A particle distribution source method based on needle channel feasible region constraint and recursive particle removal

Abstract

The application discloses a kind of particle source distribution method and system based on needle path feasible region constraint and recursive particle removal, belong to tumor radiotherapy dose calculation and optimization technical field.The method first obtains patient image data, and the planned target volume, critical organ and bone structure are segmented and three-dimensional reconstruction is carried out, and three-dimensional needle path feasible region model is established;Secondly, a multi-objective puncture angle optimization model is constructed and an improved particle swarm optimization algorithm is used to solve, and the optimal puncture angle of each slice layer is obtained, and the candidate needle position is generated to form the candidate particle placement area;Then, the neighborhood threshold mechanism is used to place particles in the candidate area, the dose gap field is calculated and layered compensation is carried out, and the initial particle solution set is obtained;Finally, a recursive particle removal strategy is used, and the weighted neighborhood distance is used as the criterion to iteratively remove redundant particles, and the dose-volume histogram is used to evaluate and ensure the target coverage requirement, and the final source distribution result is output.The application places needle path implementability constraint in source distribution optimization, and through the synergistic effect of dose gap field driven compensation and recursive particle removal, redundant particles are effectively eliminated under the premise of ensuring target dose coverage, and the clinical implementability, dose distribution uniformity and calculation efficiency of the source distribution scheme are significantly improved.

Description

Technical Field

[0001] This invention relates to the field of brachytherapy and medical image computational optimization technology, specifically to a particle source placement method based on needle path feasible domain constraints and recursive particle removal. Background Technology

[0002] Brachytherapy, a widely used internal radiation therapy method for cancer treatment, works by using high-energy rays released from radioactive isotopes (such as iodine-125 and palladium-103) to disrupt the mitotic processes of tumor cells, thereby achieving a therapeutic effect. Because the radiation range of the radioactive particles is limited, the radiation dose is mainly concentrated in the tumor area, effectively reducing damage to surrounding healthy tissues compared to external beam radiation therapy.

[0003] In radioactive particle implantation therapy, the planning of the particle source location is a core element determining the quality of treatment. Previous methods heavily relied on the physician's clinical experience for both calculating the particle radiation dose and selecting the particle source location. While the advent of radioactive particle implantation therapy planning systems (TPS systems) has assisted physicians in calculating the radionuclide radiation dose, the particle source location still heavily depends on the physician's experience and manual adjustments.

[0004] While existing simulated annealing algorithms offer flexibility and global search capabilities, they suffer from limitations such as long computation times, difficulty in parameter tuning, and the potential to get trapped in local optima. Chinese patent CN118949298B proposes a particle placement method and system for particle implantation therapy. This method accelerates calculations by pre-encapsulating radiation dose templates for individual particles and employs a hybrid optimization algorithm to optimize the placement location, thus improving computational efficiency and placement quality to some extent. However, this method primarily focuses on dosimetric indicators and does not adequately consider the feasibility of actual puncture needle tracts, potentially leading to difficulties in clinical implementation of the placement scheme.

[0005] On the other hand, in the field of needle path planning, Chinese patent application CN202310078661.5 discloses a method and device for optimizing needle paths in afterloading implantation planning. This method achieves automatic optimization of the needle path by constructing a three-dimensional coordinate system, reconstructing the spatial information of the target area and organs at risk, calculating all possible needle paths, and using a penalty function to evaluate the path's merits. However, this method focuses on the geometric feasibility and safety of the needle path, paying insufficient attention to whether the specific distribution of particles within the needle path can achieve optimal dose coverage. Furthermore, the dose distribution evaluation in its penalty function is based on a macroscopic estimation at the needle path level and lacks the ability to guide fine-grained source optimization at the individual particle level.

[0006] Furthermore, existing particle placement methods generally suffer from a common problem: their optimization logic is primarily additive, focusing on "how to increase particles to meet dosage requirements," while lacking a mechanism for "how to systematically remove redundant particles." This easily leads to locally overly dense placement, redundant placement, or uneven dose distribution. Simultaneously, traditional methods often employ a global search strategy across the entire three-dimensional space, resulting in a heavy computational burden, and the optimization results are highly sensitive to the initial particle distribution. While some existing technologies study needle path planning and particle placement separately, none have systematically and synergistically unified needle path feasibility constraints with particle placement optimization.

[0007] In summary, there is an urgent need for a particle distribution method that can incorporate needle path feasibility constraints into the particle distribution optimization process and systematically identify and eliminate redundant particles while ensuring dose coverage, so as to improve the clinical feasibility of distribution schemes, dose distribution uniformity, and treatment plan quality. Summary of the Invention

[0008] To address the shortcomings of existing technologies, this invention provides a particle source placement method based on needle-path feasible region constraints and recursive particle removal.

[0009] To achieve the above-mentioned technical objectives, the technical solution adopted by the present invention is as follows:

[0010] A particle source placement method based on needle-path feasible region constraints and recursive particle removal includes the following steps:

[0011] 1) Acquire patient image data, and segment and reconstruct the planned target area, organs at risk, bony structures, vascular and nerve structures, and candidate needle insertion areas in the image data to establish a three-dimensional needle path feasible domain model.

[0012] 2) Construct a multi-target puncture angle optimization model based on the three-dimensional needle path feasible domain model. The multi-target puncture angle optimization model includes at least target area traversal constraints, endangered organ avoidance constraints, needle path feasibility constraints, and adjacent layer angle continuity constraints.

[0013] 3) An improved particle swarm optimization algorithm is used to solve the multi-target puncture angle optimization model to obtain the optimal puncture angle for each slice layer; the number of iterations, the maximum number of iterations, and the historical best particle source distribution results are initialized.

[0014] 4) Based on the optimal puncture angle, candidate needle positions are generated, and a candidate particle placement area is formed;

[0015] 5) Place particles within the candidate particle placement area using a neighborhood thresholding mechanism and calculate the dose gap field; perform layered compensation on the initial particle placement based on the dose gap field to obtain the initial particle solution set;

[0016] 6) Evaluate the dose-volume histogram of the initial particle set to determine if it meets the preset target coverage requirements: if it does, proceed to step 8); if not, proceed to step 7).

[0017] 7) Compare the current particle solution set with the historical best particle placement result; if the dose-volume histogram evaluation result corresponding to the current particle solution set is better than the historical best particle placement result, then update the historical best particle placement result to the current particle solution set; then increment the current loop count by 1, and determine whether the current loop count has reached the maximum loop count; if the maximum loop count has been reached, output the historical best particle placement result as the final placement scheme; if the maximum loop count has not been reached, return to step 4).

[0018] 8) Initialize the collection Let it be equal to the initial particle solution set, and calculate the... The weighted neighborhood distance values ​​of each particle;

[0019] 9) Sort the neighborhood values ​​in descending order of weighted distance. The particles in the process are removed one by one; and the particles are removed one by one. The remaining particles are evaluated using a dose-volume histogram; if the results still meet the preset target area dose coverage constraints, then... The particle is permanently deleted; otherwise, the removal operation is undone, and the particle is restored to its original state. middle;

[0020] 10) Return until all particles have been verified. As the final particle source distribution scheme.

[0021] Furthermore, the multi-target puncture angle optimization model in step 2) includes the following objective function:

[0022]

[0023] in, For the first The angle of the puncture needle on the slice. Indicates the first The planned target area is in Position on the slice, Indicates the target area at an angle The span length below, Indicates the number of planned target areas. Indicates the first The first organ at risk in the Position on the slice, Indicates the first The distance between an organ at risk and the nearest planned target area For safe distance threshold, Indicates the number of organs at risk. This is a feasibility item for needle insertion. For the continuity of angles between adjacent slices, , , and Indicates the weighting coefficient;

[0024] The objective function aims to minimize the target area span, minimize the radiation exposure of organs at risk, minimize the penalty for needle path infeasibility, and minimize the angle change between adjacent layers. The needle path infeasibility is determined by the accessibility of the needle entry point, avoidance of anatomical restricted areas, needle path length limitations, and instrument posture limitations. Hard constraints are imposed on inaccessible areas, and penalty constraints are imposed on potentially traversable areas.

[0025]

[0026]

[0027] in, The set angle threshold, and This is a preset penalty coefficient.

[0028] Furthermore, the improved particle swarm optimization algorithm includes dynamic inertia weight adjustment, and incorporates information from adjacent slices, local needle path feasibility, and angle continuity during particle velocity updates to suppress abrupt changes in puncture angles between adjacent slices.

[0029]

[0030]

[0031] in, Indicates the first The particle in the first The speed obtained in the second iteration Indicates the first The optimal solution for each particle. Indicates the first The particle in the first The solution obtained in the second iteration. This is the globally optimal solution. , and It is a random number. , and For coefficients, and These are the initial and final sections, representing the start and end points for placing the puncture needle. For correction items, For the feasible region distance field gradient term, The inertia weights vary with the number of iterations:

[0032]

[0033]

[0034]

[0035] in, As the initial inertia weight, To terminate the inertia weight, , As the attenuation factor, The maximum number of iterations, For the trajectory curvature term.

[0036] In step 4), the candidate needle positions satisfy the following relationship:

[0037]

[0038] in, Indicates the first In the layer slice The first planned target area The location of the puncture needle, To achieve the optimal puncture angle, This is the needle spacing coefficient. For the minimum particle spacing, For boundary compensation terms, This represents the offset between adjacent slices. The adjustment amount varies with the number of iterations:

[0039]

[0040] in, This represents the current loop count. The maximum number of loops. Indicates the first The planned target area is in Layer slices, angles The horizontal span length below.

[0041] Furthermore, boundary compensation items The distance between the candidate needle position and the target area boundary, the minimum safe distance between the candidate needle position and the organ at risk, and the margin of the feasible domain boundary of the needle path are jointly determined to keep the candidate needle position within the feasible needle path.

[0042] Furthermore, the neighborhood threshold mechanism described in step 5) includes:

[0043] 51) Based on the minimum particle spacing Initialize the collection Set the set to empty and initialize the minimum neighborhood value. with the maximum neighborhood value ,in:

[0044] ,

[0045] and ;

[0046] 52) Select any one of the candidate particle placement regions as the initial seed point;

[0047] 53) Using the initial seed point as the center, search for the location in the same layer and adjacent layers. and Place particles at candidate points between them, and add the particles to... ;

[0048] 54) For Each particle is judged one by one. Does the neighborhood contain other particles? If so, then from... Remove the particle from the list;

[0049] 55) Judgment Is it empty? If not empty, then from... Choose any particle as the initial seed point for the next round and repeat steps 53) to 54). If the result is empty, return the current source placement result as the initial particle solution set.

[0050] Furthermore, the dose gap field in step 5) is used to characterize the spatial region within the target area where the prescribed dose requirement is not met, and the neighborhood threshold is dynamically updated as the dose gap field changes.

[0051] Furthermore, the method for layered compensation of the initial dose gap is as follows: the candidate point closest to the geometric center of the local dose gap is used as the seed point, and candidate points that meet the updated neighborhood threshold are searched in the same layer and adjacent layers for placement. The local dose gap field is updated after each placement to achieve layer-by-layer compensation of the particle dose gap.

[0052] Furthermore, the expression for the weighted neighborhood distance value in step 8) is:

[0053]

[0054] in, Represents particles With particles The Euclidean distance between them Neighborhood weight function:

[0055]

[0056] For particles Particles in the same layer and adjacent layers, For its total quantity, The maximum neighborhood value .

[0057] The beneficial effects of this invention are as follows:

[0058] 1. Significantly improves the clinical feasibility of the source distribution protocol.

[0059] This invention incorporates anatomical no-entry zones, bony structure avoidance, needle length limitations, and instrument posture constraints into the particle distribution optimization process through three-dimensional needle path feasible domain modeling. This ensures that all candidate needle positions in the generated distribution scheme meet clinical needle insertion requirements, fundamentally solving the defect of existing dose optimization methods that neglect physical constraints, making the distribution scheme unfeasible in actual puncture operations. The scheme can be directly applied to clinical practice without additional manual correction.

[0060] 2. Accurately identify and compensate for target dose cold spots

[0061] This invention accurately identifies spatial regions within the target area that do not meet the prescribed dose requirements through a dose gap field-driven hierarchical compensation mechanism. It uses the candidate point closest to the geometric center of the local dose gap as the seed point for hierarchical compensation and dynamically updates the local dose gap field after each placement until all dose gaps are effectively filled. This overcomes the shortcomings of traditional global optimization algorithms in terms of insufficient coverage of local under-dose regions within the target area.

[0062] 3. Effectively eliminate redundant particles and optimize particle utilization efficiency.

[0063] This invention employs a recursive particle removal strategy, prioritizing the removal of candidate particles from the densest local regions based on weighted neighborhood distance. After each removal, the target coverage requirement is evaluated and verified using a dose-volume histogram. Under the premise of ensuring dose coverage, redundant particles with low contribution to the target dose are systematically identified and eliminated, effectively reducing the number of redundant particles in the source placement scheme, avoiding excessively dense local source placement, and reducing unnecessary irradiation of normal tissues and puncture trauma to patients.

[0064] 4. Significantly improves the uniformity and conformity of dose distribution in the target area.

[0065] This invention controls particle spacing through a neighborhood threshold mechanism to make particle arrangement regular and orderly. Combined with the synergistic effect of dose gap field-driven hierarchical compensation and recursive particle removal strategy, the dose distribution inside the target area is more uniform. At the same time, the recursive removal of redundant particles at the edge greatly improves the fit between the prescription dose line and the target area boundary. The fit between the dose distribution and the target area shape is significantly better than that of existing technologies that only target macroscopic dose indicators.

[0066] 5. Effectively suppresses abrupt changes in the puncture angle of adjacent sections.

[0067] This invention introduces an angle continuity constraint term between adjacent layers into the objective function and uses an improved particle swarm optimization algorithm sensitive to trajectory curvature to solve it. This keeps the puncture angle variation between adjacent slices within a clinically acceptable smooth range, effectively overcoming the angle mutation problem caused by traditional independent layer-by-layer optimization, and improving the overall continuity of the needle path and the feasibility of clinical operation.

[0068] 6. Significantly reduce computing costs and improve optimization efficiency.

[0069] This invention reduces the particle search space from the entire three-dimensional volume to a finite set within the feasible region by constraining the feasible region of the needle path. Combined with the neighborhood threshold mechanism, it effectively controls the particle spacing, significantly reducing the number of candidate positions to be evaluated. The standard dose template pre-encapsulation strategy reduces the dose calculation to a simple accumulation operation, significantly reducing the computational burden of the optimization process. At the same time, the recursive removal strategy further reduces the number of particles in the final solution. The synergistic effect of multiple mechanisms makes the overall computational efficiency significantly better than the existing technology. Attached Figure Description

[0070] Figure 1 This is a flowchart illustrating the overall process of the method of the present invention.

[0071] Figure 2 This is a schematic diagram of the feasible region model for the three-dimensional needle path.

[0072] Figure 3 This is a schematic diagram of the neighborhood threshold mechanism;

[0073] Figure 4 Flowchart for DVH assessment and results output;

[0074] Figure 5 A schematic diagram of the needle placement result. Detailed Implementation

[0075] To make the objectives, technical solutions, and advantages of this application clearer, the application is described and illustrated below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application. All other embodiments obtained by those skilled in the art based on the embodiments provided in this application without inventive effort are within the scope of protection of this application.

[0076] Obviously, the accompanying drawings described below are merely some examples or embodiments of this application. Those skilled in the art can apply this application to other similar scenarios based on these drawings without any inventive effort. Furthermore, it is understood that although the efforts made in this development process may be complex and lengthy, for those skilled in the art related to the content disclosed in this application, any changes to design, manufacturing, or production based on the technical content disclosed in this application are merely conventional technical means and should not be construed as insufficient disclosure of the content of this application.

[0077] This embodiment uses imaging data from an abdominal tumor case as an example to illustrate the complete implementation process of the method of the present invention.

[0078] Figure 1 The overall flow of the method of the present invention is shown. Figures 2 to 5 The following sections illustrate the three-dimensional needle path feasible region model, neighborhood threshold mechanism, DVH evaluation process, and needle placement results. Figures 1 to 5 The specific implementation methods are described in detail.

[0079] like Figure 1 As shown, the overall process of this method is as follows:

[0080] Step 1): Acquire patient image data and establish a 3D needle path feasible region model. Acquire the patient's enhanced CT image data, with a slice thickness of 2.5mm and a total of 65 slices. After importing the image data into the system, the system automatically identifies and segments the planned target area (PTV) and performs 3D reconstruction. Simultaneously, it segments and reconstructs organs at risk (OARs) such as the liver, kidneys, duodenum, and spinal cord, and identifies and models bony structures (ribs, spine), vascular and neural structures, and candidate needle insertion areas. Based on the above 3D reconstruction results, a 3D needle path feasible region model is established.

[0081] The model construction process is as follows: First, the starting and ending surfaces of the needle path are determined based on the skin surface of the candidate needle insertion area and the target area surface; second, anatomical restricted areas are defined by combining the location of bony and vascular / nerve structures; then, feasible needle paths are screened based on needle path length restrictions (not exceeding 200mm) and instrument posture restrictions (puncture angle not exceeding 60°); finally, all needle path sets that meet the conditions are merged to form a three-dimensional feasible needle path domain model. Figure 2 As shown, the three-dimensional needle path feasible domain model includes the target area, organs at risk, bony structures, and feasible needle insertion areas.

[0082] Step 2): Constructing a multi-target puncture angle optimization model. Based on the 3D needle tract feasible region model, a multi-target puncture angle optimization model is constructed. The multi-target puncture angle optimization model includes the following objective function:

[0083]

[0084] in, For the first The angle of the puncture needle on the slice. Indicates the first The planned target area is in Position on the slice, Indicates the target area at an angle The span length below, Indicates the number of planned target areas. Indicates the first The first organ at risk in the Position on the slice, Indicates the first The distance between an organ at risk and the nearest planned target area For safe distance threshold, Indicates the number of organs at risk. This is a feasibility item for needle insertion. For the continuity of angles between adjacent slices, and Indicates the weighting coefficient;

[0085] In this embodiment, This indicates that there is only one planned target area, and the number of organs at risk considered is four, including the liver, kidneys, duodenum, and spinal cord. The angle variation threshold between adjacent layers is 20° to 70°. Set at 5°, the threshold for safe distance from organs. Set to 2.5mm, weighting coefficient , , , The values ​​were set to 0.4, 0.3, 0.15, and 0.15 respectively.

[0086] The objective function aims to minimize the target area span, minimize the radiation exposure of organs at risk, minimize the penalty for needle path infeasibility, and minimize the angle change between adjacent layers. The needle path infeasibility is determined by the accessibility of the needle entry point, avoidance of anatomical restricted areas, needle path length limitations, and instrument posture limitations. Hard constraints are imposed on inaccessible areas, and penalty constraints are imposed on potentially traversable areas.

[0087]

[0088]

[0089] in, The set angle threshold, and This is a preset penalty coefficient.

[0090] In this embodiment, Values . Values .

[0091] Step 3): The PTV is located in 18 slice layers. The optimal puncture angle on each slice is obtained using the improved particle swarm optimization algorithm. Initialize the loop count. Maximum number of loops The historical best particle source distribution result is an empty set.

[0092] Step 4): Generate candidate needle positions based on the optimal puncture angle:

[0093]

[0094] in, Indicates the first In the layer slice The first planned target area The location of the puncture needle, To achieve the optimal puncture angle, This is the needle spacing coefficient. For the minimum particle spacing, For boundary compensation terms, This represents the offset between adjacent slices. As the number of loops increases Correction amount:

[0095]

[0096] in, This represents the current loop count. The maximum number of loops. Indicates the first The planned target area is in Layer slices, angles The horizontal span length below.

[0097] In the initial loop of this embodiment, , , , , .

[0098] Step 5): In this embodiment, a total of 38 needles were placed, forming 95 candidate particle positions. In the neighborhood threshold mechanism, , And obtain the initial particle solution set. .

[0099] Step 6): Perform a dose-volume histogram evaluation on the initial particle set. V100 = 98.12%, meeting the requirement of V100 ≥ 95%, and D90 is 153.72 Gy, meeting the prescribed dose of 145 Gy. Figure 4 As shown, the DVH assessment process is used to determine whether the target coverage requirements are met, and decides whether to proceed to step 8) or step 7 based on the assessment results.

[0100] Step 7): Skip.

[0101] Step 8): Initialize the collection And calculate the weighted neighborhood distance values ​​for 95 particles:

[0102]

[0103] in, Represents particles With particles The Euclidean distance between them Neighborhood weight function:

[0104]

[0105] For particles Particles in the same layer and adjacent layers, For its total quantity, To initialize the maximum neighborhood value, we set it to 22.

[0106] Step 9): After sorting .

[0107] First, from After removing the particle, the dose-volume histogram result for the subsequent particle set is: V100 = 97.45%, satisfying the requirement of V100 ≥ 95%, and D90 is 152.23 Gy, satisfying the prescribed dose of 145 Gy. This indicates that removing the particle has a small impact on the source coverage result, and the source coverage result of the remaining particles can still meet the preset target area dose coverage constraint. The particle will be permanently deleted.

[0108] Next, from After removing the particle, the dose-volume histogram result for the subsequent particle set is: V100 = 94.12%, which does not meet the requirement of V100 ≥ 95%, while D90 is 147.33 Gy, which meets the prescribed dose of 145 Gy. This indicates that removing this particle has a significant impact on the source distribution results. Therefore, the removal operation is revoked, and the particle is restored to its original position. middle.

[0109] The above operations were then performed on the remaining particles.

[0110] Step 10): After processing 95 particles, gather... The remaining 73 particles will be used as the final particle distribution scheme for output. For example... Figure 5 As shown, the final needle placement result in this embodiment is regularly distributed within the target area.

[0111] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. 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.

Claims

1. A particle source placement method based on needle-path feasible region constraints and recursive particle removal, characterized in that, Includes the following steps: 1) Acquire patient image data, and segment and reconstruct the planned target area, organs at risk, bony structures, vascular and nerve structures, and candidate needle insertion areas in the image data to establish a three-dimensional needle path feasible domain model. 2) Construct a multi-target puncture angle optimization model based on the three-dimensional needle path feasible domain model. The multi-target puncture angle optimization model includes at least target area traversal constraints, endangered organ avoidance constraints, needle path feasibility constraints, and adjacent layer angle continuity constraints. 3) The improved particle swarm optimization algorithm is used to solve the multi-target puncture angle optimization model to obtain the optimal puncture angle corresponding to each slice layer; the current number of cycles, the maximum number of cycles, and the historical best particle source distribution results are initialized. 4) Based on the optimal puncture angle, candidate needle positions are generated, and a candidate particle placement area is formed; 5) Place particles within the candidate particle placement area using a neighborhood thresholding mechanism and calculate the dose gap field; perform layered compensation on the initial particle placement based on the dose gap field to obtain the initial particle solution set; 6) Evaluate the dose-volume histogram of the initial particle set to determine if it meets the preset target coverage requirements: if it does, proceed to step 8); if not, proceed to step 7). 7) Compare the current particle solution set with the historical best particle placement result; if the dose-volume histogram evaluation result corresponding to the current particle solution set is better than the historical best particle placement result, then update the historical best particle placement result to the current particle solution set; then increment the current loop count by 1, and determine whether the current loop count has reached the maximum loop count; if the maximum loop count has been reached, output the historical best particle placement result as the final placement scheme; if the maximum loop count has not been reached, return to step 4). 8) Initialize the collection Let it be equal to the initial particle solution set, and calculate the... The weighted neighborhood distance values ​​of each particle; 9) Sort the neighborhood values ​​in descending order of weighted distance. The particles in the process are removed one by one; and the particles are removed one by one. The remaining particles are evaluated using a dose-volume histogram; if the results still meet the preset target area dose coverage constraints, then... The particle is permanently deleted; otherwise, the removal operation is undone, and the particle is restored to its original state. middle; 10) Return until all particles have been verified. As the final particle source distribution scheme.

2. The particle source placement method based on needle-path feasible region constraints and recursive particle removal according to claim 1, characterized in that, The multi-target puncture angle optimization model in step 2) includes the following objective function: in, For the first The angle of the puncture needle on the slice. Indicates the first The planned target area is in Position on the slice, Indicates the target area at an angle The span length below, Indicates the number of planned target areas. Indicates the first The first organ at risk in the Position on the slice, Indicates the first The distance between an organ at risk and the nearest planned target area For safe distance threshold, Indicates the number of organs at risk. This is a feasibility item for needle insertion. This is a constraint term for the continuity of angles between adjacent slices. , , and This represents the weighting coefficient.

3. The particle source placement method based on needle-path feasible region constraints and recursive particle removal according to claim 2, characterized in that, The objective function aims to minimize the target area span, minimize the radiation exposure of organs at risk, minimize the penalty for needle path infeasibility, and minimize the angle change between adjacent layers. The needle path infeasibility is determined by the accessibility of the needle entry point, avoidance of anatomical restricted areas, needle path length limitations, and instrument posture limitations. Hard constraints are imposed on inaccessible areas, and penalty constraints are imposed on potentially traversable areas. in, The set angle threshold, and This is a preset penalty coefficient.

4. The particle source placement method based on needle-path feasible region constraints and recursive particle removal according to claim 1, characterized in that, The improved particle swarm optimization algorithm includes dynamic inertia weight adjustment, and incorporates information on adjacent slices, local needle path feasibility, and angle continuity during particle velocity updates to suppress abrupt changes in puncture angles between adjacent slices. in, Indicates the first The particle in the first The speed obtained in the second iteration Indicates the first The optimal solution for each particle. Indicates the first The particle in the first The solution obtained in the second iteration. This is the globally optimal solution. , and It is a random number. , and For coefficients, and These are the initial and final sections, representing the start and end points for placing the puncture needle. For correction items, For the feasible region distance field gradient term, The inertia weights vary with the number of iterations: in, As the initial inertia weight, To terminate the inertia weight, , As the attenuation factor, The maximum number of iterations, For the trajectory curvature term.

5. The particle source placement method based on needle-path feasible region constraints and recursive particle removal according to claim 1, characterized in that, In step 4), the candidate needle positions satisfy the following relationship: in, Indicates the first In the layer slice The first planned target area The location of the puncture needle, To achieve the optimal puncture angle, This is the needle spacing coefficient. For the minimum particle spacing, For boundary compensation terms, This represents the offset between adjacent slices. As the number of loops increases Correction amount: in, This represents the current loop count. The maximum number of loops. Indicates the first The planned target area is in Layer slices, angles The horizontal span length below.

6. The particle source placement method based on needle-path feasible region constraints and recursive particle removal according to claim 5, characterized in that, The boundary compensation item The distance between the candidate needle position and the target area boundary, the minimum safe distance between the candidate needle position and the organ at risk, and the margin of the feasible domain boundary of the needle path are jointly determined to keep the candidate needle position within the feasible needle path.

7. The particle source placement method based on needle-path feasible region constraints and recursive particle removal according to claim 1, characterized in that, The neighborhood threshold mechanism described in step 5) includes: 51) Based on the minimum particle spacing Initialize the collection Set the set to empty and initialize the minimum neighborhood value. with the maximum neighborhood value ,in: ， and ; 52) Select any one of the candidate particle placement regions as the initial seed point; 53) Using the initial seed point as the center, search for the location in the same layer and adjacent layers. and Place particles at candidate points between them, and add the particles to... ; 54) For Each particle is judged one by one. Does the neighborhood contain other particles? If so, then from... Remove the particle from the list; 55) Judgment Is it empty? If not empty, then from... Choose any particle as the initial seed point for the next round and repeat steps 53) to 54). If the result is empty, return the current source placement result as the initial particle solution set.

8. The particle source placement method based on needle-path feasible region constraints and recursive particle removal according to claim 1, characterized in that, In step 5), the dose gap field is used to characterize the spatial region within the target area where the prescribed dose requirement is not met, and the neighborhood threshold is dynamically updated as the dose gap field changes.

9. A particle source placement method based on needle-path feasible region constraints and recursive particle removal according to claim 8, characterized in that, The method for performing layered compensation of the initial dose gap is as follows: the candidate point closest to the geometric center of the local dose gap is used as the seed point, and candidate points that meet the updated neighborhood threshold are searched in the same layer and adjacent layers for placement. The local dose gap field is updated after each placement to achieve layer-by-layer compensation of the particle dose gap.

10. The particle source placement method based on needle-path feasible region constraints and recursive particle removal according to claim 1, characterized in that, The expression for the weighted neighborhood distance value in step 8) is: in, Represents particles With particles The Euclidean distance between them Neighborhood weight function: For particles Particles in the same layer and adjacent layers, For its total quantity, It is the maximum neighborhood value.

Images