acquire real-time tissue images and extract tissue feature points from them; register the feature points with corresponding points in the 3D model; calculate tissue deformation based on the registration results; if the deformation exceeds a preset threshold, it is determined that tissue deformation exists.

[0108] In this scheme, real-time tissue images are acquired using intraoperative imaging equipment to generate first image data; the first image data is preprocessed using denoising and edge enhancement algorithms to obtain second image data; feature points are extracted from the second image data, and three-dimensional feature points are further extracted; the positions of key points are calculated based on voxel gradients. And the description vector, further generating a feature point set. Feature point set Each point contains position coordinates. and description vector. The feature point set Predefined point set in the 3D model Registration is performed, and the transformation matrix is ​​calculated using the iterative nearest-point algorithm. The registration result is obtained; based on the transformation matrix in the registration result... Calculate organizational deformation And obtain the numerical values ​​of the deformation variables.

[0109] Where, transformation matrix For: feature point set to the corresponding point set Spatial mapping;

[0110] Organizational deformation For: feature point set Each point in the transformation matrix Transformed and corresponding point set The mean Euclidean distance between corresponding points.

[0111] If organizational deformation Exceeding the preset threshold If the deformation is detected, a deformation status identifier is generated.

[0112] If organizational deformation Less than or equal to the preset threshold If no significant deformation is found, a normal state identifier is generated.

[0113] Based on the deformation state identifiers described above, the 3D model parameters are updated, and the geometric parameters of the corresponding regions in the model are adjusted using the deformation state identifiers to obtain the updated 3D model data.

[0114] This embodiment provides the following examples to further illustrate the above technical solution: During the first image data preprocessing, speckle noise is suppressed by anisotropic diffusion filtering (iteration 4 times, conduction coefficient k=0.05); then, contrast-limited histogram equalization (CLAHE) is used to enhance image edges, thereby further ensuring image clarity; feature points are extracted using the 3D-SIFT algorithm, with a contrast threshold of 0.04 and an edge threshold of 10, and approximately 200-300 stable feature points can be extracted from each frame.

[0115] Furthermore, during registration, the iterative nearest neighbor algorithm is used for iterative optimization. The initial matching is accelerated by the KD-Tree for nearest neighbor search. The maximum number of iterations is set to 50, the convergence threshold is 0.001mm, and the final registration error is controlled within 0.5mm.

[0116] Deformation analysis is performed on the registered feature point cloud. When the deformation of a local area exceeds the preset threshold of 2 mm, the system automatically triggers a deformation alarm and classifies the deformation type. For example, when the displacement of a certain area in the right lobe of the liver reaches 2.3 mm, a deformation alarm is triggered. The alarm labels for the deformation types are: elastic deformation is labeled as the first warning, and permanent damage is labeled as the second alarm.

[0117] In one embodiment, such as Figure 1 As shown, this invention provides a method for precise dynamic matching and navigation of percutaneous puncture based on an AI surgical robot, comprising:

[0118] The location of the lesion area after tissue deformation is determined based on the registration results; the start and end points of the puncture path are recalculated and the spatial parameters of the puncture path are updated, including the path angle and length; the pose of the robot end effector is recalculated based on the updated spatial parameters of the puncture path; and the robot joint motion trajectory is updated using inverse kinematics algorithms.

[0119] In this scheme, registration results are obtained, and tissue deformation data is extracted using image processing technology to determine the spatial coordinates of the lesion area after deformation, thus obtaining the lesion area coordinates. Based on the lesion area coordinates and a geometric algorithm, the starting and ending coordinates of the puncture path are calculated, resulting in the path start and end coordinates. The vector direction and distance are calculated using these coordinates, and the path angle and length are then updated to obtain spatial parameters. Based on these spatial parameters, the pose of the robot's end effector is calculated, resulting in the end effector pose. If the end effector pose exceeds the robot's workspace, the joint angles are adjusted using an inverse kinematics algorithm to generate joint motion trajectories, resulting in the joint motion trajectory. Based on the joint motion trajectory, a motion planning algorithm is used to optimize the robot's motion path, resulting in the optimized robot trajectory. Based on the optimized robot trajectory, control commands are generated, and the robot's motion parameters are updated based on these commands, resulting in the final motion parameters.

[0120] This embodiment provides the following example to further illustrate the above technical solution: In the registration result processing stage, the tissue deformation field is first calculated using a two-way registration algorithm, and the preoperative CT coordinate system and the intraoperative ultrasound coordinate system are spatially mapped. Assume that the coordinates of the lesion center in the preoperative CT are... After registration, it was found that the lesion had shifted to [location] in the intraoperative coordinate system. The deformation vector is .

[0121] In actual clinical practice, this example may involve non-uniform deformation of other tissues along the puncture path, such as skin, fascia, or other flexible or resilient tissues. Therefore, improvements to the aforementioned technical issues could be considered in actual clinical practice:

[0122] Sampling multiple control points along the puncture path, such as sampling one point every 10 mm; updating the path segment by segment based on the local deformation field; and combining the finite element deformation model to predict the tissue displacement distribution along the path, and using the predicted tissue displacement distribution to plan the puncture path, is more conducive to the success rate of puncture positioning.

[0123] The puncture path parameters were recalculated using spatial geometric transformations, with the original path starting from... Adjusted to The endpoint is updated synchronously to the deformed coordinates. The original path start point is set as the skin entry point; the endpoint is set as the lesion target point.

[0124] Furthermore, the path angle is recalculated using vector operations, employing direction vectors. Calculations are performed to obtain the new pitch angle. Yaw angle The path length is updated to 55.7 mm using the Euclidean distance formula. In this example, the specific values ​​are for calculation purposes only. In actual clinical applications, or during puncture experiments on experimental models, anatomical feasibility checks and limitations on the puncture angle range should be performed before path planning to reduce the risk of complications. Anatomical feasibility checks involve CT 3D reconstruction of the rib and organ positions; an example of a puncture angle range is 40°≤θ≤70°.

[0125] In this embodiment, before surgery, reachability analysis is used to verify the target pose and determine whether the robot's workspace is within the preset working range. Inverse kinematics calculations are then performed using a DH parameter model. This model integrates an obstacle distance field to identify and avoid physical obstructions within the robot's movement range during the surgery.

[0126] In one embodiment, such as Figure 1 As shown, this invention provides a method for precise dynamic matching and navigation of percutaneous puncture based on an AI surgical robot, comprising:

[0127] A finite element simulation model is established based on the 3D model, and the ablation process is simulated to obtain the expansion range of the ablation area; the coverage rate between the ablation area and the lesion area is determined based on the expansion range; if the coverage rate is lower than a preset threshold, the puncture point position and ablation parameters in the initial surgical planning data are adjusted; and the initial surgical planning data is updated.

[0128] In this scheme, geometric structure data is obtained from a 3D model to construct a finite element simulation model. A discretized mesh is generated using a mesh generation algorithm, and simulation boundary conditions are determined to obtain the finite element simulation model. The ablation process is simulated based on the finite element simulation model. The temperature distribution is calculated using the heat conduction equation to determine the ablation zone expansion range. The coverage ratio between the ablation zone and the lesion zone is calculated using the expansion range and the geometric data of the lesion zone. The overlap volume ratio is calculated using stereoscopic methods to obtain coverage data; alternatively, the overlap volume ratio between the ablation zone and the lesion zone is calculated using 3D geometric Boolean operations to obtain coverage data. If the coverage ratio is lower than a preset threshold, the puncture point position and ablation parameters are adjusted. The puncture point coordinates and ablation power are optimized using gradient descent or particle swarm optimization (PSO) algorithms to determine the adjusted puncture point position and ablation parameters. Based on the adjusted puncture point position and ablation parameters, the surgical planning data is updated, generating a new surgical planning dataset to obtain the updated surgical planning data. New puncture point locations and ablation parameters are obtained from the updated surgical planning data. A new finite element simulation model is then reconstructed, and a new discretized mesh is generated using a mesh generation algorithm. New simulation boundary conditions are determined, resulting in a new finite element simulation model. Based on the new finite element simulation model, the ablation process is re-simulated. The new temperature distribution is calculated using the heat conduction equation, and the new ablation region's expansion range and coverage are determined, resulting in optimized surgical planning data.

[0129] This embodiment provides the following example to further illustrate the above technical solution: Based on the patient's liver CT scan data, a three-dimensional model containing the tumor lesion is reconstructed using Mimics software, with the model accuracy set to 0.5 mm voxel resolution. A finite element simulation model is established using ANSYS, with the liver tissue thermal conductivity set to 0.5 W / (m·K), specific heat capacity to be 3600 J / (kg·K), and blood perfusion rate to be 0.03 s⁻¹. In the radiofrequency ablation simulation, the temperature field distribution is calculated using the Pennes biothermal equation, with the electrode power set to 50 W and the action time to be 300 seconds. The tissue damage range is calculated using the Arrhenius damage integral model, and a tissue damage range value greater than 1 is considered an ablation zone.

[0130] At this point, the simulation results showed that the maximum diameter of the ablation zone was 3.2 cm, with a matching degree of only 78% with the preoperatively planned 2.8 cm lesion area, which was lower than the clinically required threshold. The system then automatically initiated an optimization algorithm, using gradient descent or particle swarm optimization (PSO) to iteratively adjust the puncture point coordinates in 3D space. Each iteration calculated the new ablation coverage and safety boundary. When the puncture point was moved 4 mm cephalad and the power was increased to 55W, the simulation showed that the ablation zone expanded to 3.5 cm, and the coverage increased to 93%. The system automatically updated the surgical planning data, increasing the minimum distance between the new puncture path and intrahepatic vessels from 1.2 mm to 2.1 mm, while predicting that the safe distance between the ablation zone and the bile duct would remain above 2.5 mm. The entire optimization process took 2 hours and 17 minutes, completing 36 parameter combination evaluations, and the final planning data was directly transmitted to the navigation system.

[0131] In one embodiment, such as Figure 1 As shown, this invention provides a method for precise dynamic matching and navigation of percutaneous puncture based on an AI surgical robot, comprising:

[0132] The motion control system acquires real-time position data of each puncture trajectory; calculates the deviation between the real-time position data and the target trajectory; if the deviation exceeds a preset threshold, it adjusts the puncture parameters, including puncture speed and angle; it synchronously monitors the trajectory deviation of the multi-needle puncture mode; and updates the robot motion control commands according to the adjusted puncture parameters.

[0133] In this scheme, the motion control system acquires real-time position data of the puncture needle and uses sensors to collect three-dimensional coordinate information to obtain a dynamic position dataset of the puncture trajectory. The dynamic position dataset is compared with a preset target trajectory, and the real-time deviation is calculated using Euclidean distance to obtain the deviation value. If the deviation value exceeds a preset threshold, the puncture speed v and angle θ are adjusted using a gradient descent algorithm or a PID control algorithm to obtain the adjusted puncture parameters. A parallel processing module synchronously monitors the trajectory deviations of multiple punctures and uses timestamps to align the data of each needle to obtain a multi-needle deviation dataset.

[0134] The Euclidean distance calculation formula is as follows:

[0135] ,

[0136] in, For real-time coordinates, The target coordinates;

[0137] Furthermore, based on the multi-needle deviation dataset, it is determined whether any needle deviation exceeds a preset threshold. If so, the adjustment parameters of each needle are merged using a weighted average method to obtain a unified parameter adjustment value.

[0138] The formula for the weighted average algorithm is as follows:

[0139] ,

[0140] in, For single-needle parameters, As weight,

[0141] Furthermore, when the multi-needle trajectory deviation exceeds a threshold, weights are assigned based on the reciprocal of the deviation of each needle:

[0142] ,

[0143] The parameters after fusion are:

[0144] .

[0145] Based on unified parameter adjustment values, robot motion control commands are generated. Joint angles are calculated using inverse kinematics algorithms to obtain updated motion commands. The updated motion commands are transmitted to the motion control system through a real-time feedback loop. Finally, closed-loop control is used to adjust the puncture needle trajectory to obtain optimized puncture trajectory data.

[0146] In this embodiment, to account for system latency, feedforward control and timestamp prediction are taken into account, thereby achieving the purpose of compensating for the latency between the sensing device and the execution device.

[0147] This embodiment provides the following examples to further illustrate the above technical solution: The motion control system acquires real-time position data of each puncture trajectory, and uses an optical positioning system or electromagnetic sensor for real-time monitoring to ensure that the position data accuracy reaches 0.1 mm; the deviation between the real-time position data and the target trajectory is calculated, and the deviation is calculated using the Euclidean distance formula; if the deviation exceeds a preset threshold of 0.5 mm, the puncture parameters, including puncture speed and angle, are adjusted dynamically using a gradient descent algorithm or a PID control algorithm;

[0148] The speed adjustment formula is as follows: The angle adjustment formula is: ,

[0149] in and Given the initial velocity and angle, and This is for adjusting the coefficient.

[0150] The system synchronously monitors the trajectory deviations in the multi-needle puncture mode and uses multi-threading technology to process the trajectory data in parallel, ensuring real-time performance and accuracy. Based on the adjusted puncture parameters, the robot motion control commands are updated and sent to each actuator via the CAN bus, ensuring the synchronization and precision of the robot's motion control. Ultimately, this achieves high-precision positioning and dynamic correction of the puncture trajectory.

[0151] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A percutaneous puncture precision dynamic matching and navigation system based on an AI surgical robot, characterized in that, The system utilizes a precise dynamic matching and navigation method for percutaneous puncture based on an AI surgical robot to perform dynamic matching and navigation during the percutaneous puncture procedure, and provides intelligent guidance for assisting the percutaneous puncture procedure based on dynamic matching and navigation. The method includes: S1, acquire preoperative medical images and generate surgical planning data based on the medical images; S2, based on surgical planning data, determine the ablation path and ablation area; S3, the puncture operation is performed by the robotic system to complete the ablation treatment of the target area; Step S3 further includes: acquiring real-time intraoperative image data multiple times, and dynamically adjusting the ablation path and ablation parameters in real time based on each real-time intraoperative image data. Acquire the patient's preoperative imaging data, construct a three-dimensional model of the lesion area based on the preoperative imaging data, and use the three-dimensional model of the lesion area to generate initial surgical planning data; The spatial parameters of the puncture path are calculated using the initial surgical planning data, and the coverage effect of the target area is verified by simulation. The robot joint motion trajectory corresponding to the puncture path is calculated and the puncture operation is controlled. Real-time tissue images are acquired during the puncture process and registered with the three-dimensional model to determine the tissue deformation state. If tissue deformation exists, the spatial parameters of the puncture path are recalculated based on the registration results and the robot motion trajectory is updated. The three-dimensional model is simulated and analyzed. Based on the simulation analysis results, the expansion range of the ablation area is predicted and the initial surgical planning data is adjusted. When multiple needles are punctured, the motion control system is used to synchronously monitor the positioning accuracy of each puncture trajectory and dynamically correct the puncture parameters. After each adjustment of the path parameters, the finite element simulation verification and robot motion trajectory calculation are re-executed until the preset ablation coverage conditions are met. The starting and ending points of the puncture path are determined based on the initial surgical planning data; the spatial parameters of the puncture path, including the path angle and length, are calculated; the puncture process is simulated using finite element simulation; the coverage rate of the target area is determined based on the simulation results; if the coverage rate is lower than a preset threshold, the spatial parameters of the puncture path are adjusted. The endpoint of the puncture is fixed at the center of the lesion, and the location of the needle entry point or the direction of the path are optimized. Design a multi-objective optimization function, and optimize the needle entry point position or path direction by calculating the comprehensive optimization objective function E: ； in, To comprehensively optimize the objective function, by minimizing Achieve the optimal puncture path; All are weighting coefficients; C is the coverage rate; (1−C) represents the penalty term for non-coverage; safety risk is the risk of damage to surrounding healthy tissues quantified by the puncture path; The motion control system acquires real-time position data of each puncture trajectory; calculates the deviation between the real-time position data and the target trajectory; if the deviation exceeds a preset threshold, it adjusts the puncture parameters, including puncture speed and angle; it synchronously monitors the trajectory deviation of the multi-needle puncture mode; and updates the robot motion control commands according to the adjusted puncture parameters.

2. The percutaneous puncture precision dynamic matching and navigation system based on an AI surgical robot as described in claim 1, characterized in that, Acquire the patient's preoperative medical image data, determine the lesion location and tissue characteristics based on the medical image data; construct a three-dimensional model based on the lesion location and tissue characteristics; Surgical planning data is generated from a 3D model, where the surgical planning data is the surgical route.

3. The percutaneous puncture precision dynamic matching and navigation system based on an AI surgical robot as described in claim 2, characterized in that, Based on the surgical planning route, surgical planning parameters are obtained, and the insertion position and angle of the ablation needle are calculated based on these parameters. The insertion position and angle are then optimized so that the puncture path can cover the target area. It also includes: verifying the coverage effect of the puncture path on the target area.

4. The percutaneous puncture precision dynamic matching and navigation system based on an AI surgical robot as described in claim 3, characterized in that, The spatial coordinates of the robot joints are calculated based on the surgical planning data; the motion trajectory of the robot joints is determined using the spatial coordinates; and the robot is controlled to precisely locate the ablation needle and perform the puncture operation based on the motion trajectory.

5. The percutaneous puncture precision dynamic matching and navigation system based on an AI surgical robot as described in claim 4, characterized in that, Real-time tissue images are acquired using intraoperative image acquisition equipment. Based on these images, tissue displacement or deformation information is obtained. The ablation path and parameters are then adjusted according to the tissue displacement or deformation information.

6. The percutaneous puncture precision dynamic matching and navigation system based on an AI surgical robot as described in claim 5, characterized in that, Simulation analysis is performed on medical image data, and the ablation effect is predicted based on the simulation results. The predicted ablation effect is used to assist in the generation of surgical planning data and the optimization of surgical procedures.

7. The percutaneous puncture precision dynamic matching and navigation system based on an AI surgical robot as described in claim 6, characterized in that, The robot controls the trajectory of the ablation needle; performs single-point multi-needle puncture based on the trajectory of the ablation needle; and monitors the puncture accuracy and stability in real time for single-point multi-needle puncture.

8. The percutaneous puncture precision dynamic matching and navigation system based on an AI surgical robot as described in claim 7, characterized in that, The intraoperative real-time images were registered with the preoperative model and then compared. Based on the comparison results, the tissue changes were determined, and the ablation path and ablation parameters were dynamically adjusted according to the tissue changes.

Images