Robot-assisted execution system and method based on pre-planned puncture scheme

Abstract

The invention discloses a robot-assisted execution system and method based on a pre-planned puncture scheme, and belongs to the technical field of medical robots. The method comprises the following steps: acquiring medical image data including a target area and a forbidden area; for the target area, a discrete candidate trajectory set is generated in a preset candidate trajectory generation cone, and each candidate trajectory is a space path connecting one point in the generation cone and the target area; performing a rigid geometric constraint check on the discrete candidate trajectory set to screen out candidate trajectories having geometric interference with the forbidden area to obtain a preliminarily purified candidate trajectory subset; executing a flexible quality index evaluation on each candidate trajectory in the candidate trajectory subset, and obtaining a comprehensive quality index for each trajectory; selecting a preferred execution track from the subset according to the comprehensive quality index; and the robot execution system is controlled, and puncture operation is completed according to the optimal execution track. The safety and clinical interpretability of a puncture path are improved.

Description

Technical Field

[0001] This invention relates to the field of medical robot technology, and in particular to a robot-assisted execution system and method based on a pre-planned puncture scheme. Background Technology

[0002] Prostate biopsy is a decisive method for diagnosing prostate cancer. This procedure requires precise guidance of the needle to the suspected lesion area within the prostate for tissue sampling, while strictly avoiding critical anatomical structures such as the urethra, rectal wall, and neurovascular bundles. Traditional manual biopsy relies heavily on the physician's experience and spatial awareness, resulting in low positive rates, increased risk of complications, and poor consistency in procedures.

[0003] To overcome these challenges, robot-assisted puncture technology has been introduced into clinical practice. Existing robotic systems typically follow a plan-execute model. Before the procedure, the physician plans the target site and ideal puncture path based on imaging data such as magnetic resonance imaging (MRI). During the procedure, the robot is responsible for precisely moving the puncture guide device to the planned position and orientation to assist the physician in completing the puncture. While this model improves the stability and repeatability of positioning, its core remains a static, open execution process.

[0004] The fundamental limitation of this static planning method lies in the lack of adaptability of the pre-defined single path to complex anatomical environments. While the path planned by the physician may be ideal, it may not be the safest or optimal choice in actual three-dimensional space. For example, a straight path might glide past the edge of an important blood vessel with a very small gap; although it doesn't directly cross it, the potential risk is extremely high. Furthermore, when multiple possible needle insertion angles exist, systematically and objectively evaluating and selecting the optimal path that balances safety, target hit rate, and ease of operation presents a complex three-dimensional decision-making problem for the planner.

[0005] Some studies have attempted to introduce model-based optimization algorithms, such as constructing artificial potential field models, defining the restricted area as a repulsive force source and the target area as a gravitational force source, and planning the puncture trajectory by solving for the lowest energy path in the potential field. While theoretically feasible, these methods face inherent technical challenges. First, constructing and solving for accurate and smooth potential field models is computationally intensive and may not be suitable for intraoperative environments requiring rapid response. Second, the model parameters (such as the strength and range of repulsive and gravitational forces) are sensitive, lack intuitive physical interpretation, and are difficult to debug. More importantly, the decision-making process of these continuous field optimization methods is like a black box for clinicians, and the rationality of the output path is difficult to intuitively explain and verify. Summary of the Invention

[0006] This invention aims to address the technical problems in existing technologies where robot puncture path planning relies excessively on static planning or complex abstract models, resulting in poor path adaptability, high computational overhead, and opaque decision-making processes. To this end, this invention provides a robot-assisted execution system and method based on a programmed generation and screening process.

[0007] In a first aspect, this application provides a robot-assisted execution method based on a pre-planned puncture scheme. The method includes: acquiring medical image data characterizing a target operation area, wherein the medical image data identifies at least one target area and at least one restricted area; generating a discrete candidate trajectory set within a preset candidate trajectory generation cone for the at least one target area, wherein each candidate trajectory in the discrete candidate trajectory set is a spatial path connecting a preset needle insertion point within the candidate trajectory generation cone and a preset target point within the target area; performing a rigid geometric constraint check on the discrete candidate trajectory set to filter out candidate trajectories that geometrically interfere with the at least one restricted area, thereby obtaining a pre-purified subset of candidate trajectories; performing a flexible quality index evaluation on each candidate trajectory in the subset of candidate trajectories to obtain a comprehensive quality index for each candidate trajectory; selecting a preferred execution trajectory from the subset of candidate trajectories based on the comprehensive quality index; and controlling a robot execution system to perform a puncture operation on the at least one target area according to the preferred execution trajectory. This technical solution breaks down the complex path optimization problem into clear geometric judgment and evaluation steps through a paradigm of first generating massive amounts of data and then rigorously screening it. It solves the technical problem of how to efficiently and reliably find the safe and optimal path in a complex anatomical environment, bringing the beneficial effects of transparent decision-making process, high computational efficiency, and strong path security.

[0008] In one possible implementation of the first aspect, generating a discrete candidate trajectory set includes: defining a candidate trajectory generation cone around a central needle insertion point on the surface of the target operating area or the surface of a puncture probe; generating multiple discrete needle insertion points within the bottom surface or volume of the candidate trajectory generation cone using a preset sampling strategy; and connecting the multiple discrete needle insertion points to preset target points within the target area to form the discrete candidate trajectory set. This method efficiently generates candidate trajectories covering all reasonable needle insertion angles through structured sampling.

[0009] In one possible implementation of the first aspect, performing a rigid geometric constraint check includes: representing the restricted area and each candidate trajectory as geometric objects in three-dimensional space; for each candidate trajectory, detecting whether there is a spatial intersection between the geometry it represents and the geometry represented by the restricted area; if a spatial intersection exists, removing the candidate trajectory from the set of discrete candidate trajectories. This method, through deterministic three-dimensional collision detection, achieves rapid and explicit screening of unsafe paths, ensuring the basic safety of subsequent paths.

[0010] In one possible implementation of the first aspect, the execution of a flexible quality index evaluation includes acquiring at least two quality indices for each candidate trajectory. These at least two quality indices include: a minimum safety margin index, representing the shortest spatial distance between the path of the candidate trajectory and the boundaries of all non-interfering restricted areas; and a target hit accuracy index, representing the distance between the endpoint of the candidate trajectory and the preset geometric center of the target area. This multi-index evaluation system ensures that the finally selected path is the optimal solution that comprehensively considers both safety and effectiveness.

[0011] In one possible implementation of the first aspect, the flexible quality index assessment further includes obtaining a puncture ergonomic index, which characterizes the angle between the candidate trajectory and the normal vector of the plane containing the puncture insertion point. This approach takes into account both operational convenience and the degree of tissue damage, making the path decision more comprehensive and clinically sound.

[0012] In one possible implementation of the first aspect, obtaining a comprehensive quality index includes: normalizing the at least two quality indicators; and weighted summing the normalized quality indicators according to preset weights to obtain the comprehensive quality index. This approach provides a single, clear optimization objective for the final decision, simplifying the comparison of the merits of different candidate trajectories.

[0013] In one possible implementation of the first aspect, after completing the puncture operation on the first target area, the method further includes: acquiring intraoperative medical imaging data characterizing the current target operation area; registering the intraoperative medical imaging data with the initially acquired medical imaging data to determine the tissue displacement field caused by the previous puncture operation; applying the tissue displacement field to update the spatial coordinates of all remaining target areas and restricted areas in the medical imaging data; and based on the updated spatial coordinates, re-executing the steps of generating a discrete candidate trajectory set, performing rigid geometric constraint verification, performing flexible quality index evaluation, and selecting the preferred execution trajectory for the remaining target areas. This scheme constructs a feedback correction closed loop, solving the technical problem of subsequent planning failure caused by tissue deformation due to the puncture operation itself, enabling the entire multi-point puncture process to have adaptive correction capabilities and improving the accuracy of subsequent punctures.

[0014] In one possible implementation of the first aspect, during the process of controlling a robotic execution system to perform a puncture operation, the method further includes: acquiring puncture impedance profile data of the puncture needle tip in real time; comparing the real-time puncture impedance profile data with a preset reference impedance profile; and triggering an intelligent firing command of the robotic execution system when the matching degree between the real-time puncture impedance profile data and the reference impedance profile exceeds a preset threshold. This solution introduces force feedback information, and by analyzing the mechanical characteristics caused by changes in tissue structure during the puncture process, it achieves accurate confirmation of the puncture needle tip position, solves the error problem that may exist if image positioning is relied upon alone, and achieves a higher level of automation and standardization.

[0015] Secondly, this application provides a robot-assisted execution system based on a pre-planned puncture scheme. The system includes: a data acquisition module; a trajectory generation module; a trajectory verification module; a trajectory evaluation module; a trajectory selection module; and a robot control module, each module being configured to execute the method described in the first aspect.

[0016] In one possible implementation of the second aspect, the system further includes: a dynamic correction module, used to acquire intraoperative medical imaging data after completing the puncture operation on the first target area, determine the tissue displacement field caused by the previous puncture operation, apply the tissue displacement field to update the spatial coordinates of all remaining target areas and restricted areas, and trigger the trajectory generation module, trajectory verification module, trajectory evaluation module and trajectory selection module to regenerate the preferred execution trajectory for the remaining target areas. Attached Figure Description

[0017] Figure 1 Schematic diagrams of the structure of a robot-assisted execution system provided for some embodiments of this application; Figure 2A flowchart illustrating a robot-assisted execution method provided for some embodiments of this application; Figure 3 This is a timing diagram illustrating the operation of a robot-assisted execution system provided for some embodiments of this application. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0019] Hereinafter, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "multiple" means two or more.

[0020] In this application, unless otherwise expressly specified and limited, the term "connection" should be interpreted broadly. For example, a connection can be a fixed connection, a detachable connection, or an integral part; it can be a direct connection or an indirect connection through an intermediate medium. Furthermore, the term "electrical connection" can refer to the manner in which electrical connections are used to achieve signal transmission.

[0021] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be described in detail below. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0022] Please see Figure 1 This invention provides a robot-assisted execution system 100 based on a pre-planned puncture procedure. The hardware structure of system 100 can be deployed in an operating room and interface with the hospital's imaging information system. At the hardware level, system 100 mainly consists of three core parts: an image acquisition and processing unit, a robot execution unit, and a human-computer interaction unit.

[0023] The image acquisition and processing unit is typically a high-performance computing workstation equipped with a dedicated graphics processing unit (GPU) to accelerate 3D rendering and computational tasks. This workstation connects to external medical imaging equipment (such as PACS or intraoperative 3D ultrasound) via the Digital Imaging and Communication Standard (DICOM) protocol. The robotic actuator can be a tandem industrial robotic arm with six rotary joints, fitted with a specially designed puncture needle guide on its end flange, whose pose can be precisely controlled by the system. The human-machine interface unit includes a high-definition 3D display, keyboard, mouse, and possibly force feedback joysticks, allowing clinicians to plan, monitor, and intervene as needed during surgery.

[0024] like Figure 1 As shown, the core logic of the method described in this invention is implemented as software modules on the computing workstation. These modules work together as a whole to complete the entire process from data input to physical execution.

[0025] For example, the system 100 may include a data acquisition module 110, a trajectory generation module 120, a trajectory verification module 130, a trajectory evaluation module 140, a trajectory selection module 150, and a robot control module 160. Additionally, the system 100 may optionally integrate a dynamic correction module 170.

[0026] The data acquisition module 110, with its input end connected to external medical imaging equipment and its output end connected to the trajectory generation module 120, is responsible for loading, processing, and 3D reconstruction of all initial data related to surgical planning. It integrates 3D reconstruction algorithms and surface generation algorithms (e.g., the moving cube algorithm), enabling it to transform the original 2D medical image sequence into a computable 3D digital twin scene composed of a semantically labeled 3D mesh model (i.e., target area and no-entry area).

[0027] The trajectory generation module 120 has its input connected to the data acquisition module 110 and its output connected to the trajectory verification module 130. Based on the 3D scene model output by the data acquisition module 110, the trajectory generation module 120 generates a set containing multiple potential puncture paths. The trajectory generation module 120 defines a geometrically reasonable exploration space, namely a candidate trajectory generation cone, and performs high-density discrete sampling within this space to provide raw data for subsequent screening steps.

[0028] The trajectory verification module 130 has its input connected to the trajectory generation module 120 and its output connected to the trajectory evaluation module 140. The trajectory verification module 130 performs a geometric logic-based rigid constraint check on the candidate trajectory set output by the trajectory generation module 120, performs collision detection in three-dimensional space on each candidate trajectory with all known restricted areas, and eliminates paths that geometrically interfere with the restricted areas. This results in a preliminarily purified subset of trajectories that meets basic safety requirements being passed to the next module.

[0029] The trajectory evaluation module 140 has its input connected to the trajectory verification module 130 and its output connected to the trajectory selection module 150. This module quantitatively evaluates a subset of paths that have passed safety verification from multiple dimensions. Specifically, the trajectory evaluation module 140 calculates a set of continuous numerical indicators for each path that reflect its clinical merits, such as safety, accuracy, and ergonomics, and outputs the trajectory data with these evaluation indicators to the trajectory selection module 150.

[0030] The trajectory selection module 150 has its input connected to the trajectory evaluation module 140 and its output connected to the robot control module 160. The trajectory selection module 150 receives multi-dimensional index data from the evaluation module 140 and, through a preset mathematical framework reflecting clinical priorities, transforms the multi-objective optimization problem into a single-objective optimization problem. It calculates a unique comprehensive quality index for each qualified candidate path, sorts them according to this index, and finally selects the path with the highest score as the unique preferred execution trajectory. The geometric parameters of this trajectory are then transmitted to the robot control module 160.

[0031] The robot control module 160 has its input connected to the trajectory selection module 150 and its output connected to an external robot execution system. The robot control module 160 is responsible for translating the geometric information output by the trajectory selection module 150 into low-level motion commands that the robot can understand and execute, thereby driving the robot arm to the designated position and providing stable guidance for the puncture operation.

[0032] Optionally, the dynamic correction module 170 constructs a feedback loop. Its input receives real-time data from external medical imaging equipment (such as intraoperative ultrasound) and execution status information from the robot control module 160, while its output is connected to the trajectory generation module 120. In multi-point puncture tasks, this module 170 is activated after each puncture, calculates tissue deformation by registering intraoperative and preoperative images, updates the coordinates of the remaining target area and restricted area in the 3D scene model, and then triggers the trajectory generation module 120 to begin a new path planning for the next target point based on the updated model.

[0033] like Figure 2 , Figure 3 As shown, the method includes: S100. Acquire medical image data characterizing the target operation area, wherein the medical image data identifies at least one target area and at least one restricted area.

[0034] In one embodiment, the data acquisition module 110 establishes a communication session with the hospital's image archiving and communication system via the hospital's local area network using the DICOM protocol. The operator enters the patient's identifier into the system interface, and module 110 encapsulates this information into a C-FIND request and sends it to the PACS server. The server responds to the request, returning a list of all imaging examinations for that patient. The operator selects the examination relevant to the current surgical plan, typically a high-resolution pelvic magnetic resonance imaging (MRI) examination completed within one week prior to surgery. The data acquisition module 110 then sends a C-MOVE request, instructing the PACS server to push the selected image data to the system's workstation. This dataset typically contains multiple sequences; this method primarily focuses on axial scan sequences of T2-weighted imaging (T2WI).

[0035] A typical T2WI sequence may contain 40 consecutive two-dimensional slice images, each image having a pixel matrix of... While receiving data, the data acquisition module 110 parses the metadata header of each DICOM file, extracting and storing a series of key geometric and physical parameters, such as: pixel spacing (e.g., (0.5mm, 0.5mm); slice thickness (e.g., 3mm); image position (a three-dimensional vector); and image orientation (composed of two three-dimensional unit vectors). These parameters together form the basis for mapping two-dimensional pixel coordinates to three-dimensional physical space.

[0036] After acquiring the original two-dimensional image sequence, the data acquisition module 110 performs three-dimensional volume data reconstruction. The data acquisition module 110 converts the received discrete two-dimensional slice sequence into a spatially continuous three-dimensional volume data. Since the resolution of the original data along the axial direction (Z-axis) (determined by the slice thickness, such as 3mm) is usually much lower than the in-plane resolution (determined by the pixel spacing, such as 0.5mm), direct stacking would cause distortion along the axial direction. To construct isotropic volume data, i.e., volume data with the same resolution in all directions, the system employs a high-quality image interpolation algorithm, such as a cubic B-spline-based interpolation algorithm. This algorithm can smoothly and accurately interpolate the voxel (three-dimensional pixel) gray values ​​between slices based on the pixel gray values ​​on adjacent slices. Through this process, a sequence of 40 images with a 3mm slice thickness is reconstructed with high accuracy into a volume data with a resolution of, for example, 3mm. mm³, with dimensions of A three-dimensional voxel grid.

[0037] After obtaining this 3D voxel grid, it is necessary to establish a mathematical mapping relationship between the voxel coordinate system and the world coordinate system. The voxel coordinate system is a system with a corner point of the grid (e.g., the center of the voxel in the first slice, first row, or first column) as its origin. Using the three principal axes of the raster as coordinate axes and the voxel index as the coordinate system... The world coordinate system is a discrete coordinate system. The world coordinate system, typically defined by the MRI scanner, is a standard right-handed Cartesian coordinate system with units in physical units (e.g., millimeters). For the system to perform calculations in real physical space, it must be ensured that each voxel index in the raster is uniquely mapped to a three-dimensional physical coordinate in the world coordinate system through a rigorous mathematical transformation.

[0038] This transformation can be achieved through a... affine transformation matrix This is achieved by transforming voxel coordinates (represented in homogeneous coordinates) into world coordinates. Its specific form is as follows:

[0039] Affine transformation matrix It is not an abstract entity, but rather a matrix constructed from a series of key tags extracted from DICOM metadata during the data access process. This matrix can be decomposed into a combination of rotation, scaling, and translation.

[0040] The definitions and sources of the parameters in the matrix are as follows: Rotation and direction ( ): The top left corner of the matrix This part defines the orientation of the voxel coordinate system relative to the world coordinate system. Its column vectors are determined by the DICOM tag Image Orientation (Patient) (0020, 0037) and the calculated slice normal vectors. Image Orientation (Patient) provides two 3D unit vectors, representing the row vectors of the voxel raster. and column vectors The direction in the world coordinate system. The third direction is the slice normal vector. It can be obtained by calculating the cross product of these two vectors: These three orthogonal unit vectors form the columns of the rotation matrix.

[0041] Scaling and Spacing ): and These represent the physical dimensions of the voxels in the row and column directions, and their values ​​are derived directly from the second and first values ​​in the DICOM tag Pixel Spacing (0028, 0030). This represents the physical size (layer thickness) of the voxel in the slice direction, derived from the DICOM tag Slice Thickness (0018, 0050) or the interlayer spacing calculated from the Image Position (Patient) of adjacent slices.

[0042] Translation and origin ( ): The last column of the matrix It is a translation vector that defines the origin of the voxel coordinate system, i.e., its index is... The center of the voxel, its physical location in the world coordinate system. This value is directly derived from the DICOM label Image Position (Patient) (0020, 0032) of the first slice.

[0043] By constructing this matrix, the system establishes an accurate mathematical model for the three-dimensional voxel grid, ensuring that all subsequent voxel-based operations (such as distance calculation and collision detection) can be performed in a unified world coordinate system with real physical meaning.

[0044] For example, we can use a specific set of DICOM values ​​to calculate the transformation from a voxel index to world coordinates. Assume the following information is read from the DICOM metadata: Image Position (Patient) (First slice): mm Image Orientation (Patient):

[0045] Pixel Spacing: mm / pixel, therefore

[0046] Slice Thickness: mm, therefore

[0047] First, construct the rotation matrix. Therefore, the rotation matrix is ​​the identity matrix: .

[0048] Next, construct the complete affine transformation matrix. :

[0049] Calculate the voxel index as The physical coordinates of the voxel center in the world coordinate system Write the voxel index in homogeneous coordinate form. And perform matrix multiplication:

[0050]

[0051]

[0052]

[0053] Therefore, the voxel index is The voxel center, its physical location in the world coordinate system is mm.

[0054] After the 3D reconstruction is completed, clinicians use the system's human-computer interface to segment the reconstructed 3D volume data into anatomical structures, identifying at least one target area and at least one restricted area. The clinician uses software segmentation tools to delineate each key structure. Specifically, the clinician needs to identify at least one target area, such as a suspicious lesion that appears as a diffuse low signal with unclear boundaries on T2WI and is reported by radiologists as a PI-RADS 4 level; and at least one restricted area, which is a critical physiological structure that the puncture needle must strictly avoid touching or passing through under any circumstances, such as the urethra, the anterior rectal wall, and the neurovascular bundle surrounding the prostate. After the clinician completes the delineation of these structures, the data acquisition module 110 internally calls its integrated surface generation algorithm (e.g., the Marching Cubes algorithm) to traverse the boundaries of the masked image and generate a 3D mesh model (stored in STL format) representing the surface of the structure, composed of numerous tiny triangular facets. Each generated mesh model is assigned a clear semantic label by the system (such as target area-T1, restricted area-urethra-Z1) and its precise world coordinates along with all its vertices are stored in the scene database created for this surgery.

[0055] For example, suppose that after the doctor completes the annotation, the system database generates the following key geometric objects: Target Area A three-dimensional mesh model of an irregular ellipsoid-like structure composed of 2560 triangular facets. The system determines its geometric center in the world coordinate system by calculating the arithmetic mean of the coordinates of all its vertices. The coordinates are measured in millimeters, and its volume is calculated to be 1.23 cubic centimeters. For example, the coordinates of the 100th vertex in its vertex list are... Restricted Area (Urethra): A slender and curved tubular 3D mesh model composed of 8192 triangular facets, whose mesh data depicts its complex orientation and shape in 3D space. For example, the coordinates of the 5123rd vertex in its mesh vertex list are... Restricted Area (Rectal wall): A sheet-like three-dimensional mesh model with a large curvature, consisting of 12,544 triangular facets.

[0056] S200. For the at least one target area, a discrete candidate trajectory set is generated within a preset candidate trajectory generation cone, wherein each candidate trajectory in the discrete candidate trajectory set is a spatial path connecting a preset needle entry point within the candidate trajectory generation cone and a preset target point within the target area.

[0057] After the 3D digital twin scene is constructed, the trajectory generation module 120 performs this step for the target area. S200 may include: S210. On the surface of the target operating area or a puncture probe, a cone is generated around a central needle insertion point to define the candidate trajectory.

[0058] In a typical transperineal puncture scenario, the base of the trajectory generation cone lies on the plane representing the perineal skin, i.e., the surface of the target manipulation area. In the system's world coordinate system, this plane can be defined as... The trajectory generation module 120 first reads the target area from the scene database. Geometric center coordinates The trajectory generation module 120 will... Vertical projection to On the plane, the center needle entry point is obtained. Subsequently, the system revolved around Define a geometric region as the source region for all candidate needle insertion points, i.e., the base of the cone. The shape and size of this region can be determined based on clinical practice and safety considerations; for example, this region can be a circular region with a radius of... It is a configurable parameter, for example, it can be set to... Millimeters. Ultimately, this... With center and radius as A circular region of mm in diameter forms the base of the candidate trajectory generation cone. The vertex of this cone is then designated as the target point within the target region, i.e., the preset target point within the target region. Thus, a region consisting of a circular base and vertices... The resulting cone, which has definite boundaries in three-dimensional space, is defined.

[0059] S220. Within the bottom surface or volume of the candidate trajectory generation cone, generate multiple discrete needle entry points using a preset sampling strategy.

[0060] After defining the base of the candidate trajectory generation cone, the system performs high-density discrete sampling within this continuous two-dimensional region to generate a large number of discrete needle entry points. In this application, the system employs a uniform grid sampling strategy, constructing a grid around a circular source region with a radius of 15 mm. Let the coordinate axis be a circumscribed square with its center and sides parallel to the coordinate axes. The side length of this square is... mm. Then, the system generates a two-dimensional grid within the square region at a preset sampling step size, such as 1 mm. This grid contains Each of the 961 grid points is then processed by the system, which iterates through these points and performs a check on each point: calculating the distance from that point to the center of the circle. The distance, if the distance is less than or equal to the radius If the value is mm, then the point is accepted as a valid candidate needle insertion point. Using this method, the system can generate approximately... Discrete needle insertion points Each accepted needle insertion point is assigned a precise three-dimensional coordinate, with its z-component set to 0.

[0061] S230. Connect the plurality of discrete needle entry points to the preset target points in the target area respectively to form the discrete candidate trajectory set.

[0062] The system will target area geometric center This serves as the preset target point. Then, module 120 iterates through all 707 candidate needle insertion points generated in S220. For each needle insertion point, the system defines a connection line to that point. With the target point A straight line segment in space is used as a candidate trajectory. Each trajectory is stored in the system using a standardized data structure, which typically includes: a unique trajectory ID, starting point coordinates, ending point coordinates, and a unit direction vector pointing from the starting point to the ending point. and the total length of the trajectory. After traversing all needle entry points, a discrete candidate trajectory set containing 707 precisely parameterized candidate trajectories is obtained. It is then generated.

[0063] For example, with a candidate trajectory Taking the generation process as an example, suppose the system generates a candidate needle entry point within the source region through grid sampling. The target point is .

[0064] To generate candidate trajectories The system performs the following calculations: Calculate the direction vector:

[0065] Calculate the trajectory length: mm.

[0066] Calculate the unit direction vector: .

[0067] Ultimately, the trajectory By: {ID:" The values ​​in the set are stored in the form of ",Start:(40.21,52.54,0),End:(35.21,48.54,62.08),Dir:(-0.08011,-0.06409,0.99473),Len:62.409}. The other 706 trajectories in the set were also precisely defined in the same way.

[0068] S300. Perform a rigid geometric constraint check on the discrete candidate trajectory set to filter out candidate trajectories that have geometric interference with the at least one forbidden zone, thereby obtaining a preliminary purified subset of candidate trajectories.

[0069] After generating a set of discrete candidate trajectories Next, the trajectory verification module 130 performs this step. This step mainly includes three stages: puncture needle geometric modeling, collision detection coarse screening, and collision detection fine screening.

[0070] In the geometric modeling stage of the puncture needle, the system establishes a geometric model with physical volume for each abstract candidate trajectory to realistically simulate the space occupied by the puncture needle within the tissue. The system will then process each candidate trajectory... In three-dimensional space, it is visualized as a slender capsule with hemispherical ends. The central axis and trajectory of this capsule... The line segments completely overlap, and their radius is... The radius is set to the radius of the puncture needle used in actual surgery. For example, for an 18G puncture needle, its outer diameter is approximately 1.27 mm, therefore the radius... It was set to 0.635mm. Compared to a simple cylinder, using a capsule as a model allows for a smoother and more accurate representation of the tip and tail of the puncture needle, resulting in more accurate results in collision detection.

[0071] In the collision detection coarse screening stage, to efficiently handle the interaction between hundreds of trajectories and the restricted area mesh containing tens of thousands of triangular faces, the system employs an acceleration structure based on an axis-aligned bounding box (AABB) hierarchical tree. An AABB is a hexahedron whose faces are parallel to the coordinate axes of the coordinate system, capable of completely containing a complex geometry. The AABB hierarchical tree is a binary tree structure used to spatially partition a complex mesh model (such as a urethral model).

[0072] The construction process of the AABB hierarchical tree is as follows: Calculate the entire urethral mesh model. The system constructs an AABB (Alternating Elemental Block), which becomes the root node of the tree. Along the longest axis of this AABB (e.g., the x-axis), all triangles within the AABB are sorted according to the x-coordinate of their midpoints, and the set of triangles is divided into left and right subsets at the middle position. Next, the AABBs that tightly enclose these two subsets are calculated, and these two new AABBs become the left and right children of the root node. This process is repeated recursively until each leaf node's AABB contains only a few (e.g., fewer than 8) triangles. In this way, a complex, unorganized set of triangles is organized into a tree structure with spatial proximity. This hierarchical tree is pre-calculated and stored for each forbidden zone model after the scene construction is completed in step S100.

[0073] When the verification module 130 needs to detect a candidate trajectory (from capsule body) (Representative) Whether it is related to the urethra During a collision, first calculate The AABB. Then, a recursive query algorithm is executed: Will An intersection test is performed between the root node of the AABB tree and the urethral AABB tree, that is, to compare whether the coordinate ranges of the two AABBs on the x, y, and z axes overlap. If they do not intersect, then it can be determined that... If no collision is possible with the entire urethra model, the query terminates immediately and returns "no collision". If the two intersect, the query is recursively performed... The AABB of the current node is intersected with the AABBs of its left and right children. This recursive process continues, exploring all nodes that intersect with the AABBs of the current node downwards. The branches where AABB intersect.

[0074] Ultimately, this coarse screening process will output a result that is possibly related to... A collection of a very small number of triangular facets that collide. These triangular facets are all located within the triangles whose AABB shape is perpendicular to the triangles in the triangle. In the leaf nodes where AABB intersect.

[0075] In the collision detection screening stage, only the small subset of triangular facets that were filtered out in the coarse screening stage will enter the capsule body. The precise intersection test is performed. This step employs a variant algorithm based on the Separating Axis Theorem (SAT). The core idea of ​​SAT is that if two convex bodies do not intersect, then there must exist a separating axis such that the projections of the two objects onto this axis do not overlap. For the detection of the capsule and the triangular facet, the potential separating axes to be tested include: the normal vector of the triangular facet, the two perpendicular vectors of the capsule's axis, and the three vectors obtained by cross products of the three sides of the triangular facet with the capsule's axis. For each potential separating axis, the algorithm calculates the projection intervals of the capsule and the triangular facet onto that axis. As long as any axis is found such that the two projection intervals do not overlap, it can be immediately determined that the two are not colliding, and the next triangular facet can be tested. Only when, after testing all potential separating axes, it is found that the projections of the two objects overlap on every axis, can a final determination be made that the two are colliding.

[0076] The verification module 130 executes the above process for any candidate trajectory. As long as its corresponding capsule body If a trajectory is determined to collide with any triangle in any restricted area model within the scene during the final screening stage, it will be immediately marked as unqualified and removed from the active candidate set. Removed from the list. This removal process will continue until... All trajectories were validated. Ultimately, the remaining trajectories formed a subset of candidate trajectories that had undergone preliminary purification. .

[0077] For example, suppose the urethra The AABB hierarchical tree has been constructed. The AABB range of its root node is... .

[0078] Validate trajectory Its starting point ,end ,radius mm. The system first calculates its capsule size. AABB.

[0079] Its coordinate range is approximately , , Recorded as .

[0080] Coarse screening: System comparison and .

[0081] X-axis: and There is overlap.

[0082] Y-axis: and There is overlap.

[0083] Z-axis: and There is overlap.

[0084] Because all three axes overlap, the AABB tree intersects. The system begins a recursive query. After multiple levels of recursion, the query eventually pinpoints two leaf nodes of the AABB tree, whose AABBs intersect with each other. The intersections contain a total of 32 triangular facets.

[0085] Fine screening: The system compares each of the 32 triangular facets with the capsule body. Perform fine-grained SAT detection. Suppose that when detecting the 18th triangular facet, the algorithm finds a separating axis (e.g., the normal vector of the facet) such that the projections of the capsule body and the facet onto that axis do not overlap. The algorithm immediately terminates the detection of that facet and continues to the next one.

[0086] Ultimately, after testing all 32 faces, no collisions were found. The system determined... With the urethra No collision. Subsequently, its contact with the rectal wall was tested in the same manner. The relationship was also reported as no collision. Therefore, the trajectory... It successfully passed the rigid geometric constraint verification and was retained.

[0087] S400. Perform a flexible quality index evaluation on each candidate trajectory in the subset of candidate trajectories.

[0088] After passing the rigid verification of S300, a subset of candidate trajectories that can ensure safety was obtained. Subsequently, the trajectory assessment module 140 performs this step, evaluating the quality of each feasible trajectory from multiple clinically relevant dimensions.

[0089] S410. Obtain a minimum safety margin index, which represents the shortest spatial distance between the path of the candidate trajectory and the boundaries of all non-interfering restricted areas.

[0090] 140 pairs of evaluation modules Each trajectory in Perform a three-dimensional spatial distance calculation to obtain its minimum safety margin index. This metric directly quantifies the safety redundancy of the trajectory. The system models the trajectory path as a radius of... capsule body The module then calculates the shortest 3D spatial distance between the capsule and the surfaces of all restricted area 3D mesh models in the scene. This calculation uses the GJK (Gilbert-Johnson-Keerthi) distance algorithm, which efficiently iteratively finds the shortest distance between two sets of convex bodies. For non-convex restricted area meshes, the system can pre-decompose them into multiple convex parts. The module rigorously calculates the trajectory. The minimum distance to each restricted area is calculated, and then the smallest of these calculated distance values ​​is selected as the final minimum safety margin index for the trajectory. .

[0091] For example, for trajectory Module 140 starts the GJK distance calculation program. The program reports, representing... The capsule body and the urethra The shortest distance between the mesh surface and the rectal wall is 3.82 mm. The shortest distance on the grid surface is 9.15 mm. Taking the smaller of the two values ​​yields... The minimum safety margin index is mm.

[0092] S420. Obtain a target hit accuracy index, which represents the distance between the end point of the candidate trajectory and the preset geometric center of the target area.

[0093] The evaluation module 140 then quantifies the accuracy of each trajectory in hitting the target point, that is, obtains the target point hit accuracy index. In this embodiment, since the endpoints of all candidate trajectories are set as the geometric center of the target area during generation... Therefore, the distance between the endpoint and the preset geometric center is zero. To provide a more meaningful evaluation, the system calculates the target center crossing degree. The system calculates the trajectory. This infinitely extended line, and its target point The shortest vertical distance between them. This indicator This vertical distance is defined as the distance in millimeters, and the smaller the value, the higher the precision.

[0094] For example, for trajectory Because of its generation method, it is guaranteed to pass through precisely. Therefore, the distance it crosses through the center of the target is 0.0 mm, that is... However, in other embodiments, the trajectory may be defined as non-linear (e.g., a Bézier curve or a piecewise linear path used to bypass obstacles). In this case, the endpoint of the trajectory... With the preset target There will be a non-zero distance error between them. In this more realistic model, the S420 step will have key distinguishing and optimization significance. Or in some algorithms, the planning target is not the geometric center. This is not a single point, but the target area. The entire volume. The planning algorithm may seek a path whose endpoint falls within this volume, but it does not necessarily have to be the center. In this case, different paths will result in the endpoint falling in different positions within the target volume, and S420 measures these landing points relative to the ideal bullseye. The degree of similarity.

[0095] S430. Obtain a puncture ergonomic index, wherein the puncture ergonomic index characterizes the angle between the candidate trajectory and the normal vector of the plane where the puncture needle insertion point is located.

[0096] Assessment module 140 evaluates the pathway from the perspective of clinical ease of operation and obtains puncture ergonomic indicators. System calculates trajectory unit direction vector With the plane of needle insertion (e.g.) normal vector of the plane (pointing into the patient's body) the angle between The included angle is calculated using the dot product formula for two vectors: This calculated included angle (Unit: degrees or radians) is the ergonomic index for puncture. The smaller the value, the better the ergonomic performance.

[0097] For example, for trajectory Its unit direction vector is It is related to the normal vector pointing out of the body. The dot product is Therefore, the angle between it and the normal. Degree. Therefore, Spend.

[0098] S500: Obtain a comprehensive quality index for each candidate trajectory.

[0099] After calculating multi-dimensional quality indicators for each qualified candidate trajectory, the trajectory selection module 150 performs this step to calculate a single comprehensive score for each trajectory. S500 includes: S510. Normalize the at least two quality indicators.

[0100] The trajectory selection module 150 first normalizes the various heterogeneous indicators to make them comparable. The module uses a min-max normalization method. For indicators where larger values ​​equate to better performance, the normalization formula is: For indicators where smaller values ​​are considered better, the opposite formula is used: After normalization, all metrics are mapped to... The range is defined, and a larger value indicates a better performance.

[0101] S520. According to the preset weights, the normalized multiple quality indicators are weighted and summed to obtain the comprehensive quality index.

[0102] After all indicators are normalized, the system introduces a set of preset weighting coefficients. ,For example For each candidate trajectory Its comprehensive quality index The calculation is performed using the following linear weighted summation formula:

[0103] For example, assuming that statistically, The range is [0.80, 9.50] mm. The range is [0.0, 0.5] mm. The range is [5.0, 30.0] degrees.

[0104] For trajectory Its original indicator is Its normalized index is calculated as follows:

[0105]

[0106]

[0107] Its overall quality index is: .

[0108] S600. Based on the comprehensive quality index, select a preferred execution trajectory from the subset of candidate trajectories.

[0109] Track selection module 150 is Each trajectory in the calculation has its unique value. Value. Subsequently, module 150 applies its values ​​to all trajectories. The values ​​are sorted in quick descending order. The trajectory at the top of the list, i.e., the one with the highest overall quality index, is ultimately determined as the preferred execution trajectory for this planning process. .

[0110] S700: Control a robot execution system to perform a puncture operation on the at least one target area according to the preferred execution trajectory.

[0111] This step is performed by the robot control module 160. The control module 160 utilizes a pre-calibrated homogeneous transformation matrix. The preferred execution trajectory will be selected. The geometric parameters are transformed from the image coordinate system to the robot's base coordinate system. Then, module 160 calls the inverse kinematics solver to calculate the joint angles required for the robot's end effector to reach the target pose. Next, the motion planner generates a smooth, collision-free trajectory in joint space. Finally, this trajectory is sent to the underlying servo controller to precisely drive the robot arm to move to the target pose and lock it in place. This step is a common technique used by those skilled in the art. After obtaining the above data through this application, those skilled in the art can control the robot to execute corresponding instructions, and this application will not elaborate further on it here.

[0112] In one embodiment, the method of the present invention further includes a dynamic correction mechanism to address problems caused by tissue deformation during multi-point puncture. That is, the method further includes: S800: After completing the puncture operation on the first target area, acquire intraoperative medical imaging data characterizing the current target operation area.

[0113] Complete the target area with robot assistance Following the puncture and sampling, the dynamic correction module 170 is automatically triggered. This module 170 immediately sends a command to the intraoperative 3D ultrasound probe integrated with the system to perform a rapid volumetric data scan, acquiring an intraoperative 3D ultrasound image reflecting the current true state of the tissue. .

[0114] S810. Register the intraoperative medical imaging data with the initially acquired medical imaging data to determine the tissue displacement field caused by the previous puncture operation.

[0115] Get Subsequently, the core task of module 170 is to compare it with the preoperative MRI images used as the planning benchmark. Accurate non-rigid registration is performed. The system employs a Demons registration algorithm based on image intensity. This algorithm iteratively calculates a three-dimensional displacement vector field (DVF), where each vector in the DVF... Both describe the three-dimensional spatial displacement of the tissue at that location from its preoperative state to its current state.

[0116] S820. Apply the tissue displacement field to update the spatial coordinates of all remaining target areas and restricted areas in the medical image data.

[0117] After calculating the DVF, the update process begins. Module 170 iterates through all target areas in the scene database that have not yet been punctured (e.g., ) and all restricted areas ( For each vertex of these geometric models. Module 170 queries the DVF for the displacement vector corresponding to that point. Then calculate its updated coordinates: By performing this operation on all vertices of a model, the system obtains a new geometric model of the anatomical structure after deformation.

[0118] S830. Based on the updated spatial coordinates, re-execute the steps of generating a discrete candidate trajectory set, performing rigid geometric constraint verification, performing flexible quality index evaluation, and selecting the preferred execution trajectory for the remaining target area.

[0119] After the geometric models of all remaining target areas and restricted areas have been updated, the dynamic correction module 170 triggers a global replanning instruction. The system uses these updated geometric models, reflecting the current real-world situation, as new input, starting from S200, for the next target (e.g., The entire process from S200 to S600 is executed again to find the new optimal puncture path under the current dynamic map and pass it to S700 for execution.

[0120] In another optional embodiment, the method of the present invention also integrates force feedback-based intelligent firing control to further improve the automation and standardization of puncture. The method further includes: S900: Real-time acquisition of puncture impedance profile data at the tip of the puncture needle.

[0121] In the S700, after the robot positions the guide to the needle insertion point along the preferred trajectory, the doctor inserts the puncture needle and initiates the robot-controlled automated needle insertion procedure. As the puncture needle advances at a constant speed, a high-frequency force / torque sensor mounted at the end of the needle advance device or on the needle hub begins to collect data in real time at a sampling rate of, for example, 1 kHz, of the resistance experienced by the puncture needle in the axial direction (i.e., the puncture direction). These continuous force data points constitute a puncture impedance profile that varies with time (or depth).

[0122] S910. Compare the real-time puncture impedance profile data with a preset reference impedance profile.

[0123] The system pre-stores one or more reference impedance profiles. These profiles are typical mechanical characteristic curves representing the sequential penetration of the puncture needle through different tissue layers, obtained through numerous in vitro and in vivo experiments. A typical reference profile shows a sharp increase in resistance when penetrating the tough prostate capsule, followed by a sharp drop in force when the needle tip enters the looser prostate parenchyma. The system employs a Dynamic Time Warping (DTW) algorithm to continuously and continuously compare the real-time acquired puncture impedance profile data stream with the reference impedance profiles. The DTW algorithm can calculate the similarity between two unequal time sequences, thereby effectively identifying patterns that match reference features (such as the capsule penetration peak).

[0124] S920. When the matching degree between the real-time puncture impedance profile data and the reference impedance profile exceeds a preset threshold, the intelligent firing command of the robot execution system is triggered.

[0125] The system sets a matching threshold (e.g., DTW distance less than a preset value). When the real-time monitored DTW matching degree first exceeds this threshold, the system determines that the puncture needle tip has successfully penetrated the prostate capsule and entered the target tissue area. At this point, the system can perform two operations: one is to highlight a prompt on the screen, suggesting that the doctor fire the biopsy gun at the current location; the other, with the doctor's prior authorization, is for the system to directly send an electrical signal to the biopsy gun actuator integrated with the robot control system, triggering its intelligent firing command to complete tissue sampling. This method, by fusing force sensory information, provides image-independent verification for determining the position of the puncture needle tip, reducing the delay and error of human judgment.

[0126] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects.

[0127] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0128] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. These modifications or substitutions do not cause the essence of the corresponding technical solutions to depart from the spirit and scope of the technical solutions of the embodiments of this application.

[0129] Through the above description of the embodiments, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0130] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another apparatus, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0131] The units described as separate components may or may not be physically separate. A component shown as a unit can be one physical unit or multiple physical units; that is, it can be located in one place or distributed in multiple different places. Depending on actual needs, some or all of the units can be selected to achieve the purpose of this embodiment.

[0132] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit described above can be implemented in hardware.

[0133] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A robot-assisted execution method based on a pre-planned puncture scheme, characterized in that, include: Acquire medical image data characterizing the target operation area, wherein the medical image data identifies at least one target area and at least one restricted area; For the at least one target area, a discrete candidate trajectory set is generated within a preset candidate trajectory generation cone, wherein each candidate trajectory in the discrete candidate trajectory set is a spatial path connecting a preset needle entry point within the candidate trajectory generation cone and a preset target point within the target area. A rigid geometric constraint check is performed on the discrete candidate trajectory set to filter out candidate trajectories that have geometric interference with the at least one forbidden zone, thereby obtaining a preliminary purified subset of candidate trajectories. A flexible quality index evaluation is performed on each candidate trajectory in the subset of candidate trajectories to obtain a comprehensive quality index for each candidate trajectory. Based on the comprehensive quality index, a preferred execution trajectory is selected from the subset of candidate trajectories. The preferred execution trajectory is used by the robot execution system to perform a puncture operation on the at least one target area.

2. The method according to claim 1, characterized in that, The generation of a discrete candidate trajectory set includes: On the body surface of the target operating area or on the surface of a puncture probe, a cone is generated around a central needle insertion point to define the candidate trajectory. Within the bottom surface or volume of the candidate trajectory generation cone, multiple discrete needle entry points are generated using a preset sampling strategy. The discrete needle insertion points are connected to the preset target points in the target area to form the discrete candidate trajectory set.

3. The method according to claim 1, characterized in that, The execution of a rigid geometric constraint verification includes: The restricted area and each candidate trajectory are represented as geometric objects in three-dimensional space; For each candidate trajectory, detect whether there is a spatial intersection between the geometry it represents and the geometry represented by the restricted area; If there is a spatial intersection, the candidate trajectory is removed from the set of discrete candidate trajectories.

4. The method according to claim 1, characterized in that, The execution of a flexible quality index evaluation includes obtaining at least two quality indices for each candidate trajectory, the at least two quality indices including: A minimum safety margin index characterizes the shortest spatial distance between the path of the candidate trajectory and the boundaries of all non-interfering restricted zones; A target hit accuracy index characterizes the distance between the end point of the candidate trajectory and the preset geometric center of the target area.

5. The method according to claim 4, characterized in that, The flexible quality index assessment also includes obtaining a puncture ergonomic index, which characterizes the angle between the candidate trajectory and the normal vector of the plane where the puncture needle insertion point is located.

6. The method according to claim 4, characterized in that, The process of obtaining a comprehensive quality index includes: Normalize the at least two quality indicators; Based on preset weights, the normalized multiple quality indicators are weighted and summed to obtain the comprehensive quality index.

7. The method according to claim 1, characterized in that, After completing the puncture operation on the first target area, the method further includes: Acquire intraoperative medical imaging data characterizing the current target surgical area; The intraoperative medical imaging data is registered with the initially acquired medical imaging data to determine the tissue displacement field caused by the previous puncture operation. The spatial coordinates of all remaining target areas and restricted areas in the medical image data are updated by applying the tissue displacement field. Based on the updated spatial coordinates, the steps of generating a discrete candidate trajectory set, performing rigid geometric constraint verification, performing flexible quality index evaluation, and selecting the preferred execution trajectory are re-executed for the remaining target area.

8. The method according to claim 1, characterized in that, The method further includes: controlling a robotic execution system to perform a puncture operation, this step including: Real-time acquisition of puncture impedance profile data at the tip of the puncture needle; The real-time puncture impedance profile data is compared with a preset reference impedance profile. When the matching degree between the real-time puncture impedance profile data and the reference impedance profile exceeds a preset threshold, the intelligent firing command of the robot execution system is triggered.

9. A robot-assisted execution system based on a pre-planned puncture scheme, characterized in that, include: A data acquisition module is used to acquire medical image data that characterizes the target operation area, wherein the medical image data is identified with at least one target area and at least one restricted area; A trajectory generation module is used to generate a discrete candidate trajectory set within a preset candidate trajectory generation cone for the at least one target area, wherein each candidate trajectory in the discrete candidate trajectory set is a spatial path connecting a preset needle entry point within the candidate trajectory generation cone and a preset target point within the target area. A trajectory verification module is used to perform a rigid geometric constraint verification on the discrete candidate trajectory set to screen out candidate trajectories that have geometric interference with the at least one forbidden zone, thereby obtaining a preliminary purified subset of candidate trajectories. A trajectory evaluation module is used to perform a flexible quality index evaluation on each candidate trajectory in the candidate trajectory subset and obtain a comprehensive quality index for each candidate trajectory. A trajectory selection module is used to select a preferred execution trajectory from the subset of candidate trajectories based on the comprehensive quality index; A robot control module is used to control a robot execution system to perform a puncture operation on the at least one target area according to the preferred execution trajectory.

10. The system according to claim 9, characterized in that, Also includes: A dynamic correction module is used to acquire intraoperative medical imaging data after the puncture operation is completed on the first target area, determine the tissue displacement field caused by the previous puncture operation, apply the tissue displacement field to update the spatial coordinates of all remaining target areas and restricted areas, and trigger the trajectory generation module, trajectory verification module, trajectory evaluation module and trajectory selection module to regenerate the preferred execution trajectory for the remaining target areas.

Images