A flexible needle puncture path planning method fusing improved A algorithm and artificial potential field method

Abstract

The present application relates to a kind of flexible needle puncture path planning methods of fusing improved A algorithm and artificial potential field method, comprising: obtaining medical image, extracting the target point of medical image and the obstacle needing to be avoided, generates planning coordinate plane;In the planning coordinate plane, the node is expanded using puncture needle simplified model, based on multilevel heuristic function, and the guide mechanism of introducing artificial potential field is generated evaluation function:Through the evaluation function, the node after expansion is evaluated, according to the evaluation result, the planning path of flexible needle puncture is generated.The present application can obtain high-precision, high-security autonomous path planning solution, so as to better cope with the challenge of non-uniform obstacle distribution and dynamic anatomical structure, optimize the clinical executability of puncture trajectory, improve the calculation efficiency, and improve the success rate of operation and patient safety performance.

Description

A flexible needle puncture path planning method integrating improved A algorithm and artificial potential field method Technical Field

[0001] This invention relates to the field of flexible needle puncture path planning technology, and in particular to a flexible needle puncture path planning method that integrates an improved A algorithm and an artificial potential field method. Background Technology

[0002] As a core tool in modern minimally invasive medicine, puncture and interventional techniques, with their ability to achieve precise diagnosis and treatment through millimeter-level incisions, significantly reduce the risk of postoperative complications and shorten the recovery period. They are widely used in areas such as tumor biopsy, nerve block anesthesia, radioactive particle implantation, and targeted ablation therapy. However, traditional manual puncture is limited by the physician's experience and physical fatigue, easily leading to trajectory deviations and posing radiation exposure risks. With the advancement of technology, robot-assisted medicine has made great progress, especially in minimally invasive procedures requiring high precision. Using robots to replace traditional manual puncture has become an important topic in the integration of medicine and engineering.

[0003] Currently, most puncture needles used in clinical practice are rigid needles, which can only puncture in a straight line. However, the flexible needles used by puncture robots can use the angle of the needle tip to form a certain bend during the puncture process, thereby avoiding important tissues, blood vessels, nerves, etc., and then reaching the target point.

[0004] During flexible needle puncture, obstacles that need to be avoided are usually unevenly distributed. Therefore, the path needs to be planned before the puncture is performed. A suitable path plan helps the robot to perform the puncture better and improve its safety and accuracy. However, flexible puncture needles have non-holonomic constraints, and different needle models are used for different tumor sites. Therefore, how to plan a suitable path according to the constraints of flexible needles is a challenge.

[0005] Currently, in the field of flexible needle puncture path planning, there are three main planning methods:

[0006] Rapidly-exploring Random Tree (RRT) is a path planning method based on probabilistic sampling. Its core idea is to gradually explore the unknown space and generate feasible paths through random sampling and tree-structure expansion. Numerous researchers have guided and optimized its expansion method to achieve faster convergence, and various curve smoothing techniques are then used to connect the generated path points that satisfy the constraints.

[0007] Artificial Potential Field (APF) is a local path planning algorithm based on virtual force fields. It guides a robot or machine from a starting point to a target point by simulating the "attraction" and "repulsion" forces in a physical field, while simultaneously avoiding obstacles. Its core idea is to construct an attractive force field for the target point and a repulsive force field for obstacles. A collision-free path is generated through gradient descent of the resultant force field, and various curve smoothing methods are then used to connect the generated path points that satisfy the constraints.

[0008] Path planning algorithms based on swarm intelligence and evolutionary computation (such as genetic algorithms, ant colony algorithms, and particle swarm algorithms) optimize the candidate path set in the solution space iteratively by embedding various constraints. By simulating natural biological behavior or physical phenomena, they converge faster and gradually approach the global optimum to find the best path.

[0009] The first two methods mentioned above generate discrete path points and then use certain methods to generate a smooth curve that satisfies the constraints. The third method uses an iterative search approach. These methods often require significant time and computational resources. In puncture and interventional procedures, the patient's tissues are not homogeneous, and there are also differences between the tissues of different patients. Furthermore, during the procedure, interference from the patient's heartbeat and breathing can cause deviations in the pre-planned trajectory, often requiring replanning during the procedure. Therefore, to meet real-time requirements, ensure that the planned path satisfies kinematic constraints, and avoids getting trapped in local optima, this invention proposes a flexible needle puncture path planning method based on a combination of an improved A* algorithm and an artificial potential field method.

[0010] The common drawbacks of the above three methods are:

[0011] The paths found do not conform to kinematic constraints and are not smooth enough: the paths found by the fast expanding random tree algorithm and the artificial potential field method do not conform to the non-holonomic kinematic constraints of the flexible needle, and a lot of time needs to be spent using path smoothing methods to smooth the paths, which undoubtedly increases the computational cost.

[0012] The safety of the paths sought is not good enough: the random sampling mechanism of the fast-expanding random tree algorithm leads to excessive randomness in the path search, and the generated path may get too close to the tissues, blood vessels, nerves, etc. that need to be avoided, resulting in a certain risk in the smoothly generated trajectory, which cannot always remain within a safe range; the artificial potential field method may have a balance point of "attraction" and "repulsion", thus forming a local minimum and causing path planning failure; path planning algorithms based on swarm intelligence and evolutionary computation may converge too early and get trapped in local extrema, which may generate unsafe paths.

[0013] Insufficient real-time performance: The path generated by the Fast Expanded Random Tree algorithm depends on the sampling density, and the planning time is exponentially related to the scene complexity, resulting in high computational complexity; the paths generated by the artificial potential field method and the Fast Expanded Random Tree algorithm both require optimization using path smoothing methods, which leads to a sharp increase in the planning time; path planning algorithms based on swarm intelligence and evolutionary computation require a large amount of computing resources for iterative updates, making it difficult to meet the replanning needs during the operation. Summary of the Invention

[0014] To address the problems existing in the prior art, the purpose of this invention is to provide a flexible needle puncture path planning method that integrates the improved A algorithm and the artificial potential field method. This method can achieve a high-precision and high-safety autonomous path planning solution, thereby better addressing the challenges of non-uniform obstacle distribution and dynamic anatomical structures, optimizing the clinical feasibility of the puncture trajectory, improving computational efficiency, and enhancing surgical success rate and patient safety.

[0015] To achieve the above objectives, the present invention provides the following solution:

[0016] A flexible needle puncture path planning method integrating an improved A algorithm and an artificial potential field method includes:

[0017] Acquire medical images, extract target points and obstacles to be avoided from the medical images, and generate a planning coordinate plane;

[0018] In the planned coordinate plane, the model is simplified and nodes are expanded using a puncture needle. Based on a multi-level heuristic function and guided by an artificial potential field, an evaluation function is generated:

[0019] The expanded nodes are evaluated using the evaluation function, and a planned path for flexible needle puncture is generated based on the evaluation results.

[0020] Optionally, establishing a simplified model of the puncture needle includes:

[0021] Obtain the base coordinate system of the puncture environment and the base coordinate system of the needle tip. Determine the kinematic model of the needle tip based on the axial needle insertion direction and the axial needle rotation direction:

[0022] ξ B =v·ξ1+ω·ξ2

[0023] Where, ξ B For the needle tip base coordinate system after defining position and attitude, ξ1 is the axial needle insertion direction, ξ2 is the axial needle rotation direction, v is the input amount for needle base insertion, and ω is the input amount for needle base rotation control.

[0024] Based on the needle tip kinematic model, the relationship between the puncture environment base coordinate system and the needle tip base coordinate system is determined as follows:

[0025]

[0026] Where, ξ B Using a needle tip-based coordinate system, Let α be the first derivative of the position and orientation of the puncture environment base coordinate system, α, β, and γ be the rotation angles about the corresponding coordinate systems, and k be the curvature formed by the single wheel abstracted from the needle tip moving along the radius r during puncture.

[0027] When the curvature of the puncture trajectory is changed during needle insertion while maintaining the desired γ angle, the rotational control value during puncture becomes the target value. Based on the relationship between the puncture environment base coordinate system and the needle tip base coordinate system, the motion model of the puncture needle is simplified, resulting in the simplified model of the puncture needle.

[0028]

[0029] in, It is the first derivative of the rotation angle about the corresponding coordinate system.

[0030] Optionally, generating the planned path for the flexible needle puncture includes:

[0031] Step 1: Initialize the OPEN and CLOSE tables, and place the insertion point into the OPEN table;

[0032] Step 2: Determine the node corresponding to the minimum f value from the OPEN table and judge whether the node corresponding to the minimum f value has reached the target point. If it has, stop the search. If it has not, expand the node in the planning coordinate plane using the puncture needle simplified model and put the expanded node into the CLOSE table.

[0033] Step 3: Set the evaluation function: f(x) = (1-δ)g(x) + δh(x); where f(x) is the evaluation function, δ is the weight coefficient, g(x) is the path length loss, and h(x) is the multi-level heuristic function; evaluate the expanded nodes in the CLOSE table according to the evaluation function;

[0034] Step 4: Place the evaluated node into the OPEN table. Repeat steps 2-3 until the node corresponding to the minimum f value reaches the target point. Backtrack the node that reaches the target point and search for the parent node until backtracking to the needle insertion point to form the planned path of the flexible needle puncture.

[0035] Optionally, the path length loss includes:

[0036] g(x) = i·v·dt

[0037] Where i represents the i-th expansion, and v·dt represents the step size of the expansion.

[0038] Optionally, the multi-level heuristic function includes:

[0039] Directional heuristic function:

[0040]

[0041] Where, α goa l is the direction angle formed by the coordinates of the current node and the coordinates of the target point, α new The kinematic constraint is a discretized direction angle that is uniform within the reachable domain, Δα. max The maximum angle α;

[0042] Heuristic function for target distance:

[0043]

[0044] Among them, y g , z g Let y be the coordinates of the target point. i , z i The coordinates of the current node under the i-th expansion are the node coordinates.

[0045] Path smoothing heuristic function:

[0046] h smooth =|(α′) new -α′)-(α new -α)|

[0047] Where α is the direction angle of the current node, α' new For α new The angle between α and the current α node, α' is α new The angle between the current node B and the current node B.

[0048] Optionally, the multi-level heuristic function further includes:

[0049] Obstacle avoidance heuristic function:

[0050] Define a gravitational field and a repulsive field, and divide the repulsive field into multiple components; the components include: a first component force and a second component force.

[0051] Based on the first component force and the second component force, a final repulsive force is formed. An attractive force is obtained through an artificial potential field. The final repulsive force is combined with the attractive force to obtain a final resultant force.

[0052] The difference between the direction of the final resultant force and the direction of the node to be evaluated is used as the obstacle avoidance heuristic function.

[0053] Optionally, the repulsive field can be divided into multiple components, including:

[0054]

[0055] Among them, F req1 (i) is the first component force, F req2 (i) is the first component force, ρ(q) i ,q obs ) represents the Euclidean distance from the current point's coordinates to the obstacle, ρ0 represents the obstacle's influence distance, and ρ ξ (q i ,q g ), where η is the repulsive field coefficient, a scaling factor used to adjust the intensity of the repulsive field, ρ(q) is the correction term, and η is the repulsive field coefficient. i ,q obs ρ represents the Euclidean distance from the current path point to each obstacle. ξ-1 (q i ,q g ) represents the derivative of the distance correction term, used to calculate the gradient of the repulsive field, q i Let q be the coordinates of the current calculation point in the path planning. obs Let q be the spatial position of the obstacle. g Let ξ be the coordinates of the target point, and ξ be the repulsive field shape adjustment index, which affects the curvature of the repulsive force as a function of distance. Let ρ be the abbreviation for the distance variable in the formula when calculating the partial derivative, and R be the distance variable. k Let R be the radius of the Kth obstacle. k Let K be the radius of the Kth obstacle.

[0056] Optionally, obtaining the attraction includes:

[0057]

[0058] Among them, F att (i) is attraction, For gradient operator, U att (i) represents the attraction potential field, k att This is the attraction gain coefficient.

[0059] Optionally, obtaining the final resultant force includes:

[0060]

[0061] Among them, F total For the final combined force, F att Let F be the attraction, n be the total number of obstacles, and F be the number of obstacles. reqLet be the radius of the k-th obstacle.

[0062] The beneficial effects of this invention are as follows:

[0063] The kinematically constrained expansion without path smoothing steps: The A* algorithm of this invention is a mature heuristic search algorithm widely used in path planning problems since its inception. It continuously evaluates the cost of neighboring nodes and obtains the optimal neighboring node until the target point is reached. Traditional A* algorithms typically operate on known mesh graphs, while the improved A* algorithm node expansion method proposed in this invention simplifies the puncture needle model and incorporates the expansion of non-holonomic kinematic constraints of the flexible needle. The searched nodes are constrained and do not require excessive subsequent curve smoothing.

[0064] Ensuring each step is a highly safe extension: This invention designs a multi-level heuristic function to simultaneously evaluate the cumulative cost of the path and the direction of the potential force during path node extension, and integrates global optimal guidance and local obstacle avoidance response, thus avoiding trajectories with poor safety caused by random search.

[0065] Ensuring a high level of safety at every step: This invention introduces an Artificial Potential Field (APF) guidance mechanism into the multi-level heuristic function design. The "attraction" of the target point and the "repulsion" generated by obstacles in the environment combine to form a potential force. The direction of this resultant force points towards a direction that integrates "approaching the target" and "moving away from obstacles," effectively avoiding the poor safety trajectory caused by random searches.

[0066] Excellent real-time performance: The kinematically constrained expansion eliminates the need for path smoothing, significantly reducing computational and time costs. Furthermore, simultaneous evaluation of cumulative path cost and potential force direction during path node expansion greatly narrows the node search range, leading to faster convergence to the optimum. The significantly reduced path search time maintains excellent real-time performance, making it suitable for intraoperative reprogramming.

[0067] Kinematic constraint compatibility: The improved A* algorithm proposed in this invention integrates the non-holistic kinematic constraints of the flexible needle (such as curvature restrictions and irreversible steering), so that the extended nodes strictly satisfy the device dynamics boundary conditions, and can generate clinically executable paths without additional trajectory smoothing processing.

[0068] Path safety: By improving the A* algorithm and adopting a multi-level heuristic function design, the cumulative cost of the path and the gradient direction of the potential field are simultaneously optimized and evaluated. By superimposing the vector of path guidance and local obstacle repulsion, the risk of suboptimal path caused by random expansion is avoided, ensuring the mechanical rationality of the puncture trajectory and the safe critical distance.

[0069] Real-time performance: The kinematic constraint-driven node expansion mechanism eliminates post-processing steps, significantly reducing computational resource consumption and time overhead; in addition, the dynamic coupling evaluation of path cost and potential vector compresses the node search space, enabling rapid convergence of the target region, ensuring a shorter planning cycle, and meeting the needs for real-time correction during the procedure. Attached Figure Description

[0070] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0071] Figure 1 is a schematic diagram of the puncture needle motion model according to an embodiment of the present invention;

[0072] Figure 2 is a schematic diagram of the duty cycle method according to an embodiment of the present invention;

[0073] Figure 3 is a simplified model diagram of an embodiment of the present invention;

[0074] Figure 4 is a schematic diagram of the multi-step extension of an embodiment of the present invention;

[0075] Figure 5 is a schematic diagram of the forces obtained by the artificial potential field method according to an embodiment of the present invention;

[0076] Figure 6 is a schematic diagram of the artificial potential field method guided by an embodiment of the present invention;

[0077] Figure 7 is a flowchart of a flexible needle puncture path planning method that integrates the improved A algorithm and the artificial potential field method according to an embodiment of the present invention. Detailed Implementation

[0078] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0079] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0080] As shown in Figure 7, this embodiment discloses a flexible needle puncture path planning method that integrates the improved A algorithm and the artificial potential field method. The method includes: acquiring medical images, extracting target points and obstacles to be avoided from the medical images, and generating a planning coordinate plane; expanding nodes in the planning coordinate plane using a simplified model of the puncture needle, generating an evaluation function based on a multi-level heuristic function and the guidance mechanism of an artificial potential field; evaluating the expanded nodes through the evaluation function, and generating a flexible needle puncture planning path based on the evaluation results.

[0081] Furthermore, using information obtained from medical imaging, the target points and obstacles to be avoided are identified, and then placed into the environment of the planned path to generate a planning coordinate plane.

[0082] Furthermore, establishing a simplified model of the puncture needle includes: obtaining the puncture environment base coordinate system and the needle tip base coordinate system; determining the needle tip kinematic model based on the axial needle insertion direction and the axial needle rotation direction; determining the relationship between the puncture environment base coordinate system and the needle tip base coordinate system based on the needle tip kinematic model; when the curvature of the puncture trajectory is changed during needle insertion, while keeping the desired γ angle constant, the rotation control quantity during puncture is the target value, thereby simplifying the motion model of the puncture needle based on the relationship between the puncture environment base coordinate system and the needle tip base coordinate system, and obtaining a simplified model of the puncture needle.

[0083] Specifically, a puncture needle model is established as follows: The flexible puncture needle utilizes the force generated between the beveled tip and the tissue, causing it to form an arc trajectory during puncture based on its own elasticity. A "bicycle" model is used to effectively describe the motion of the puncture needle. This model abstracts the movement of the needle tip as the motion of a bicycle, and can also be simplified to the motion of a "unicycle."

[0084] As shown in Figure 1, in the puncture environment base coordinate system {A} and the needle tip base coordinate system {B}, the position and orientation are defined as follows: ξ A =[x,y,z,α,β,γ] T ξ B =[x ' ,y ' ,z ' ,α ' ,β ' ,γ ' ] T Where x, y, z represent the positions in the corresponding coordinate system, and α, β, γ represent the rotation angles around the corresponding coordinate system. Let ξ1 = [0, 0, 1, k, 0, 0] corresponding to the axial needle advance direction. T ξ2, corresponding to the axial direction of the needle rotation, is [0,0,1,k,0,0]. TWhere k = 1 / r is the curvature k formed by the movement of the needle tip, abstracted as a single wheel, along radius r during puncture, which is determined by the mechanical properties of the puncture needle and the force exerted by the patient's tissue. Let v and ω be the input quantities for needle insertion and rotation control acting on the needle base, respectively, then the kinematic model of the needle tip can be obtained: ξ B =v·ξ1+ω·ξ2; where v is the input amount of the needle base.

[0085] Based on the relationship between the puncture environment base coordinate system {A} and the needle tip base coordinate system {B}, we can obtain: In the formula:

[0086]

[0087] in, The transpose of the rotation matrix between the puncture environment base coordinate system {A} and the needle tip base coordinate system {B} can be used to obtain:

[0088]

[0089] Where, ξ B This defines the tip-based coordinate system after position and orientation are defined.

[0090] During puncture, arcs of varying radii are often required, as a single-radius arc is insufficient to effectively avoid multiple obstacles. Common surgical techniques include rotating the needle during insertion to avoid these obstacles. A method utilizing duty cycles proposes altering the curvature of the puncture trajectory by advancing at high speed along the desired γ angle for a certain time, followed by a non-rotating advance at the γ angle, as shown in Figure 2. High-speed rotation ensures the trajectory remains within the yz two-dimensional plane, satisfying real-time feedback from the ultrasound 2D image. Controlling the rotation time proportion within a single time step yields different curvatures, and the arc formed by completely non-rotating insertion represents the maximum curvature the needle can achieve within that time step. The region between the arcs formed by the maximum curvatures on both sides is the reachable region. The formula is as follows:

[0091]

[0092] Where n is the number, T rotation For the rotation insertion time, T step For a single time step, T insertion This is for direct time insertion.

[0093] The arc formed by using the duty cycle method for needle insertion is equivalent to only changing the curvature k, without escaping the two-dimensional plane. Furthermore, the angle γ remains constant after the needle insertion rotation ends, and no rotation occurs at other times. Therefore, the process can be considered as the rotation control variable being 0. Thus, by simplifying the model, we can obtain:

[0094]

[0095] The discretized model is:

[0096]

[0097] The direction of k is determined by the angle γ. When the needle is inserted into the yz plane, the angle γ is 0° or 180°, that is, the bevel direction of the needle tip is respectively the left and right, as shown in Figure 3.

[0098] Further, the planning path for generating flexible needle puncture includes: Step 1, initializing the OPEN and CLOSE tables, and placing the needle insertion point into the OPEN table; Step 2, determining the node corresponding to the minimum f value from the OPEN table and judging whether the node corresponding to the minimum f value has reached the target point. If it has, the search stops; if not, the node is expanded in the planning coordinate plane using a simplified model of the puncture needle, and the expanded node is placed into the CLOSE table; Step 3, setting an evaluation function and evaluating the expanded nodes in the CLOSE table according to the evaluation function; Step 4, placing the evaluated node into the OPEN table, repeating steps 2-3 until the node corresponding to the minimum f value reaches the target point, backtracking the node that has reached the target point, searching for the parent node, until backtracking to the needle insertion point, thus forming the planning path for flexible needle puncture.

[0099] Specifically: the planned path for generating flexible needle puncture includes:

[0100] Step 1: Initialize the OPEN and CLOSE tables, and place the needle entry point into the OPEN table.

[0101] Step 2: Check if the OPEN table is empty. If it is not empty, and each expanded node will have an f value after evaluation, find the node with the smallest f value in the OPEN table and check if it has reached the target point. If it meets the requirement, stop searching; otherwise, continue expanding.

[0102] Step 3 differs from the traditional A* algorithm in that it requires expansion based on the model established in step 2. Taking the entry point as the starting point, several new α direction angles are uniformly generated within a discrete expansion range that meets the maximum curvature constraint. Based on these direction angles, corresponding coordinates are generated using the discretized model, and the expanded nodes are placed into the CLOSE table.

[0103] As shown in Figure 4, A, B, and C represent nodes on the current planned path, with B being the parent node of C and A being the parent node of B. At node A, the direction angle is α. Based on α, the maximum left and right angles α can be generated, and the included angle between them is Δa. max From several α-direction angles obtained by uniform discretization within the kinematically constrained reachable domain, the selected direction is Δa. new The direction angle, which makes an angle Δα1 with the current α, can be obtained similarly. ' new ,Δα2,α ' , α".

[0104] Step 4: Design the evaluation function f(x) = (1-δ)g(x) + δh(x) to evaluate the expanded nodes in the CLOSE table. g(x) is the path length loss. To improve computational efficiency, the value of g(x) is obtained by multiplying the expansion step size v·dt by the number of times the current node is expanded on the expanded path tree. That is, g(x) = i·v·dt.

[0105] h(x) is the heuristic function in the search algorithm, which consists of multiple levels of heuristics:

[0106] Directional heuristic function: Where α goal The direction angle formed by the coordinates of the current node and the coordinates of the target point.

[0107] Heuristic function for target distance Where y g , z g y and z are the coordinates of the target point, and y and z are the coordinates of the current node.

[0108] Path smoothing heuristic function: h smooth =|(α' new -α')-(α new -α)|, in order to improve computational efficiency, the difference between the two changes in the α angle can reflect the change in the curvature before and after.

[0109] Obstacle avoidance heuristic function: It uses the resultant force direction obtained based on the APF algorithm as a guide, as follows:

[0110] The idea behind the artificial potential field method is to create a virtual gravitational potential field based on the information of the target point, and a repulsive potential field based on the information of the obstacles, starting from the starting point. The path with the fastest gradient descent is found through the interaction of the two.

[0111] The definition of a gravitational field is: Where ρ(q) i ,qg () represents the Euclidean distance from the current point to the target point.

[0112] The definition of a repulsive field is:

[0113]

[0114] Where η is the repulsion coefficient, ρ(q) i ,q obs () represents the Euclidean distance from the current point's coordinates to the obstacle. ρ0 is the distance of influence of the obstacle, R i Let ρ be the radius of the obstacle. To ensure that no situation becomes unreachable, a correction term ρ is introduced. ξ (q i ,q g This allows the final repulsive force to be divided into two components, F. req1 and F req2 This allows one force to move away from the obstacle, while another can generate a virtual thrust to move towards the target point.

[0115] As shown in Figure 5, the forces obtained from the potential field are as follows:

[0116] attraction

[0117] Component 1:

[0118] Component 2:

[0119] As shown in Figure 6, the two component forces eventually form a total repulsive force F. rep .

[0120] The final combined force is: h is defined as the difference between the direction of the resultant force and the direction of the current node to be evaluated. APF .

[0121] Adding the weighting factor, we get h = ε1·h dir +ε2·h dis +ε3·h smooth +ε4·h APF ; where h APF This is a function for obstacle avoidance heuristics.

[0122] Step 5: Add the evaluated points to the OPEN table and return to Step 2.

[0123] Step 6: If the condition of reaching the target point is met, stop searching the path. Then, backtrack step by step from the last point to search the parent node until the needle entry point is reached, thus forming a path.

[0124] The present invention proposes a flexible needle puncture path planning method that integrates the improved A algorithm and the artificial potential field method. It is a path generation method with embedded nonholonomic constraints, which can directly embed the nonholonomic motion constraints of flexible needle curvature into the node extension rules, so that the output path naturally meets the clinical execution requirements without much subsequent processing.

[0125] The present invention proposes a flexible needle puncture path planning method that integrates the improved A algorithm and the artificial potential field method. It introduces a composite loss function driven by multi-objective constraints, which can minimize the risk of trajectory deviation caused by sudden changes in path curvature and insufficient safe distance from obstacles, thereby reducing the probability of clinical complications.

[0126] The present invention proposes a flexible needle puncture path planning method that integrates the improved A algorithm and the artificial potential field method. This is a highly efficient real-time dynamic path generation mechanism that can effectively replace the high computational load and low real-time performance defects of traditional fast expanding random tree algorithm, artificial potential field method and swarm intelligence algorithm.

[0127] The present invention proposes a flexible needle puncture path planning method that integrates the improved A algorithm and the artificial potential field method. It is a real-time processing scheme that combines dynamic pruning and parallel computing. It can quickly re-plan the path according to dynamic scenarios such as intraoperative tissue deformation and instrument displacement, and realize real-time path correction.

[0128] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A flexible needle puncture path planning method integrating an improved A-algorithm and an artificial potential field method, characterized in that, include: Acquire medical images, extract target points and obstacles to be avoided from the medical images, and generate a planning coordinate plane; In the planned coordinate plane, a simplified model is expanded using a puncture needle. Based on a multi-level heuristic function and guided by an artificial potential field, an evaluation function is generated. The multi-level heuristic function includes: a directional heuristic function. ;in, The direction angle formed by the coordinates of the current node and the coordinates of the target point. The kinematically constrained, discretized direction angle is uniform within the reach domain. For the largest Angle; heuristic function for target distance: ;in, ， The coordinates of the target point, ， The coordinates of the current node under the i-th expansion; path smoothing heuristic function: ;in, The direction angle of the current node. for With the current Node angle, for With the current Node angle; Obstacle avoidance heuristic function: Define an attractive field and a repulsive field, and divide the repulsive field into multiple components; the components include: a first component and a second component; based on the first component and the second component, a final repulsive force is formed, and an attractive force is obtained through an artificial potential field. The final repulsive force is combined with the attractive force to obtain the final resultant force; the difference between the direction of the final resultant force and the direction of the node to be evaluated is used as the obstacle avoidance heuristic function; the expanded node is evaluated through the evaluation function, and a flexible needle puncture planning path is generated according to the evaluation result; the generation of the flexible needle puncture planning path includes: Step 1, initializing the OPEN table and the CLOSE table, and placing the needle insertion point into the OPEN table; Step 2, determining the node corresponding to the minimum f value from the OPEN table and judging whether the node corresponding to the minimum f value has reached the target point. If it meets the condition, the search stops; if it does not meet the condition, the node is expanded in the planning coordinate plane using the puncture needle simplified model, and the expanded node is placed into the CLOSE table; Step 3, setting the evaluation function: ;in, For the evaluation function, These are the weighting coefficients. For path length loss, It is a multi-level heuristic function; the nodes in the expanded CLOSE table are evaluated according to the evaluation function; step 4: the evaluated nodes are placed into the OPEN table, and steps 2-3 are repeated until the node corresponding to the minimum f value reaches the target point. The nodes that reach the target point are backtracked to search for the parent node until the needle insertion point is reached, thus forming the planned path of the flexible needle puncture.

2. The flexible needle puncture path planning method integrating the improved A algorithm and the artificial potential field method according to claim 1, characterized in that, Establishing the simplified model of the puncture needle includes: obtaining the puncture environment base coordinate system and the needle tip base coordinate system, and determining the needle tip kinematic model based on the axial needle insertion direction and the axial needle rotation direction. ;in, To define the tip-based coordinate system after determining its position and orientation, This refers to the axial direction of needle insertion. Axial needle direction, The input amount for needle insertion at the base of the needle. The input quantity for needle base rotation control; based on the needle tip kinematic model, the relationship between the puncture environment base coordinate system and the needle tip base coordinate system is determined: ;in, Using a needle tip-based coordinate system, Let be the first derivative of the position and orientation in the puncture environment base coordinate system. Let be the rotation angle about the corresponding coordinate system. Let be the curvature formed by the movement of a single wheel, abstracted from the needle tip, along radius r during puncture; and let be the curvature of the puncture trajectory changed during needle insertion while maintaining the desired curvature. If the angle remains unchanged, the rotation control value during the puncture process is the target value. Therefore, based on the relationship between the puncture environment base coordinate system and the needle tip base coordinate system, the motion model of the puncture needle is simplified, and the simplified model of the puncture needle is obtained: ;in, It is the first derivative of the rotation angle about the corresponding coordinate system.

3. The flexible needle puncture path planning method integrating the improved A algorithm and the artificial potential field method according to claim 1, characterized in that, The path length loss includes: ;in, For the first Secondary expansion, For the step size of the expansion.

4. The flexible needle puncture path planning method integrating the improved A algorithm and the artificial potential field method according to claim 1, characterized in that, The repulsive field is divided into multiple components, including: ;in, As the first component force, As the first component force, The distance of the obstacle's influence. For correction items, This is the repulsive field coefficient, a scaling factor used to adjust the intensity of the repulsive field. This represents the Euclidean distance from the current path point to each obstacle. The derivative of the distance correction term is used to calculate the gradient of the repulsive field. These are the coordinates of the current calculation point in the path planning. The spatial location of the obstacle. The coordinates of the target point, The repulsive field shape adjustment index, To influence the curvature that affects the repulsive force as a function of distance, This is a simplified representation of the distance variable in the formula when calculating partial derivatives. Let K be the radius of the Kth obstacle.

5. The flexible needle puncture path planning method integrating the improved A algorithm and the artificial potential field method according to claim 4, characterized in that, Acquiring the attraction includes: ;in, For attraction, This is the gradient operator. For attraction potential field, This is the attraction gain coefficient.

6. The flexible needle puncture path planning method integrating the improved A algorithm and the artificial potential field method according to claim 5, characterized in that, Obtaining the final resultant force includes: ;in, For the final synergy, For attraction, There are a total of n obstacles. For the first The radius of the obstacle.

Images