Methods and systems for planning the puncture oocyte retrieval pathway for assisted reproduction

Abstract

This invention relates to the field of surgical procedure planning, specifically to a method and system for planning oocyte retrieval pathways for assisted reproduction. The method includes: updating the position of each particle using a particle swarm optimization algorithm, such that all particles constitute the oocyte retrieval pathway; dividing all particles into several groups and assigning a group evaluation value to each group; obtaining a global evaluation value for all particles based on the needle rotation motion between the linear distribution characteristics of different groups; filtering out deviating and approaching particles based on the change in group evaluation value with respect to the global evaluation value; reorganizing each group according to the deviating and approaching particles to re-obtain each group; and updating the position of each particle again using the re-obtained group. This invention ensures the accuracy of the generated oocyte retrieval pathway, while the optimized oocyte retrieval pathway can address or optimize multiple objectives or problems.

Description

Technical Field

[0001] This invention relates to the field of surgical procedure planning, and more specifically to a method and system for planning the puncture oocyte retrieval pathway for assisted reproduction. Background Technology

[0002] Oocyte retrieval via puncture is a crucial step in assisted reproductive technology (ART), primarily used to obtain mature oocytes during in vitro fertilization-embryo transfer (IVF-ET). Traditional puncture oocyte retrieval is highly dependent on the operator's experience and is technically challenging. Therefore, developing an automated puncture oocyte retrieval pathway planning system has significant clinical and social value. For example, using particle swarm optimization algorithms for preoperative automated planning of the puncture oocyte retrieval pathway can effectively reduce the difficulty of the procedure.

[0003] To ensure efficient and low-risk oocyte retrieval, multiple goals and issues need to be considered or optimized. For example, the following goals need to be achieved: avoiding the myometrium, bladder, intestines, and important blood vessels around the uterus and ovaries to prevent accidental puncture of these structures from causing bleeding, infection, or more serious complications; adopting a "straight-line puncture with internal turning" strategy to optimize the traditional procedure requiring multiple needle insertions and withdrawals into a single needle insertion, thereby shortening the operation time and reducing patient pain and anesthesia risks; and retrieving oocytes from mature follicles whenever possible.

[0004] Because optimizing the oocyte retrieval path requires considering or addressing multiple objectives or issues, it is difficult to accurately generate the oocyte retrieval path when using the particle swarm optimization algorithm. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides a method and system for planning the puncture oocyte retrieval pathway for assisted reproduction.

[0006] The present invention provides a method and system for planning the puncture oocyte retrieval pathway for assisted reproduction, which adopts the following technical solution:

[0007] In a first aspect, the present invention provides a method for planning the puncture oocyte retrieval pathway for assisted reproduction, the method comprising the following steps:

[0008] Several particles are randomly generated in the 3D model of the pelvic cavity, and the position of each particle is updated using a particle swarm optimization algorithm, so that all particles form the puncture and egg retrieval path; the process of updating the position of each particle includes:

[0009] Before updating the position of each particle, all particles are divided into several groups, so that all particles in each group exhibit a linear distribution characteristic; the group evaluation value of each group is obtained based on the maturity of the follicles at the positions of all particles in each group; the global evaluation value of all particles is obtained based on the needle rotation movement between the linear distribution characteristics of different groups.

[0010] The particle swarm optimization algorithm updates the position of each particle using the group evaluation value and the global evaluation value of each group, and records the average displacement generated by each particle. Based on the change of the group evaluation value of each group with the global evaluation value, it selects deviating particles from particles within each group whose average displacement deviates from the linear distribution feature, and selects approaching particles from particles outside each group whose average displacement does not deviate from the linear distribution feature. Based on the deviating particles and approaching particles, each group is reorganized to obtain each group again. The position of each particle is updated again using the group evaluation value of each group and the global evaluation value of the re-obtained group.

[0011] Preferably, the specific steps of dividing all particles into several groups, such that all particles within each group exhibit a linear distribution characteristic, are as follows:

[0012] The Hough transform algorithm is used to detect straight lines among all particles in the ovarian region, resulting in several straight lines. For any straight line, all particles whose distance from the straight line is less than a first preset threshold are grouped together.

[0013] The least squares method is used to refit the particles within the group into a straight line. The distribution line segment of the particles within the group on the refitted straight line is denoted as the linear distribution feature.

[0014] Preferably, the specific steps for obtaining the group evaluation value for each group based on the maturity of the follicles at the locations of all particles in each group are as follows:

[0015] For all particles in any group, the sum of the maturity of all follicles at the location of all particles is denoted as the overall maturity. The mean of the tracing characteristics of all particles in the group is calculated. The group evaluation value of each group is positively correlated with the overall maturity and the mean of the tracing characteristics, respectively. The tracing characteristics of a particle are equal to the distance between the particle and the linear distribution characteristics of each group.

[0016] Preferably, the step of obtaining the global evaluation value of all particles based on the linear distribution characteristics of different groups includes the following specific steps:

[0017] The linear distribution characteristics of any two groups are denoted as line segments A and B; the connection path between line segments A and B is obtained by connecting the closest endpoints of line segments A and B, and the connection path represents the puncture needle rotation action between the linear distribution characteristics of any two groups; the average maturity of all follicles traversed by the connection path is denoted as the first maturity; the connection suitability of any two groups is obtained, and the connection suitability is negatively correlated with the length of the connection path and positively correlated with the first maturity.

[0018] The linear distribution features of the two groups with the highest connection suitability and the connection path are used to form a first sub-path. The first sub-path is extended by sequentially connecting the first sub-path with the linear distribution features of all groups to obtain the final sub-path, so that all connection paths in the final sub-path have the highest connection suitability.

[0019] Connect all particles outside the group to obtain the second sub-path. The second sub-path and the final sub-path are spliced ​​together to form the puncture and egg retrieval path composed of all particles. The global evaluation value is obtained based on the second sub-path and the final sub-path.

[0020] Preferably, the specific steps of recording the average displacement generated by each particle, and selecting deviating particles from particles within each group whose average displacement deviates from the linear distribution characteristic based on the change of the group evaluation value of each group with the global evaluation value, and selecting approaching particles from particles outside each group whose average displacement does not deviate from the linear distribution characteristic, are as follows:

[0021] The position of each particle is updated several times using the group evaluation value and the global evaluation value. Each time the position of each particle is updated, a displacement vector of each particle is recorded. After updating the position of each particle several times, the mean of all displacement vectors of each particle is recorded as the average displacement.

[0022] The average displacement deviates from the linear distribution characteristics of each group and all particles within each group are denoted as set S1; in set S1, particles whose distance from the linear distribution characteristics of each group is greater than a second preset threshold are selected and denoted as deviation particles; wherein the second preset threshold is positively correlated with the change of the group evaluation value of each group with the global evaluation value;

[0023] Among all particles outside each group, the particles that do not deviate from the linear distribution characteristics of each group are denoted as set S2; in set S2, the particles whose distance from the linear distribution characteristics of each group is less than a third preset threshold are selected and denoted as approaching particles.

[0024] Specifically, the position of each particle is updated several times using the group evaluation value and the global evaluation value of each group. After each update of the position of each particle, the group evaluation value and the global evaluation value of each group are calculated. After updating the position of each particle several times, all the group evaluation values ​​obtained by each group are recorded as an evaluation value sequence, and all the global evaluation values ​​obtained are recorded as a global evaluation value sequence. The ratio of the average change of adjacent elements in the evaluation value sequence to the average change of adjacent elements in the global evaluation value sequence is recorded as the group evaluation value change rate of each group. The second preset threshold and the third preset threshold are positively correlated with the group evaluation value change rate of each group.

[0025] Preferably, the specific steps of reorganizing each group based on the deviating particles and approaching particles to obtain each group again include the following:

[0026] Any deviated particle in each group, and any particle in each group that is outside of all deviated particles but within the group, are marked as mutually exclusive particles.

[0027] All particles that deviate from the group but are within the group, and any particle that approaches the group, are labeled as complementary particles.

[0028] After updating the position of each particle using the group evaluation value and the global evaluation value, all particles are regrouped to obtain new groups. Within each newly obtained group, mutually exclusive particles are removed. It is then determined whether the complementary particle of any particle in each group belongs to that group. If not, the complementary particle of that particle is added to that group.

[0029] Preferably, the method of extending the first sub-path to obtain the final sub-path by sequentially connecting the first sub-path with the linear distribution characteristics of all groups, so that all connected paths in the final sub-path have the maximum connection suitability, includes the following specific steps:

[0030] D1: The groups corresponding to the two linear distribution features that constitute the first sub-path are denoted as G1 and G2; among all groups other than G1 and G2, the group with the maximum connection suitability with G2 is denoted as G3, and the group with the maximum connection suitability with G1 is denoted as G4.

[0031] When the connection suitability between G3 and G2 is greater than or equal to the connection suitability between G4 and G1, G3 is recorded as the group to be connected; otherwise, G4 is recorded as the group to be connected. The group to be connected is denoted as g. The linear distribution features in G1, G2, and g, as well as the connection paths between the linear distribution features in G1, G2, and g, constitute the path after the extension of the first sub-path.

[0032] D2: Extend the first sub-path and use it as the first sub-path again. Then repeat D1 several times and record the resulting extended first sub-path as the final sub-path.

[0033] Preferably, the specific steps for updating the position of each particle using the group evaluation value and the global evaluation value are as follows:

[0034] Update the position of each particle using the position and velocity of any given particle.

[0035] The difference between the optimal value of each group and the evaluation value of each group is denoted as group_t; the difference between the global optimal value and the global evaluation value is denoted as global_t; the magnitude of the particle's velocity is positively correlated with group_t and global_t.

[0036] The optimal value of each group is equal to the maximum value of the group evaluation value of each group during the history of the particle swarm algorithm updating all particle positions; the global optimal value is equal to the maximum value of the global evaluation value during the history of the particle swarm algorithm updating all particle positions.

[0037] Preferably, the specific steps for obtaining the global evaluation value based on the second sub-path and the final sub-path are as follows:

[0038] The sum of the connection suitability of all connecting paths in the final sub-path is recorded as the first evaluation value; the ratio of the length of the second sub-path to the length of the oocyte retrieval path is recorded as x, and exp(-x) is recorded as the second evaluation value, where exp() represents an exponential function with the natural constant as the base;

[0039] The overall evaluation value is positively correlated with the first evaluation value and the second evaluation value, respectively.

[0040] Secondly, the present invention provides a puncture oocyte retrieval pathway planning system for assisted reproduction, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes a puncture oocyte retrieval pathway planning method for assisted reproduction when running the computer program.

[0041] The beneficial effects of the technical solution of the present invention are:

[0042] Compared to directly using particle swarm optimization (PSO) for oocyte retrieval path planning, this invention divides all particles into several groups before updating the position of each particle, ensuring that all particles within each group exhibit a linear distribution. It then obtains a group evaluation value for each group based on the maturity of the follicles at the locations of all particles within that group. Finally, it obtains a global evaluation value for all particles based on the needle rotation motion between the linear distribution characteristics of different groups. Finally, it plans the oocyte retrieval path based on the group evaluation values ​​and the global evaluation values. This process analyzes a local number of particles as a whole (i.e., as a group), allocating a portion of the issues to be considered or optimized in the global evaluation value of the oocyte retrieval path to the group evaluation value for optimization. This initially avoids the problem of difficulty in quickly and accurately obtaining the optimal result when directly using the global evaluation value for PSO optimization, as the issues to be considered or optimized come from multiple aspects.

[0043] Furthermore, based on the change in the group evaluation value of each group with the global evaluation value, this invention filters out deviating particles from particles within each group whose average displacement deviates from the linear distribution characteristic, and filters out approaching particles from particles outside each group whose average displacement does not deviate from the linear distribution characteristic. Each group is then reorganized based on the deviating and approaching particles to obtain a new group. The position of each particle is updated again using the group evaluation value of the newly obtained group and the global evaluation value. In this process, deviating particles are detrimental to further optimization of each group when optimizing the oocyte retrieval path (i.e., they are detrimental to ensuring that all particles within each group exhibit a stable, linear distribution near mature follicles). Approaching particles in this process refer to particles that are beneficial for further optimization of each group when optimizing the oocyte retrieval path but are not included in each group. This invention reorganizes groups using deviating and approaching particles, which not only distributes a portion of the problems that need to be considered or optimized in the global evaluation value of the egg retrieval path to the group evaluation value for optimization, but also avoids the problem of difficulty in quickly and accurately obtaining the optimal result when directly using the global evaluation value for particle swarm optimization, because the problems that need to be considered or optimized come from multiple aspects; it also avoids the problem of easily obtaining local optimal solutions or losing the global optimal solution when optimizing local particles as a whole, which restricts the degree of freedom of each particle as an independent individual.

[0044] Ultimately, this invention ensures the accuracy of the generated oocyte retrieval path, while the optimized oocyte retrieval path can take into account or optimize multiple objectives or problems. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art 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.

[0046] Figure 1 This is a flowchart illustrating the steps of a puncture oocyte retrieval pathway planning method for assisted reproduction, provided in one embodiment of the present invention. Detailed Implementation

[0047] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the puncture oocyte retrieval pathway planning method and system for assisted reproduction proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0048] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0049] The following description, in conjunction with the accompanying drawings, details the specific scheme of the puncture oocyte retrieval path planning method and system for assisted reproduction provided by this invention.

[0050] Example 1:

[0051] This embodiment provides a puncture oocyte retrieval pathway planning system for assisted reproduction. The system is used for pathway planning before puncture oocyte retrieval. The system includes a high-frequency (usually ≥7.5MHz) ultrasound imaging instrument and a three-dimensional volume vaginal probe, which are used to perform multi-section scanning of the patient's pelvis, reconstruct a three-dimensional model of the pelvis, and include the ovaries and nearby tissues such as the bladder, intestines, uterus and important blood vessels. The three-dimensional model of the pelvis is also visualized.

[0052] The system also includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When running, the computer program reads a puncture oocyte retrieval path planning method based on a 3D pelvic model and executes this method for assisted reproduction. This method generates an optimal puncture oocyte retrieval path based on the distribution and physiological state of various tissues in the 3D pelvic model.

[0053] Comparative Example:

[0054] This example presents a general method for planning the puncture egg retrieval path in assisted reproduction. This method uses a particle swarm optimization algorithm to optimize the puncture egg retrieval path.

[0055] The specific process includes:

[0056] Step S001: Randomly generate several particles in the 3D model of the pelvis.

[0057] Specifically, a Region of Interest (ROI) is artificially drawn in the 3D model of the pelvis. This ROI is a three-dimensional spatial region containing tissues that can be punctured; for example, in this embodiment, the entire area outside the bone region is designated as the ROI. Multiple particles are randomly generated within the ROI; in this embodiment, 300 particles are randomly generated. Each particle corresponds to a unit coordinate in the 3D pelvic model, and its position is recorded. An initial velocity is randomly assigned to each particle.

[0058] Step S002: Calculate the evaluation value of each particle and the global evaluation value of all particles.

[0059] The evaluation value of each particle and the global evaluation value of all particles are obtained based on the position of each particle. The evaluation value of each particle describes whether the position of each particle is optimal. In this embodiment, the larger the evaluation value of each particle, the more optimal the position of each particle. The global evaluation value describes whether the positions of all particles are optimal overall. In this embodiment, the puncture oocyte retrieval path refers to the path formed by the positions of all particles. Therefore, the global evaluation value is also used to describe whether the generated puncture oocyte retrieval path is optimal. In this embodiment, the larger the global evaluation value, the more optimal the puncture oocyte retrieval path is.

[0060] Step S003: Update the position of each particle.

[0061] (1) Calculate the new position of any particle using its position and the velocity it is assigned (i.e., update the position of the particle), and denot the new position of the particle as the updated position of the particle.

[0062] For all particles, obtain the updated positions of each particle individually.

[0063] (2) Re-obtain the evaluation value of each particle and the global evaluation value of all particles based on the updated position of each particle.

[0064] The maximum value between the re-acquired evaluation value and the individual optimal value of each particle is taken as the individual optimal value of that particle. The maximum value between the re-acquired global evaluation value and the global optimal value is taken as the global optimal value. The position of each particle when it obtains its individual optimal value is recorded as its optimal position.

[0065] This process is equivalent to taking the maximum value of the evaluation value of each particle in the historical update process and the maximum value of the global evaluation value as the optimal value of the evaluation value of each particle (i.e., the individual optimal value) and the optimal value of the global evaluation value (i.e., the global optimal value).

[0066] It should be noted that the initial values ​​of the individual optimal value and the global optimal value of each particle are both 0. That is, if the individual optimal value and the global optimal value of each particle are not read, then 0 is used as the individual optimal value and the global optimal value of each particle.

[0067] (3) Reacquire the velocity of each particle based on the individual optimal value and the optimal global target value (i.e., update the velocity of each particle).

[0068] (4) Repeat (1) to (3) above until the position of each particle no longer changes, or until the number of repetitions exceeds the set threshold.

[0069] (5) The path formed by all the particles is the optimal path for puncturing and retrieving eggs.

[0070] This concludes the example.

[0071] The above embodiment is a common method for applying particle swarm optimization to the planning of the puncture egg retrieval path. In this process, the particle positions are continuously updated by repeatedly executing (1) to (3). That is, each execution of (1) to (3) can update the position of all particles once, so that the position of each particle and the puncture egg retrieval path formed by all particles are optimal.

[0072] Example 2:

[0073] In this embodiment, to ensure efficient and low-risk oocyte retrieval during the procedure, the following objectives or issues need to be considered or optimized: Firstly, it is necessary to avoid the uterine muscle layer, bladder, intestines, and important blood vessels surrounding the uterus and ovaries to prevent accidental puncture of these structures from causing bleeding, infection, or more serious complications. Secondly, a "straight-line puncture with internal turning" strategy needs to be adopted to optimize the traditional procedure requiring multiple needle insertions and withdrawals into a single needle insertion, thereby shortening the operation time, reducing patient pain and anesthesia risks. Additionally, it is necessary to retrieve oocytes from mature follicles whenever possible.

[0074] This makes the process of obtaining the global evaluation value corresponding to the puncture oocyte retrieval path relatively complex, and there are many issues that need to be considered and optimized. When using the particle swarm optimization algorithm to update the position of particles, it is difficult to make all particle positions reach the optimal, which is not conducive to generating the puncture oocyte retrieval path quickly and accurately.

[0075] Based on this, such as Figure 1As shown, the puncture oocyte retrieval pathway planning method for assisted reproduction provided in this embodiment specifically includes:

[0076] Step S201: Divide all particles into several groups so that all particles in each group exhibit a linear distribution characteristic; obtain the group evaluation value of each group based on the maturity of the follicles at the locations of all particles in each group.

[0077] In this step, after randomly generating a number of particles, the particles are divided into several groups. The purpose is that, considering that a reliable puncture oocyte retrieval path consists of a large number of particles, it is necessary not only to optimize the position of each particle separately, but also to ensure that the puncture oocyte retrieval path can take into account or optimize the above-mentioned multiple issues. This makes it difficult to quickly and accurately obtain the optimal result when optimizing the particle swarm algorithm based on the global evaluation value of the puncture oocyte retrieval path (that is, it is difficult for the global evaluation value to quickly reach the global optimum).

[0078] In this embodiment, the particles are grouped, and each group corresponds to a group evaluation value. This group evaluation value is used to describe whether a local number of particles in the group are in the optimal position. By analyzing the local number of particles as a whole (i.e., as a group), a portion of the problems that need to be considered or optimized in the global evaluation value of the puncture egg retrieval path are distributed and assigned to the group evaluation value of each group for optimization. This initially avoids the problem that it is difficult to obtain the optimal result quickly and accurately when directly using the global evaluation value for particle swarm optimization, because the problems that need to be considered or optimized come from multiple aspects.

[0079] Furthermore, in this embodiment, when grouping, the particles are divided according to their linear distribution, and a group evaluation value is calculated based on the maturity of the follicles. The puncture path formed by the linearly distributed particles helps to collect as many mature follicles as possible along the same puncture induction line during puncture, avoiding repeated punctures of the vaginal wall and increasing oocyte retrieval efficiency.

[0080] As an example, methods for dividing particles into several groups include:

[0081] All particles located within the ovarian region were treated as follows:

[0082] (1) The Hough transform algorithm is used to detect straight lines among these particles, and several straight lines are obtained. It should be noted that the hough_transform_3d operator in Python implements the Hough transform algorithm, and the Hough transform algorithm will not be described in detail in this embodiment.

[0083] For any straight line, obtain the distance between any particle and the straight line, and record it as the first distance of the particle. Obtain all particles whose first distance is less than the first preset threshold th1. These particles are distributed near the straight line, that is, these particles exhibit the characteristic of being distributed along the straight line.

[0084] As an example, the method for obtaining th1 is as follows: obtain the average of the maximum diameter of all follicles in the ovary, which is abbreviated as the average follicle diameter, and take 75% of the average follicle diameter as the value of th1.

[0085] (2) Then, the least squares method is used to refit these particles into a straight line, and the projections of these particles onto the refitted straight line are obtained. The two particles with the largest and smallest projections are obtained, and the projection points of these two particles on the straight line are recorded as the two endpoints. The line segment between these two endpoints on the refitted straight line is recorded as the distribution of these particles along the straight line, which is simply called the straight line distribution feature (that is, the straight line distribution feature is a line segment). These particles are grouped together. At the same time, the distance of each particle to the straight line containing the line segment is recorded as the approach feature of each particle.

[0086] The method for obtaining the projection is as follows: the position of any particle is regarded as a vector, and the inner product of this vector and the unit vector of the line is the projection of any particle onto the line.

[0087] (3) For all groups detected by the Hough transform algorithm, the above method is used to group them respectively.

[0088] Thus, multiple groups are obtained, each group corresponding to a line segment, representing the linear distribution characteristics of all particles within each group.

[0089] Specifically, if the same particle is in multiple groups simultaneously, the tendency features of the particle in each group are obtained, and the particle is assigned to the group with the smallest tendency features. In other embodiments, the particle may be randomly selected to add to a group, ensuring that the same particle is not in multiple groups at the same time.

[0090] As an example, each group corresponds to a group evaluation value, and the steps include:

[0091] For all particles in any group, obtain all follicles at the location of these particles. The sum of the maturity of all follicles is recorded as the overall maturity. In this embodiment, the maturity of follicles is represented by a value in [0, 1] (without dimension). The larger the value, the more mature the follicle.

[0092] Calculate the mean of the approach characteristics of all particles within the group.

[0093] The group evaluation value of each group is positively correlated with the mean of the overall maturity and the convergence feature.

[0094] The higher the group evaluation value of each group, the more optimal each group is. In other words, when performing punctures along the linear distribution characteristics of each group, multiple mature ovaries can be punctured in the same line, resulting in high puncture efficiency.

[0095] Step S202: Obtain the global evaluation value of all particles based on the puncture needle rotation action between the linear distribution characteristics of different groups.

[0096] This step takes into account the complex distribution and maturation of follicles within the ovary, making it difficult to simultaneously ensure a large number and high maturity of follicles when performing punctures in the same direction. In such cases, it is often necessary to change the puncture direction. Based on this, this embodiment obtains the global evaluation value of all particles according to the puncture needle rotation movement between the linear distribution characteristics of different groups.

[0097] The higher the global evaluation value, the more sub-paths with linear distribution characteristics the puncture path composed of all particles can pass through, and the more appropriate puncture needle rotation action can be achieved when adjusting the puncture direction between different linear distribution characteristics. This ensures that the puncture path does not cause serious damage to the patient's tissues and reduces the risk during puncture.

[0098] As an example, the global evaluation value of all particles is obtained based on the needle rotation motion between the linear distribution characteristics of different groups, including the following methods:

[0099] (1) For any two groups, the linear distribution characteristics of these two groups are denoted as line segments A and B. Connect the endpoints of line segments A and B that are closest to each other to obtain the connection path between line segments A and B. This connection path represents the path traversed when the puncture needle rotates.

[0100] The average maturity of all follicles traversed by the connecting path is denoted as the first maturity level (if no follicles are traversed, the first maturity level is denoted as 1).

[0101] The connection suitability of any two groups is obtained. The connection suitability is negatively correlated with the length of the connection path and positively correlated with the first maturity. When the length of the connection path is smaller and the first maturity is larger, it means that when the puncture needle turns from the linear distribution feature in one group to the linear distribution feature in another group, it is easier to operate and less likely to damage immature follicles. At this time, the puncture needle rotation process is more suitable.

[0102] (2) Obtain the two groups with the highest connection suitability. The linear distribution characteristics of these two groups and the connection path between them constitute a first sub-path. Extend the first sub-path to include:

[0103] (2-1) The two groups corresponding to the straight line distribution characteristics at both ends of the first sub-path are denoted as G1 and G2;

[0104] Among all packets other than G1 and G2, the packet with the highest connection suitability with G2 is denoted as G3, and the packet with the highest connection suitability with G1 is denoted as G4.

[0105] It should be noted that, assuming the two nearest endpoints of the linear distribution features in G1 and G2 are A and B respectively; the endpoint connecting the linear distribution feature in G3 to the linear distribution feature in G2 is another endpoint besides B, meaning G1, G2, and G3 can be connected end-to-end via a connecting path. Similarly, the endpoint connecting the linear distribution feature in G4 to the linear distribution feature in G1 is another endpoint besides A, meaning G4, G1, and G2 can be connected end-to-end via a connecting path.

[0106] Furthermore, regarding the connection suitability of G3 and G2, and the connection suitability of G4 and G1, if the former is greater than or equal to the latter, then G3 is recorded as a group to be connected; if the latter is greater than the former, then G4 is recorded as a group to be connected.

[0107] Let g be the group to be connected, and let g be the path formed by the linear distribution characteristics of G1, G2, and g and the connection path between the three. Let g be the path after the extension of the first sub-path.

[0108] (2-2) Then the extended path of the first sub-path is used as the first sub-path again, and then (2-1) is repeated several times (e.g., 2 times). The final extended path of the first sub-path is recorded as the final sub-path.

[0109] In this process, the first sub-path is used as a basis to connect with the linear distribution characteristics of other groups, so that the final sub-path after the extension of the first sub-path has the maximum connection suitability. At this time, all the connecting paths contained in the final sub-path represent the rotation path when the puncture needle rotates (that is, the running path when turning from the linear distribution characteristics in one group to the linear distribution characteristics in another group).

[0110] (3) The sum of the connection suitability of all connected paths in the final sub-path is recorded as the first evaluation value.

[0111] In this embodiment, the puncture point is outside the ovary, specifically on the posterior fornix of the vagina (the specific location is specified by the puncture surgeon); using all particles outside the group, starting from the puncture point, the nearest particles are connected sequentially until the endpoint of the final sub-path is reached. The puncture path formed by this connection is called the second sub-path. The second sub-path and the final sub-path constitute the complete puncture oocyte retrieval path.

[0112] The ratio of the length of the second sub-path to the length of the oocyte retrieval path is denoted as x, and exp(-x) is denoted as the second evaluation value, where exp represents an exponential function with the natural constant as the base. The purpose of using the length of the oocyte retrieval path as the denominator to calculate the ratio is to remove the dimensions and orders of magnitude. The smaller the length of the second sub-path (i.e., the smaller x), the larger the second evaluation value, indicating that the second sub-path is more optimal.

[0113] The global assessment values ​​of all particles (or the puncture oocyte retrieval pathway) were positively correlated with the first assessment value and the second assessment value, respectively.

[0114] Step S203: The particle swarm optimization algorithm updates the position of each particle based on the evaluation value of each particle, the evaluation value of each group, and the global evaluation value.

[0115] Obtain the evaluation value for each particle, including:

[0116] Obtain the location of each particle and determine whether there are important blood vessels surrounding the uterus and ovaries at that location. If they are, set the evaluation value of each particle to 0; otherwise, perform the following checks:

[0117] Determine whether the location is within a follicle. If so, set the evaluation value of each particle to the follicle maturity level; otherwise, set the evaluation value of each particle to 1 (indicating that puncture at this location will not damage the relevant tissue).

[0118] At this point, based on the positions of all particles, we have obtained the evaluation value for each particle, the global evaluation value for all particles, and the group evaluation value for each group. The maximum value between each particle's evaluation value and its individual optimal value is used as the individual optimal value for that particle. The maximum value between the global evaluation value and the global optimal value is used as the global optimal value. The maximum value between each group's evaluation value and its group optimal value is used as the group optimal value for that group. The position of each particle when it achieves its individual optimal value is recorded as its optimal position.

[0119] This process is equivalent to taking the maximum value of the evaluation value of each particle in the historical update process, the maximum value of the evaluation value of each group, and the maximum value of the global evaluation value as the optimal value of the evaluation value of each particle (i.e., the individual optimal value), the optimal value of the evaluation value of each group (i.e., the group optimal value), and the optimal value of the global evaluation value (i.e., the global optimal value).

[0120] It should be noted that the initial values ​​of the individual optimal value, the global optimal value, and the group optimal value of each group are all 0.

[0121] Furthermore, the positions of all particles are updated based on the evaluation value of each particle, the group evaluation value of each group, and the global evaluation value.

[0122] Step S204: Record the average displacement generated by each particle. Based on the change of the group evaluation value of each group with the global evaluation value, filter out the deviating particles from the particles within each group whose average displacement deviates from the linear distribution characteristics, and filter out the approaching particles from the particles outside each group whose average displacement does not deviate from the linear distribution characteristics.

[0123] It should be noted that after all particles update their positions, the particles contained in each group remain unchanged; only the position of each particle within the group changes.

[0124] (1) For all particles in each group, reacquire the linear distribution characteristics of each group and the approach characteristics of each particle according to the method in (2) of step S201. And reacquire the global evaluation value according to step S202, and then update the position of all particles according to step S103.

[0125] (2) Repeat (1) several times; as an example, repeat 5 times (including the execution process of steps S201 to S203). During this repeated execution, the same particle is updated multiple times in position.

[0126] For these position update processes of the same particle, obtain the displacement from the position before each update to the position after each update, obtain the average displacement of all position updates, and record it as the average displacement of each particle.

[0127] During this repeated execution, the group evaluation values ​​obtained for each group are arranged in the order of acquisition, resulting in a sequence of evaluation values ​​for each group. The obtained global evaluation values ​​are then arranged in the order of acquisition, resulting in a sequence of global evaluation values.

[0128] (3) After repeating step (1) several times:

[0129] The average change of adjacent elements in the evaluation value sequence is denoted as A1, and the average change of adjacent elements in the global evaluation value sequence is denoted as A2. The ratio of A1 to A2 is recorded as the group evaluation value change rate for each group, representing the average increase in the group evaluation value of each group as the global evaluation value gradually increases. The larger the average increase in change, the more suitable each group is, and the more it helps to quickly optimize the oocyte retrieval pathway.

[0130] (4) After repeating step (1) several times, for all particles in each group, determine whether the average displacement of each particle deviates from the linear distribution characteristics of each group. All particles whose average displacement deviates from the linear distribution characteristics of each group and are in each group are denoted as set S1. Each particle in set S1 corresponds to a tendency feature (see step S201 for details). Particles in set S1 whose tendency feature is greater than the second preset threshold th2 are denoted as deviation particles; where th2 is positively correlated with the rate of change of the group evaluation value of each group.

[0131] Deviations in the process hinder further optimization of each group when optimizing the oocyte retrieval path (i.e., it prevents all particles within each group from exhibiting a stable, linear distribution near mature follicles). Furthermore, a higher rate of change in the group evaluation value indicates that the particles within each group still have the potential for further optimization, meaning that particles deviating from the linear distribution characteristics within each group are less likely to be considered as deviating particles that hinder further optimization of that group.

[0132] Furthermore, for all particles outside each group, it is determined whether the average displacement of each particle deviates from the linear distribution characteristics of each group. Particles that do not deviate from the linear distribution characteristics of each group are denoted as set S2. Each particle in set S2 corresponds to a tendency feature (see step S201 for details). Particles in set S2 whose tendency feature is less than a third preset threshold th3 are denoted as tendency particles; where th3 is positively correlated with the rate of change of the group evaluation value of each group.

[0133] In this process, the approaching particles represent particles that, while optimizing the oocyte retrieval path, are beneficial for further optimization of each group but are not included in each group. Furthermore, the greater the rate of change in the group evaluation value for each group, the greater the potential of particles outside each group to further optimize that group. This indicates a greater need to include more particles outside each group that approximate a linear distribution as particles that promote further optimization of each group.

[0134] At this point, each group corresponds to several approaching particles and deviating particles.

[0135] In a special case, if the same particle is recorded as a symptotic particle in multiple groups, one group is randomly selected as its symptotic particle, and it is no longer used as a symptotic particle in other groups.

[0136] Step S205: Reorganize each group according to the deviating particles and approaching particles to obtain each group again, and update the position of each particle again using the group evaluation value of each group and the global evaluation value.

[0137] In the above process, after dividing the particles into several groups, a portion of the optimization problem from various aspects is distributed to the local particles in each group for optimization. This avoids the problem of difficulty in achieving global optimum for all particle positions when there are too many optimization problems and optimization is based solely on the global evaluation value. However, optimizing the local particles (i.e., the particles within each group) as a whole restricts the degree of freedom of each particle as an independent individual (i.e., particles cannot perform large-scale optimization or independent optimization when they are individual), which can easily lead to local optima or the loss of the global optimum.

[0138] Based on this, this embodiment performs the following steps:

[0139] (1) For any one deviating particle in each group and any other particle outside the group but within the group, the two particles are labeled as mutually exclusive particles.

[0140] For each deviating particle within a group, for any particle outside of all deviating particles but within a group, and for any approaching particle within a group, these two particles are labeled as complementary particles.

[0141] (2) For each particle and its position, regroup all particles using the methods (1) to (3) in step S201 to obtain new groups. Within each new group, randomly select two particles and determine whether they are marked as mutually exclusive particles. If they are, remove these two particles from each group; otherwise, do not remove them. By removing mutually exclusive particles, it is ensured that any two particles within each group are no longer mutually exclusive particles.

[0142] Furthermore, it is determined whether the complementary particle of any particle in each group belongs to that group. If not, the complementary particle of that particle is added to that group. If it is, no addition is made.

[0143] In a specific case, if the complementary particle of any particle in a certain group is also in another group, the group in which the complementary particle is located and that group are denoted as the complementary group. Randomly select a complementary group and add the particle and its complementary particle to that complementary group, so that the complementary particles are in the same group.

[0144] Thus, the reorganization of each group was achieved through the above (1)~(5). A portion of the problems that need to be considered or optimized in the global evaluation value of the puncture oocyte retrieval path is distributed and assigned to the group evaluation value of each group for optimization. This not only avoids the problem of difficulty in quickly and accurately obtaining the optimal result when directly using the global evaluation value for particle swarm optimization, due to the multiple aspects of the problems that need to be considered or optimized; it also avoids the problem of easily obtaining local optimal solutions or losing the global optimal solution when optimizing local particles as a whole, which restricts the degree of freedom of each particle as an independent individual.

[0145] Then, steps S204 to S205 are repeated, continuously updating the position of each particle and simultaneously regrouping each group. When the positions of all particles no longer change, or when the positions of each particle have been updated several times (this embodiment uses 1000 updates as an example), the path formed by all particles is taken as the final puncture and oocyte retrieval path. This path is displayed in the pelvic 3D model, and the surgeon will perform the puncture and oocyte retrieval based on this path and the real-time display of the 3D model. The specific puncture and oocyte retrieval process is well-known and will not be described in detail in this embodiment.

[0146] This concludes the example.

[0147] Example 3:

[0148] In Example 2, step S201 includes: the group evaluation value of each group is positively correlated with the mean of the overall maturity and the convergence feature, respectively. As an example, the specific formulas are as follows:

[0149]

[0150] Where P0 represents the group evaluation value of each group, L1 represents the overall maturity, and L2 represents the mean of the convergent features; the purpose of using th1 as the denominator is to remove the dimensions and order of magnitude of L2.

[0151] In Example 2, step S202 includes: connecting the closest endpoints of line segments A and B to obtain a connection path between line segments A and B. This example provides another method for obtaining a connection path, ensuring a smooth connection path. As an example, this method includes:

[0152] Let the closest endpoints of line segments A and B be denoted as a and b. Extend line segment A from endpoint a to point a1, and extend line segment B from endpoint b to point b1. Connect a1 and b1 to obtain the first line segment. Draw an arc starting from endpoint a, tangent to the first line segment at point a2, and also tangent to line segment A at endpoint a. Draw an arc starting from endpoint b, tangent to the first line segment at point b2, and also tangent to line segment B at endpoint b. These two arcs, along with the line segment connecting points a2 and b2, form a smooth connecting path.

[0153] In this embodiment, when extending line segments A and B, the extension length of th1 is used as an example for description.

[0154] In other embodiments, other methods may be used to ensure that the connection path between line segments A and B is smooth, so as to guide the puncture needle to puncture smoothly.

[0155] In Example 2, step S202 includes: obtaining the connection suitability of any two groups, where the connection suitability is negatively correlated with the length of the connection path and positively correlated with the first maturity; as an example, the specific formula is as follows:

[0156]

[0157] in Indicates the suitability of the connection. Indicates the first stage of maturity. L4 represents the length of the connecting path, and L4 represents the average length of the linear distribution characteristics of any two groups. The purpose of using L4 as the denominator is to remove the dimensions and order of magnitude of L3.

[0158] In Example 2, step S202 includes: the global evaluation value of all particles (or the oocyte retrieval path) is positively correlated with the first evaluation value and the second evaluation value, respectively; as an example, the specific formula included is:

[0159] The average of the first and second evaluation values ​​is used as the global evaluation value.

[0160] In Example 2, step S204 includes: obtaining the average change of adjacent elements in the evaluation value sequence, denoted as A1; and obtaining the average change of adjacent elements in the global evaluation value sequence. As an example, the specific formula is as follows:

[0161] For any two adjacent elements in the evaluation value sequence, let the element that comes first be denoted as x1 and the element that comes last be denoted as x2. Let |x2-x1| / (x2+x1) be the change between these two adjacent elements. The purpose of using |x2+x1| as the denominator is to remove dimensions and achieve normalization.

[0162] For any two adjacent elements in the global evaluation value sequence, let the element that comes first be denoted as y1 and the element that comes last be denoted as y2. Let |y2-y1| / (y2+y1) be the change between the two adjacent elements. The purpose of using y2+y1 as the denominator is to remove dimensions and achieve normalization.

[0163] The mean of all changes in the evaluation value sequence is denoted as the average change A1. The mean of all changes in the global evaluation value sequence is then calculated, and this mean is increased by 0.1, denoted as the average change A2. The purpose of increasing the mean by 0.1 is to avoid the denominator being 0 when the ratio of A1 to A2 is recorded as the rate of change of the evaluation value of each group.

[0164] In Example 2, step S204 includes: determining whether the average displacement of each particle deviates from the linear distribution characteristics of each group. As an example, the method includes:

[0165] Starting from the position of each particle, draw a perpendicular line segment to the linear distribution feature. The unit vector parallel to this perpendicular line segment, starting from the position of the particle, is denoted as the perpendicular vector. If the projection of the average displacement onto the perpendicular vector is less than 0, it indicates that the particle deviates from the linear distribution feature; if it is greater than or equal to 0, it indicates that it has not deviated from the linear distribution feature.

[0166] In Example 2, step S204 includes: th2 is positively correlated with the rate of change of the group evaluation value of each group, and th3 is positively correlated with the rate of change of the group evaluation value of each group. As an example, the specific formula is: th2=th3=(1+w1)×th1, where w1 represents the rate of change of the group evaluation value of each group.

[0167] In Example 2, step S205 includes using the path formed by all particles as the final oocyte retrieval path. The specific process involved is described in step S202. It should be noted that the final oocyte retrieval path obtained using step S202 may not be smooth. In this case, the surgeon can use a Bézier curve to fit it into a smooth path that facilitates needle insertion. In other embodiments, the surgeon can adjust the technique of the final oocyte retrieval path based on the specific display of the pelvic 3D model to ensure the smooth progress of the procedure.

[0168] Example 4:

[0169] In Example 2, step S201 includes the maturity of follicles. As an example, the method for obtaining the maturity of follicles is as follows:

[0170] All ultrasound images obtained during sectional scanning using a three-dimensional volumetric vaginal probe were segmented using a semantic segmentation algorithm (e.g., Unet neural network) to identify each follicle region in each ultrasound image. The maximum diameter (distance between the two farthest pixels in the follicle region) and the average diameter (average distance between pixels at the edge of the follicle region) of each follicle region were obtained. The maximum value of the maximum diameter and the maximum value of the average diameter of each follicle region in all ultrasound images were obtained. The surgeon read the follicle wall thickness of each follicle. The maximum value of the maximum diameter, the maximum value of the average diameter, and the follicle wall thickness were used as morphological features of each follicle. The patient's estradiol (E2), luteinizing hormone (LH), and progesterone (P) secretion levels were detected and recorded as physiological characteristics.

[0171] The morphological characteristics of each follicle and the patient's physiological characteristics are input into a fully connected neural network to obtain the maturity of each follicle.

[0172] The fully connected neural network in this embodiment has 3 intermediate layers, each with 5 neurons. The last layer of the fully connected neural network uses a softmax function to output the maturity of each follicle, so that the maturity of each follicle is greater than or equal to 0 and less than or equal to 1.

[0173] The training methods for this fully connected neural network include:

[0174] A large number of patients' physiological characteristics and the morphological characteristics of each follicle were collected as a dataset. Each patient's follicle morphological characteristics and each patient's physiological characteristics were treated as a sample, and each sample was manually labeled. In this embodiment, the label value for each sample ranged from {0, 0.2, 0.4, 0.6, 0.8, 1.0}. A larger label value indicated greater maturity; for example, a fully mature follicle was labeled with a value of 1.0, and a completely immature follicle was labeled with a value of 0.

[0175] Using this dataset, the fully connected neural network is trained using the cross-entropy loss function, and the neuron parameters of the fully connected neural network are updated using stochastic gradient descent.

[0176] The specific structure and training process of a fully connected neural network are well known, and will not be described in detail in this implementation.

[0177] Example 5:

[0178] Step S203 in Example 2 includes: updating the positions of all particles based on the evaluation value of each particle, the group evaluation value of each group, and the global evaluation value, including:

[0179] The new position of any particle is calculated using its position and assigned velocity (i.e., the position of the particle is updated). As an example, this process includes: pos_new = pos_old + v × t.

[0180] Where pos_new represents the new position of any particle, pos_old represents the position of any particle before the update; v represents the velocity of any particle, and t represents the preset time interval.

[0181] It should be noted that in this embodiment, the particle position before the update represents the three-dimensional coordinates of the pelvic cavity three-dimensional model. The unit of each dimension in the three-dimensional coordinates is millimeters, the unit of the particle velocity is millimeters per second (note that this velocity is a three-dimensional vector), and the unit of the preset time interval is seconds. In this embodiment, only numerical values ​​are considered when updating particle positions using the particle swarm optimization algorithm, and unit conversion is no longer considered. In addition, the evaluation value of each particle, the global evaluation value, and the group evaluation value of each group in this embodiment are dimensionless and can be directly used in calculations with the numerical values ​​of the aforementioned physical quantities such as position and velocity.

[0182] In this embodiment, the preset time interval t is equal to 2.

[0183] As an example, methods for obtaining the velocity of any single particle include:

[0184] v=v_old×par+par1×r1×R(tar,pos)+par2×r2×global_t+par3×r3×group_t.

[0185] Where v represents the velocity of any particle at the current updated position, and v_old represents the velocity of each particle at the previous updated position; it should be noted that when each particle is randomly generated, its corresponding velocity is also randomly generated, specifically a random number in (0, 2).

[0186] Where par represents the inertia weight, and this embodiment uses z=1.0 as an example; par1, par2, and par3 represent the first learning factor, the second learning factor, and the third learning factor, and this embodiment uses par1=0.1, par2=0.4, and par3=0.6 as examples; r1, r2, and r3 represent random numbers between (0, 1); tar represents the optimal position of each particle; pos represents the current position of the particle; R(tar, pos) represents the displacement vector from pos to tar; global_t represents the difference between the global optimal value and the global evaluation value; and group_t represents the difference between the mean of the group optimal values ​​of all groups and the mean of the group evaluation values ​​of all groups.

[0187] It should be noted that the result of par2×r2×global_t+par3×r3×group_t is a scalar, and the result of v_old×par+par1×r1×R(tar,pos) is a vector. The addition of a scalar and a vector means that each dimension of the vector is added to the scalar.

[0188] It should also be noted that the method for updating the positions of all particles in this embodiment is a conventional extension of the particle swarm optimization algorithm in this embodiment, and the specific principles of the above formula will not be elaborated in this embodiment.

[0189] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for planning a path for oocyte retrieval by puncture for assisted reproduction, characterized in that, The method comprises the following steps: Randomly generating a plurality of particles in the pelvic cavity three-dimensional model, updating the position of each particle by using a particle swarm algorithm, so that all the particles constitute a puncture oocyte collection path; the process of updating the position of each particle comprises: Before updating the position of each particle, dividing all the particles into a plurality of groups, so that all the particles in each group exhibit a linear distribution feature; obtaining a group evaluation value of each group according to the maturity of the follicle at the position of all the particles in each group; obtaining a global evaluation value of all the particles according to the rotation action of the puncture needle between the linear distribution features of different groups; The particle swarm algorithm updates the position of each particle by using the group evaluation value of each group and the global evaluation value, and records the average displacement generated by each particle; based on the change amount of the group evaluation value of each group with the global evaluation value, the particles in each group and the average displacement deviating from the linear distribution feature are screened out as deviating particles, and the particles outside each group and the average displacement not deviating from the linear distribution feature are screened out as approaching particles; each group is recombined according to the deviating particles and the approaching particles, each group is re-obtained, and the position of each particle is updated again by using the group evaluation value of each re-obtained group and the global evaluation value.

2. The puncture oocyte retrieval path planning method for assisted reproduction according to claim 1, wherein, The specific steps of dividing all the particles into a plurality of groups so that all the particles in each group exhibit a linear distribution feature comprise the following: Detecting a straight line in all the particles in the ovary region by using a Hough transform algorithm to obtain a plurality of straight lines; for any straight line, all the particles with a distance less than a first preset threshold from the straight line are taken as a group; The particles in the group are refitted into a straight line by using a least square method, and the distribution segment of the particles in the group on the refitted straight line is recorded as a linear distribution feature.

3. The puncture oocyte retrieval path planning method for assisted reproduction according to claim 2, wherein, The specific steps of obtaining a group evaluation value of each group according to the maturity of the follicle at the position of all the particles in each group comprise the following: For all the particles in any one group, the sum of the maturity of all the follicles at the position of all the particles is recorded as an overall maturity, and the mean value of the approaching features of all the particles in the group is calculated; the group evaluation value of each group is positively correlated with the overall maturity and the mean value of the approaching features, respectively; the approaching feature of the particle is equal to the distance between the particle and the linear distribution feature of each group.

4. The puncture oocyte retrieval path planning method for assisted reproduction according to claim 2, wherein, The specific steps of obtaining a global evaluation value of all the particles according to the rotation action of the puncture needle between the linear distribution features of different groups comprise the following: The linear distribution features of any two groups are recorded as line segments A and B; the connection path between line segments A and B is obtained by connecting the end points closest to each other in line segments A and B, and the connection path represents the rotation action of the puncture needle between the linear distribution features of any two groups; The mean value of the maturity of all the follicles passing through the connection path is recorded as a first maturity; the connection suitability of any two groups is obtained, and the connection suitability is negatively correlated with the length of the connection path and positively correlated with the first maturity; The straight line distribution characteristics of the two groups with the largest connection fitness and the connection path are combined to form a first sub-path; the first sub-path is extended by sequentially connecting the first sub-path with the straight line distribution characteristics of all groups, to obtain a final sub-path, so that all connection paths in the final sub-path have the largest connection fitness; The second sub-path is obtained by connecting the particles outside all groups, the second sub-path and the final sub-path are spliced to form a puncture oocyte path formed by all particles, and a global evaluation value is obtained according to the second sub-path and the final sub-path.

5. The puncture oocyte retrieval path planning method for assisted reproduction according to claim 2, wherein, The average displacement of each particle is recorded, the variation of the group evaluation value of each group with the global evaluation value is obtained, the deviating particles are screened from the particles in each group and deviating from the straight line distribution characteristics, and the approaching particles are screened from the particles outside each group and not deviating from the straight line distribution characteristics, and the specific steps include the following: The position of each particle is updated several times by using the group evaluation value of each group and the global evaluation value, and a displacement vector of each particle is recorded after the position of each particle is updated once; after the position of each particle is updated several times, the average of all displacement vectors of each particle is recorded as the average displacement; All particles in each group and deviating from the straight line distribution characteristics are recorded as set S1; in set S1, the particles with a distance greater than a second preset threshold from the straight line distribution characteristics of each group are obtained and recorded as deviating particles; The second preset threshold is positively correlated with the variation of the group evaluation value of each group with the global evaluation value; All particles outside each group and not deviating from the straight line distribution characteristics of each group are recorded as set S2; in set S2, the particles with a distance less than a third preset threshold from the straight line distribution characteristics of each group are obtained and recorded as approaching particles; The position of each particle is updated several times by using the group evaluation value of each group and the global evaluation value, and the group evaluation value of each group and the global evaluation value are calculated after the position of each particle is updated once; after the position of each particle is updated several times, all group evaluation values obtained by each group are recorded as an evaluation value sequence, and all global evaluation values obtained are recorded as a global evaluation value sequence; the ratio of the average variation of adjacent elements in the evaluation value sequence to the average variation of adjacent elements in the global evaluation value sequence is recorded as the group evaluation value variation rate of each group; the second preset threshold and the third preset threshold are positively correlated with the group evaluation value variation rate of each group. 6.The method for planning a puncture oocyte retrieval path for assisted reproduction according to claim 1, wherein, The specific steps of the recombination of each group according to the deviating particles and the approaching particles to obtain each group again include the following: Any one deviating particle of each group and any one particle inside the group and outside all deviating particles of the group are marked as mutually exclusive particles; Any one approaching particle of each group and any one particle inside the group and outside all deviating particles of the group are marked as complementary particles; After updating the position of each particle by using the group evaluation value of each group and the global evaluation value, all the particles are regrouped to obtain each group again; in each group obtained again, the mutually exclusive particles are removed; it is judged whether the complementary particle of any particle in each group belongs to the group, and if not, the complementary particle of the particle is added to the group.

7. The puncture oocyte retrieval path planning method for assisted reproduction according to claim 4, wherein, The first sub-path is extended to obtain the final sub-path by sequentially connecting the first sub-path with the linear distribution features of all groups, so that all connection paths in the final sub-path have the maximum connection fitness, and the specific steps include the following: D1: the groups corresponding to the two linear distribution features constituting the first sub-path are denoted as G1 and G2; in all groups other than G1 and G2, the group with the maximum connection fitness with G2 is denoted as G3, and the group with the maximum connection fitness with G1 is denoted as G4; When the connection fitness of G3 and G2 is greater than or equal to the connection fitness of G4 and G1, G3 is denoted as the to-be-connected group, otherwise G4 is denoted as the to-be-connected group; the to-be-connected group is denoted as g, and the linear distribution features in G1, G2 and g and the connection paths between the linear distribution features in G1, G2 and g constitute the path after the first sub-path is extended; D2: the path after the first sub-path is extended is taken as the first sub-path again, and then the first sub-path is extended several times by repeatedly executing D1 to obtain the final sub-path. 8.The method for planning a puncture oocyte retrieval path for assisted reproduction according to claim 1, wherein, The specific steps for updating the position of each particle by using the group evaluation value of each group and the global evaluation value include the following: The position of each particle is updated by using the position of any particle and the speed of the particle; The difference between the group optimal value of each group and the group evaluation value of each group is denoted as group_t, and the difference between the global optimal value and the global evaluation value is denoted as global_t; the speed of the particle is positively correlated with group_t and global_t; The group optimal value of each group is equal to the maximum value of the group evaluation value of each group in the historical process of updating the position of all particles by the particle swarm algorithm; The global optimal value is equal to the maximum value of the global evaluation value in the historical process of updating the position of all particles by the particle swarm algorithm. 9.The method of claim 4, wherein, The specific steps for obtaining the global evaluation value according to the second sub-path and the final sub-path include the following: The sum of the connection fitnesses of all connection paths in the final sub-path is denoted as the first evaluation value; the ratio of the length of the second sub-path to the length of the puncture oocyte path is denoted as x, and exp(-x) is denoted as the second evaluation value, wherein exp() represents the exponential function with the natural constant as the base; The global evaluation value is positively correlated with the first evaluation value and the second evaluation value, respectively.

10. A puncture oocyte retrieval path planning system for assisted reproduction, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor executes the computer program to perform the puncture oocyte path planning method for assisted reproduction according to any one of claims 1-9. The processor executes the computer program to perform the puncture oocyte path planning method for assisted reproduction according to any one of claims 1-9.

Images