Pedicle screw planning methods, equipment, and computer program products

Abstract

This application relates to the field of computer-aided medical technology, providing a method, device, and computer program product for pedicle screw planning. The method includes: constructing multiple fitness functions based on the spatial relationship between the vertebral body and the planned screw for evaluating the placement planning of the corresponding screws; the multiple fitness functions include a reward function and a penalty function; evaluating the fitness score of each individual in the population using the multiple fitness functions, where any individual in the population is defined as including a first screw and a second screw planned for placement in the vertebral body; iteratively performing multiple genetic controls on the population based on the fitness scores to obtain a target population; and determining multiple target individuals from the target population, where the multiple target individuals represent multiple executable pedicle screw planning schemes. Using the above method, pedicle screw placement schemes can be planned efficiently and accurately.

Description

[0001] This application is a divisional application. The original application has the application number 202610052552.X and the original application date is January 15, 2026. The entire contents of the original application are incorporated herein by reference. Technical Field

[0002] This application belongs to the field of computer-aided medical technology, and in particular relates to a pedicle screw planning method, device and computer program product. Background Technology

[0003] Pedicle screw fixation is one of the core techniques in spinal surgery, used to treat fractures, deformities, degenerative diseases, and more. With the development of medical imaging and robot-assisted navigation technologies, precise and safe screw placement is gaining increasing acceptance from doctors and patients.

[0004] Accurate planning of the pedicle screw placement is a prerequisite for successful pedicle screw fixation. Current technologies for pedicle screw planning combined with robot-assisted navigation mainly include two methods: traditional algorithms and deep learning algorithms. Traditional algorithms are flexible and simple, but rely on manually designed vertebral geometry features for identification. Human-defined rules cannot cover all complex vertebral conditions, resulting in insufficient generalization and a tendency to get stuck in local optima. Deep learning algorithms offer high accuracy, but depend on large amounts of training data, which is difficult and expensive to obtain due to ethical concerns regarding clinical data. Furthermore, both traditional and deep learning algorithms typically only provide a single optimized plan, lacking the flexibility to offer multiple options for surgeons. Summary of the Invention

[0005] In view of this, embodiments of this application provide a pedicle screw planning method, device, and computer program product, which can provide multiple planning schemes for surgeons to choose from while ensuring the efficiency and accuracy of pedicle screw placement planning. It can not only output efficient and accurate placement schemes, but also improve the flexibility of the planning process.

[0006] The first aspect of this application provides a pedicle screw planning method, including: Based on the spatial relationship between the vertebral body and the screw, multiple fitness functions are constructed to evaluate the placement plan of the corresponding screw. The multiple fitness functions include a reward function and a penalty function. The fitness score of each individual in the population is obtained by evaluating multiple fitness functions, wherein any individual in the population is defined as including the first screw and the second screw of the vertebral body to be inserted. Based on the fitness score, the population is subjected to multiple genetic control iterations to obtain the target population; Multiple target individuals are identified from the target population, and these multiple target individuals represent multiple executable pedicle screw planning schemes.

[0007] Optionally, the first screw and the second screw each have multiple screw placement information, including screw entry point position, screw direction, screw diameter, and screw length; The reward function includes any one or more of the following functions: The region reward function is used to represent the ratio of the volume of the screw in the overlapping region of the cone to the total volume of the screw. A symmetric reward function is used to represent the angular difference between the first and second screws and the vertebral midline. The bone mineral density reward function is used to represent the ratio of the cumulative bone mineral density of the screw within the vertebral body to the overall bone mineral density of the vertebral body. The stability reward function is used to represent the ratio of screw diameter to maximum diameter, and the ratio of screw length to maximum length; The penalty function includes any one or more of the following functions: The collision penalty function is used to represent the collision state between the first screw and the second screw; The penetration penalty function is used to represent the positional relationship between the screw stop point and the vertebral body.

[0008] Optionally, if the first screw collides with the second screw, the reward / penalty value corresponding to the collision penalty function is -1; if the first screw and the second screw do not collide, the reward / penalty value corresponding to the collision penalty function is 1. If the stop point of the first screw or the second screw penetrates outside the vertebral body, the reward / penalty value corresponding to the penetration penalty function is -1; if the stop points of both the first screw and the second screw do not penetrate outside the vertebral body, the reward / penalty value corresponding to the penetration penalty function is 1.

[0009] Optionally, the step of iteratively performing genetic control on the population based on the fitness score to obtain the target population includes: Based on the fitness score, multiple strategies are used to genetically control the population to generate the next generation population. The genetic control is performed iteratively on the population multiple times to obtain the target population.

[0010] Optionally, the various strategies include elite strategies and roulette strategies; the genetic control of the population based on the fitness score, using multiple strategies simultaneously to generate the next generation population, includes: An elite strategy is employed to directly replicate a predetermined number of individuals with the highest fitness scores from the current population into the next generation; and, An individual from any region is randomly selected as the parent using a roulette wheel strategy. By performing crossover and mutation on the parent individuals, the corresponding offspring individuals are generated and placed into the next generation population.

[0011] Optionally, the step of randomly selecting an individual from any region as the parent using a roulette wheel strategy, and generating corresponding offspring individuals by performing crossover and mutation on the parent individuals and placing them into the next generation population includes: Determine the first and second parent generations from the population; Using the first parent as the base sequence and the second parent as the cross sequence, the base sequence is traversed sequentially; wherein, each sequence value in the base sequence and the cross sequence is a value corresponding to multiple pinning information; When any sequence value in the base sequence satisfies the crossover probability, the corresponding sequence value in the crossover sequence is replaced in the base sequence to obtain the offspring sequence; Iterate through the child sequence in sequence; When any sequence value in the offspring sequence satisfies the mutation probability, the sequence value of the corresponding part in the offspring sequence is modified to obtain an offspring individual, which is then placed into the next generation population.

[0012] Optionally, the fitness score of each individual in the population is obtained by evaluating multiple fitness functions, including: Determine the weights for each of the fitness functions; Based on the weights, the reward and penalty values ​​corresponding to multiple fitness functions are weighted and summed to obtain the fitness score of each individual.

[0013] Optionally, determining multiple target individuals from the target population includes: Sort multiple individuals in the target population according to the fitness score; A predetermined proportion of individuals with the highest ranking are retained as target individuals representing executable pedicle screw planning schemes.

[0014] A second aspect of this application provides a pedicle screw planning device, comprising: An initialization module is used for population initialization, wherein the population includes multiple individuals, and any individual is defined as including a first screw and a second screw, wherein the first screw and the second screw each have multiple screw placement information; The fitness evaluation module is used to evaluate the fitness of each individual in the population according to multiple fitness functions to obtain the fitness score of each individual. The fitness functions are constructed based on multiple pinning information. A genetic control module is used to genetically control the population based on the fitness score to generate the next generation population; The target population generation module is used to iteratively perform the genetic control multiple times to obtain a target population, which includes multiple target individuals, and any one of the target individuals represents a pedicle screw planning scheme.

[0015] A third aspect of this application provides a computer device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the computer device performs the method as described in any of the first aspects above.

[0016] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a computer, implements the method described in any of the first aspects above.

[0017] A fifth aspect of this application provides a computer program product, including a computer program that, when the computer program is run, causes the method described in any of the first aspects above to be executed.

[0018] Compared with the prior art, the embodiments of this application have the following beneficial effects: In this embodiment, by defining multiple individuals including a first screw and a second screw, a population comprising multiple individuals can be initialized. The various screw placement information of the first and second screws can be used to construct multiple fitness functions, thereby enabling fitness evaluation of each individual in the population based on these functions, yielding a fitness score for each individual. Based on this, genetic control of the population according to the fitness scores can generate a next-generation population. Through iterative genetic control, a target population comprising multiple target individuals can be obtained. Each target individual in the target population represents a pedicle screw planning scheme. Applying the planning method provided in this embodiment can provide multiple planning schemes for surgeons to choose from while ensuring the efficiency and accuracy of pedicle screw placement planning. This not only outputs efficient and accurate placement schemes but also improves the flexibility of the planning process. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application, 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 this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1This is a schematic diagram of a pedicle screw planning method provided in an embodiment of this application; Figure 2 This is a schematic diagram of a next-generation population generation method provided in an embodiment of this application; Figure 3 This is a schematic diagram of a pedicle screw planning process provided in an embodiment of this application; Figure 4 This is a schematic diagram of a pedicle screw planning device provided in an embodiment of this application; Figure 5 This is a schematic diagram of a computer device provided in an embodiment of this application. Detailed Implementation

[0021] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0022] As mentioned earlier, pedicle screw planning combined with robot-assisted navigation mainly includes two planning methods: those using traditional algorithms and those using deep learning algorithms. Both methods use spinal imaging data as input and output the planned screw positions, i.e., the screw's entry and exit points. Existing methods can only output a single planning scheme, failing to provide surgeons with more options and exhibiting poor flexibility.

[0023] To address the aforementioned issues, this application provides a genetic algorithm-based pedicle screw planning method. Based on the uncertainty of vertebral body geometry, it transforms clinicians' experience into optimization objectives, combining various reward and penalty mechanisms, such as screw overlap area rewards, symmetry rewards, bone density rewards, stability rewards, collision penalties, and penetration penalties. Utilizing the global search and randomness of the genetic algorithm, it outputs efficient and accurate screw planning schemes. Furthermore, applying the method provided in this application, multiple executable planning schemes can be generated, allowing doctors to select the most suitable option based on the patient's specific situation.

[0024] The technical solution of this application will be described below through specific embodiments.

[0025] Reference Figure 1 The diagram illustrates a pedicle screw planning method provided in an embodiment of this application, which may specifically include the following steps: S101, Population initialization, wherein the population includes multiple individuals, and any one of the individuals is defined as including a first screw and a second screw, wherein the first screw and the second screw each have multiple screw placement information.

[0026] It should be noted that this method can be applied to computer equipment; that is, the executing subject of this application embodiment is a computer device. This computer device can execute the various steps of this method to plan the pedicle screw placement scheme. Furthermore, the computer device executing this scheme can output multiple planning schemes for the surgeon to choose from. The aforementioned computer device can be a device with relevant data processing functions, such as a desktop computer or a cloud server. This application embodiment does not limit the specific type of computer device.

[0027] In pedicle screw placement planning, each vertebral body to be screwed requires two screws: a first screw (or screw 1) and a second screw (or screw 2). Therefore, when using a genetic algorithm for screw placement planning, each individual in the population can be defined as including both the first and second screws. For example, an individual... s It can be defined as:

[0028] in, This indicates the first screw (screw 1). This indicates the second screw (screw 2).

[0029] In this embodiment, the first screw and the second screw may each have various placement information. This placement information can be related to how the first screw and the second screw are placed or inserted during the placement process. For example, the placement information may include the screw entry point position, screw direction, screw diameter, and screw length. The screw entry point position may refer to the starting point position for screw insertion, the screw direction may refer to the direction in which the screw is inserted into the vertebral body from the aforementioned entry point position, and the screw diameter and screw length indicate the type of screw.

[0030] In one possible implementation of the embodiments of this application, in the case of s When the definition includes a first screw and a second screw, the aforementioned first screw, second screw, and their corresponding screw placement information can be represented as follows:

[0031]

[0032] in, Indicates the position of the screw insertion point. Indicates the direction of the screw. Indicates the diameter of the screw. This indicates the length of the screw. i =1,2, when i When =1, it indicates that the first screw is... ,when i When =2, it indicates the second screw, i.e. .

[0033] The diameter and length of the aforementioned screws are determined by the actual screws used. For example, the screw diameter and length values ​​can be determined based on the dimensions of various screws in a screw database. For instance, based on the actual dimensions of various screws in the screw database, the screw diameter and length values ​​could be:

[0034]

[0035] In this embodiment of the application, the population includes multiple individuals as defined above. s Each individual s includes two screws and their corresponding screw placement information. One of the purposes of the computer device executing the embodiments of this application is to use a genetic algorithm to iteratively obtain a target population including multiple target individuals from an initialized population, where each target individual represents an executable planning scheme.

[0036] In this embodiment of the application, when multiple screw placement information includes screw entry point position, screw direction, screw diameter and screw length, each individual in the initialized population includes a first screw, a second screw and its corresponding screw entry point position, screw direction, screw diameter and screw length.

[0037] During population initialization, the screw insertion point of any screw can be randomly determined within the cone-shaped space where the screw is to be placed. The screw direction can also be randomly determined, while the screw diameter and length can be randomly selected from a screw database. This completes the population initialization.

[0038] S102. The fitness of each individual in the population is evaluated according to multiple fitness functions to obtain the fitness score of each individual. The fitness functions are constructed based on multiple pinning information.

[0039] In this embodiment, the fitness function can be constructed based on multiple screw placement information and can be used to evaluate the screw placement plan of the corresponding screw. Specifically, the fitness function is determined based on the spatial relationship between the vertebral body and the screw plan, combined with actual clinical needs. That is, multiple fitness functions can be constructed based on the spatial relationship between the vertebral body and the screw plan to evaluate the screw placement plan of the corresponding screw.

[0040] In one possible implementation of this application embodiment, the constructed multiple fitness functions may include a reward function and a penalty function.

[0041] The reward function can include any one or more of the following functions: 1) Regional reward function; 2) Symmetric reward function; 3) Bone mineral density reward function; 4) Stability reward function.

[0042] The penalty function can include any one or more of the following functions: 5) Collision penalty function; 6) Penalty function for penetration.

[0043] The region reward function represents the ratio of the screw's volume in the overlapping region of the cone to the screw's volume. If this ratio is close to 1, it indicates that the screw is completely located within the cone.

[0044] The symmetry reward function is used to represent the angular difference between the two screws and the vertebral body's central axis. This function can prevent the relative angle between the two screws from being too large.

[0045] The bone mineral density (BMD) reward function represents the ratio of the cumulative BMD of the screw within the vertebral body to the total BMD of the vertebral body. This BMD value can be determined based on computed tomography (CT) scan values. A higher ratio indicates a more stable screw within the vertebral body, better meeting the requirements for clinical application.

[0046] The stability reward function represents the ratio of the screw diameter to the maximum diameter, and the ratio of the screw length to the maximum length. The maximum diameter and maximum length can refer to the maximum diameter and maximum length of screws already existing in the screw database. Generally, the larger the diameter ratio and length ratio, the higher the stability of the screw.

[0047] The collision penalty function is used to represent the collision state between the first screw and the second screw, that is, whether the two screws collide.

[0048] The penetration penalty function is used to represent the positional relationship between the screw stop and the vertebral body, i.e., whether the screw stop is located inside or outside the vertebral body.

[0049] The collision penalty function and penetration penalty function can be determined with corresponding coefficients based on the actual collision and penetration states. For example, if the first screw collides with the second screw, the collision penalty function can be represented by a collision coefficient of -1; if the first screw and the second screw do not collide, the collision coefficient can be represented by 1. Correspondingly, if the stop point of a screw is located outside the vertebral body, the penetration penalty function can be represented by a penetration coefficient of -1; if the stop points of both the first screw and the second screw are located inside the vertebral body, the penetration coefficient can be represented by 1.

[0050] The penalty coefficients for collision and penetration mentioned above can also refer to the reward / penalty values ​​corresponding to the collision penalty function and the penetration penalty function. Therefore, for the example above, if the first screw collides with the second screw, a penalty can be imposed on the corresponding fitness, and the reward / penalty value corresponding to the collision penalty function can be -1; if the first screw and the second screw do not collide, the reward / penalty value corresponding to the collision penalty function can be 1. Similarly, if the stop point of either the first or second screw penetrates outside the vertebral body, a penalty can be imposed on the corresponding fitness, and the reward / penalty value corresponding to the penetration penalty function can be -1; if the stop points of both the first and second screws do not penetrate outside the vertebral body, the reward / penalty value corresponding to the penetration penalty function can be 1.

[0051] In one possible implementation of this application, each fitness function may have a corresponding weight. A computer device can determine the weight of each fitness function and, based on that weight, perform a weighted sum of the reward and penalty values ​​corresponding to multiple fitness functions to obtain the fitness score for each individual. This fitness score can be used to represent the rationality or accuracy of the planning scheme corresponding to the individual.

[0052] In one example, the fitness score can be represented as:

[0053] Here, Aera represents the ratio of the screw volume in the overlapping area of ​​the vertebral body to the total screw volume; a ratio close to 1 indicates that the screw is completely within the vertebral body. Angle represents the angle difference between the two screws and the central axis, preventing excessively large relative angles between the two screws. HU represents the ratio of the cumulative bone mineral density (CT value) of the screws within the vertebral body to the total bone mineral density (CT value) of the vertebral body; a larger ratio indicates greater stability of the screw within the vertebral body, better meeting clinical application requirements. D represents the ratio of the screw diameter to the maximum diameter in the screw database, and L represents the ratio of the screw length to the maximum length in the screw database; larger diameter and length ratios indicate higher screw stability. C indicates whether the screws collide; for example, -1 for collision and 1 for no collision. P indicates whether the screw's insertion point is outside the vertebral body; for example, -1 for penetrating the vertebral body and 1 for penetrating the vertebral body. The penalty coefficients for collision and penetration can be adjusted according to actual conditions. (i=1,2,3,4,5,6,7) represent different weights, which can be adjusted according to different clinical needs.

[0054] S103. Genetic control is performed on the population based on the fitness score to generate the next generation population.

[0055] In the embodiments of this application, genetic control can refer to the control process of generating the next generation of population based on fitness scores during population iteration.

[0056] In one possible implementation of this application, multiple strategies can be used simultaneously to genetically control the population based on fitness scores to generate the next generation population. For example, two different population generation strategies can be used simultaneously to generate the next generation population. Exemplarily, an elite strategy and a roulette wheel strategy can be used simultaneously to generate the next generation population.

[0057] The elite strategy can refer to retaining a subset of high-performing individuals and directly replicating the highest-fitting individuals (referred to as "elites") from the current population into the next generation without crossover or mutation. To ensure algorithm convergence, the top 5%-10% of individuals with the highest fitness scores (the larger the population, the higher the proportion) can be selected for direct entry into the next generation. In other words, the elite strategy can be used to directly replicate a predetermined number of individuals with the highest fitness scores from the current population into the next generation.

[0058] The roulette wheel strategy involves randomly pointing to a region on a disk and selecting an individual from that region as its parent. The disk consists of all individuals, and regions are divided by the ratio of an individual's fitness score to the total score; individuals with higher fitness scores are more likely to be selected as parents.

[0059] For multiple individuals selected as parents using the roulette wheel strategy, corresponding offspring individuals can be generated through crossover and mutation. That is, an individual from any region can be randomly selected as a parent using the roulette wheel strategy, and the corresponding offspring individuals can be generated and placed into the next generation of the population by performing crossover and mutation on the parent individuals.

[0060] like Figure 2 The diagram shown is a schematic representation of a next-generation population generation method provided in an embodiment of this application. According to... Figure 2 The generation methods shown can be used to generate the initial population in two ways: First, an elitist strategy can be employed, directly replicating the small subset of individuals with the highest fitness scores to form offspring set 1. Second, a roulette wheel strategy can be used to generate parent set 1 and parent set 2, and then crossover mutation can be used to generate offspring set 2. Offspring set 1 and offspring set 2 together constitute the next generation population. Figure 2The next generation of the population shown can proceed in the same manner during subsequent iterations.

[0061] Combination Figure 2 The method for generating the next generation population, as shown, involves a computer device genetically controlling the population based on fitness scores to generate the next generation. First, multiple individuals in the existing population are ranked according to their fitness scores, and a predetermined number of individuals with the highest rankings are copied into the next generation population. This process is the process of an elite strategy, where the predetermined number of individuals with the highest rankings can be... Figure 2 The set of offspring individuals shown in Figure 1 includes multiple individuals.

[0062] In the roulette wheel strategy and crossover mutation process, the computer device can determine the first parent (or parent 1) and the second parent (or parent 2) from the population. Then, based on the first and second parents, crossover mutation is performed to generate multiple offspring individuals, which are then replicated into the next generation of the population. The multiple offspring individuals obtained through the roulette wheel strategy and crossover mutation process can be... Figure 2 The set of offspring individuals shown in Figure 2 includes multiple individuals.

[0063] It should be noted that multiple individuals selected through the elite strategy can serve as parent individuals in roulette wheel selection and crossover / mutation processes. For example, Figure 2 The individuals included in the neutron generation set 1 can be with Figure 2 Individuals in parent generation 1 or parent generation 2 are identical.

[0064] In this embodiment of the application, when generating multiple offspring individuals through crossover mutation based on the first and second parent generations, the first parent generation can be used as the base sequence and the second parent generation as the crossover sequence. Each sequence value in the aforementioned base sequence and crossover sequence corresponds to a value for various pin placement information. That is, the base sequence and crossover sequence can include sequence values ​​such as {position, orientation, diameter, and length of screw 1, and position, orientation, diameter, and length of screw 2}.

[0065] Computer devices can sequentially traverse the base sequence. When any sequence value in the base sequence satisfies the crossover probability, the corresponding sequence value in the crossover sequence can be replaced in the base sequence to obtain the offspring sequence.

[0066] For example, during the sequential traversal of the base sequence, if the sequence value of the length of screw 1 in the base sequence satisfies the crossover probability, the sequence value of the position, direction, diameter, length, etc. of screw 2 in the crossover sequence can be replaced in the base sequence to obtain the offspring sequence.

[0067] Based on this, multiple offspring individuals can be generated by performing sequence mutations on the offspring sequence.

[0068] Sequence mutation of the offspring sequence can be performed by sequentially traversing the sequence. If any sequence value in the offspring sequence satisfies the mutation probability, then that sequence value needs to be mutated, i.e., the corresponding sequence value in the offspring sequence is modified to obtain the offspring individual. For example, if the sequence value of the length of screw 1 in the offspring sequence satisfies the mutation probability, and the length of screw 1 is 30, then the length of screw 1 can be modified. Specifically, the length of screw 1 can be mutated by randomly selecting dimensions from the screw database.

[0069] S104. The genetic control is iterated multiple times to obtain a target population, which includes multiple target individuals, and any one of the target individuals represents a pedicle screw planning scheme.

[0070] The aforementioned S103 describes the specific process of genetically controlling the population based on fitness scores to generate the next generation. The execution of the genetic algorithm requires the computer to repeatedly perform genetic control multiple times until the number of iterations reaches a set threshold. That is, the genetic control described in the previous example can be iteratively performed on the population multiple times to obtain the target population.

[0071] Specifically, for the generated next-generation population, the computer device can repeatedly perform genetic control on the next-generation population based on fitness scores. When the number of iterations of genetic control on the next-generation population reaches a set threshold, the currently obtained population can be used as the target population, thereby identifying multiple target individuals from the target population.

[0072] In one possible implementation of this application, considering that the vertebral cone to be fitted may have structural complexity due to fractures or lesions, the planning scheme with the highest fitness score may not be applicable. Therefore, it is advisable to retain several individuals with high scores within the population. Thus, when determining several target individuals from the target population, these individuals can be ranked according to their fitness scores, and a predetermined proportion of individuals with high rankings can be retained as target individuals representing the pedicle screw planning scheme. This predetermined proportion can be 5%-10%.

[0073] For example, if the target population contains 100 individuals, 5-10 individuals can be retained (5%-10%). These 5-10 individuals represent 5-10 feasible pedicle screw placement plans. The surgeon can choose the most suitable plan from these multiple plans based on the patient's specific situation. For instance, from the final target population, the suitability of the plans for each individual can be determined by ranking them from highest to lowest fitness score, thus identifying the plan that best matches the patient's condition and has the highest fitness score.

[0074] The planning method provided in this application generates multiple executable planning schemes. Compared with existing planning algorithms that can only retain one optimal solution, this avoids the situation where a scheme is abandoned due to the poor planning effect of a certain screw.

[0075] In this embodiment, by defining multiple individuals including a first screw and a second screw, a population comprising multiple individuals can be initialized. The various screw placement information of the first and second screws can be used to construct multiple fitness functions, thereby enabling fitness evaluation of each individual in the population based on these functions, resulting in a fitness score for each individual. Based on this, genetic control of the population according to the fitness scores can generate a next-generation population. Through iterative genetic control, a target population comprising multiple target individuals can be obtained. Each target individual in the target population represents a pedicle screw planning scheme. Applying the planning method provided in this embodiment can provide multiple planning schemes for surgeons to choose from while ensuring the efficiency and accuracy of pedicle screw placement planning. This not only outputs efficient and accurate placement schemes but also improves the flexibility of the planning process.

[0076] It should be noted that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0077] To facilitate understanding, the pedicle screw planning method provided in this application embodiment will be introduced below with a complete example.

[0078] Reference Figure 3 This diagram illustrates an automatic pedicle screw planning process provided in an embodiment of this application. The overall process of this method can be summarized into six parts: data preprocessing, screw data encoding, population initialization, fitness evaluation, genetic control, and generation of multiple screw planning schemes. Below, in conjunction with... Figure 3 As shown, a detailed introduction will be provided.

[0079] 1. In the data preprocessing stage: Robot-assisted navigation surgery can be performed based on the patient's spinal imaging data. The overall spinal imaging data can be segmented into voxel data of individual vertebrae using manual or automatic segmentation, thus obtaining the vertebral midline. For example, as... Figure 3 As shown, taking a patient's CT image data as an example, through vertebral segmentation, the overall spinal CT image data can be divided into voxel data of individual vertebrae. Then, calculations are performed on the voxel data of individual vertebrae to determine the midline of that vertebra. After completing the above data preprocessing, automatic screw planning can be performed.

[0080] 2. During the screw data encoding stage: Since each cone requires the planning of two screws, the encoding of an individual in the population is related to the position, orientation, diameter, and length of the two screws. The individual s is defined as:

[0081]

[0082]

[0083] Among them, the first screw (screw 1) and the second screw (screw 2) are respectively used and express. Indicates the position of the screw insertion point. Indicates the direction of the screw. Indicates the diameter of the screw. This indicates the length of the screw. i =1,2, when i When =1, it indicates that the first screw is... ,when i When =2, it indicates the second screw, i.e. .

[0084] The range of screw diameters and lengths depends on the dimensions in the actual screw database used, and is set by the user. For example, the screw database may include:

[0085]

[0086] 3. During the population initialization phase: A population of N individuals can be randomly generated, where N is the population size. For each individual, the position of the screw can be obtained by randomly selecting a point in the cone space, the screw direction can be a random direction in space, and the diameter and length of the screw can be randomly selected from a screw database.

[0087] 4. During the fitness assessment phase: The fitness function is the core of this approach. It is determined based on the spatial relationship between the cone and screw planning, combined with actual clinical needs. For example, the evaluation of screw planning can consist of the following components: 1) The proportion of overlapping screw areas (regional reward); 2) The angle between the screw and the central axis (symmetry bonus); 3) The proportion of bone mineral density (CT value) in the screw overlap area (bone mineral density bonus); 4) Screw diameter to length ratio (stability bonus); 5) Collision status of screws (collision penalty); 6) Screw penetration status (penetration penalty).

[0088] The aforementioned evaluation factors correspond to the six fitness functions in the aforementioned embodiments, namely, the region reward function, the symmetry reward function, the bone density reward function, the stability reward function, the collision penalty function, and the penetration penalty function.

[0089] Therefore, the expression for the fitness score can be obtained as follows:

[0090] Here, Aera represents the ratio of the screw volume in the overlapping area of ​​the vertebral body to the total screw volume; a ratio close to 1 indicates that the screw is completely within the vertebral body. Angle represents the angle difference between the two screws and the central axis, preventing excessively large relative angles between the two screws. HU represents the ratio of the cumulative bone mineral density (CT value) of the screws within the vertebral body to the total bone mineral density (CT value) of the vertebral body; a larger ratio indicates greater stability of the screw within the vertebral body, better meeting clinical application requirements. D represents the ratio of the screw diameter to the maximum diameter in the screw database, and L represents the ratio of the screw length to the maximum length in the screw database; larger diameter and length ratios indicate higher screw stability. C indicates whether the screws collide; for example, -1 for collision and 1 for no collision. P indicates whether the screw's insertion point is outside the vertebral body; for example, -1 for penetrating the vertebral body and 1 for penetrating the vertebral body. The penalty coefficients for collision and penetration can be adjusted according to actual conditions. (i=1,2,3,4,5,6,7) represent different weights, which can be adjusted according to different clinical needs.

[0091] 5. In the stage of genetic control: Genetic control refers to the control during population iteration, where the scores of all individuals are calculated based on a fitness function to generate the next generation of the population. The generation methods in the embodiments of this application may include: elite strategy, roulette wheel strategy, crossover, and mutation.

[0092] The elite strategy involves retaining a subset of high-performing individuals and directly replicating the highest-fitting individuals (referred to as "elites") from the current population into the next generation without crossover or mutation. To ensure algorithm convergence, the top 5%-10% of individuals (the larger the population, the higher the percentage) can be selected for direct entry into the next generation.

[0093] The roulette wheel strategy involves randomly pointing to a region on a roulette wheel and selecting an individual within that region as its parent. The roulette wheel consists of all individuals, and regions are divided by the ratio of an individual's score to the total score; therefore, individuals with higher scores are more likely to be selected as parents. Parents selected via the roulette wheel strategy can include a first parent (parent 1) and a second parent (parent 2).

[0094] Crossover and mutation are methods for generating offspring. When performing crossover and mutation, the screw sequences of parent generation 1 and parent generation 2 can be crossovered. Assuming parent generation 1 has a lower fitness score, the sequence of parent generation 1 is used as the base sequence, and parent generation 2 is used as the crossover sequence. Both the base sequence and the crossover sequence contain sequence values ​​such as {position, orientation, diameter, and length of screw 1 and position, orientation, diameter, and length of screw 2}. It should be noted that screw 1 in the base sequence and screw 1 in the crossover sequence can be different screws; similarly, screw 2 in the base sequence and screw 2 in the crossover sequence can also be different screws.

[0095] For the base sequence, it can be traversed sequentially from left to right. When the crossover probability is met, the corresponding part of the parent 2 sequence can be replaced in the parent 1 sequence. After all traversals are completed, the offspring sequence is generated. At this point, the offspring sequence needs to undergo sequence mutation. Specifically, the offspring sequence is traversed sequentially, and when the mutation probability is met, a certain part of the sequence is mutated, modifying the value of that part. For example, if the screw length is 30, the size can be randomly selected from the screw database after mutation. After crossover and mutation, offspring are obtained, and genetic control is repeated multiple times until the size of the next generation population equals the set population size.

[0096] In this embodiment, the number of individuals in the population can be set between 50 and 150. Too few individuals result in insufficient diversity, a narrow search space, and a tendency to prematurely aggregate to a local optimum, making it impossible to find the truly optimal path; too many individuals lead to increased computational costs and excessively long algorithm execution time.

[0097] The process involves iterating repeatedly to generate a new population until the set threshold number of iterations is reached, at which point the target population is obtained.

[0098] 6. During the stage of generating multiple screw planning schemes: Because genetic algorithms are characterized by randomness and strong robustness, they can effectively optimize even for cones with significant variability. Therefore, after a single computation, the highest fitness score is selected as the first choice, followed by other high-fitness schemes within the population. This is because cones may become structurally complex due to fractures or lesions, and the highest-scoring scheme may not always be suitable. For example, if the population size is 100, the top 5%-10% of individuals can be retained, resulting in 5-10 schemes for doctors to quickly review and select. Compared to most traditional automatic screw planning algorithms that only retain one optimal result and are easily discarded due to poor performance of a particular screw plan, this method can generate multiple screw planning schemes after a single computation, allowing doctors to choose and adjust, thus offering broader applicability.

[0099] Reference Figure 4 The diagram illustrates a pedicle screw planning device according to an embodiment of this application, which may specifically include an initialization module 401, a fitness evaluation module 402, a genetic control module 403, and a target population generation module 404, wherein: An initialization module 401 is used for population initialization, wherein the population includes multiple individuals, and any individual is defined as including a first screw and a second screw, wherein the first screw and the second screw each have multiple screw placement information; Fitness evaluation module 402 is used to evaluate the fitness of each individual in the population according to multiple fitness functions to obtain the fitness score of each individual. The fitness functions are constructed based on multiple pinning information. Genetic control module 403 is used to perform genetic control on the population based on the fitness score to generate the next generation population; The target population generation module 404 is used to iteratively perform the genetic control multiple times to obtain a target population, which includes multiple target individuals, and any one of the target individuals represents a pedicle screw planning scheme.

[0100] In this embodiment of the application, the genetic control module 403 can specifically be used for: Based on the fitness score, multiple individuals in the population are sorted, and a predetermined number of individuals with the highest ranking are copied into the next generation population. The first and second parent generations are determined from the population, and crossover mutation is performed based on the first and second parent generations to generate multiple offspring individuals. The multiple offspring individuals are then copied into the next generation population.

[0101] In one possible implementation of this application embodiment, the genetic control module 403 may also be used for: Using the first parent generation as the base sequence and the second parent generation as the cross sequence, the base sequence is traversed sequentially; wherein, each sequence value in the base sequence and the cross sequence is a numerical value corresponding to various pinning information; When any sequence value in the base sequence satisfies the crossover probability, the corresponding sequence value in the crossover sequence is replaced in the base sequence to obtain the offspring sequence; Sequence mutation is performed on the offspring sequence to generate multiple offspring individuals.

[0102] In another possible implementation of this application embodiment, the genetic control module 403 may also be used for: Iterate through the child sequence in sequence; When any sequence value in the offspring sequence meets the mutation probability, the sequence value of the corresponding part in the offspring sequence is modified to obtain the offspring individual.

[0103] In this embodiment, the various types of pin placement information include screw entry point position, screw direction, screw diameter, and screw length. The screw entry point position of any screw in each individual of the initialized population is randomly determined within the vertebral space where the pin is to be placed. The screw direction is randomly determined, and the screw diameter and screw length are randomly selected from a screw database. The fitness function constructed based on these various pin placement information includes the following: The region reward function is used to represent the ratio of the volume of the screw in the overlapping region of the cone to the total volume of the screw. A symmetric reward function is used to represent the angular difference between the first and second screws and the vertebral midline. The bone mineral density reward function is used to represent the ratio of the cumulative bone mineral density of the screw within the vertebral body to the overall bone mineral density of the vertebral body. The stability reward function is used to represent the ratio of screw diameter to maximum diameter, and the ratio of screw length to maximum length; The collision penalty function is used to represent the collision state between the first screw and the second screw; The penetration penalty function is used to represent the positional relationship between the screw stop point and the vertebral body.

[0104] In this embodiment of the application, the fitness evaluation module 402 can specifically be used for: Determine the weights for each of the fitness functions; Based on the weights, the reward and penalty values ​​corresponding to multiple fitness functions are weighted and summed to obtain the fitness score of each individual.

[0105] In this embodiment of the application, the target population generation module 404 can specifically be used for: For the generated next generation population, genetic control of the next generation population is repeated based on the fitness score; When the number of iterations for genetic control of the next generation population reaches a set threshold, the current population is used as the target population. Multiple target individuals are identified from the target population.

[0106] In one possible implementation of this application embodiment, the target population generation module 404 may also be used for: Sort multiple individuals in the target population according to the fitness score; A predetermined proportion of individuals with the highest ranking are retained as target individuals representing the pedicle screw planning scheme.

[0107] This application provides a pedicle screw planning device, which can be a computer device or a module or unit with corresponding functions within a computer device. Using this device, the steps in the aforementioned method embodiments can be implemented.

[0108] As the apparatus embodiments are basically similar to the method embodiments, they are described in a relatively simple manner. For relevant details, please refer to the description in the method embodiment section.

[0109] Reference Figure 5 The diagram illustrates a computer device provided in an embodiment of this application. Figure 5 As shown, the computer device 500 in this embodiment includes: a processor 510, a memory 520, and a computer program 521 stored in the memory 520 and executable on the processor 510. When the processor 510 executes the computer program 521, it implements the steps in the various embodiments of the pedicle screw planning method described above, for example... Figure 1 The steps S101 to S104 are shown. Alternatively, when the processor 510 executes the computer program 521, it implements the functions of each module / unit in the above-described device embodiments, for example... Figure 4 The functions of modules 401 to 404 are shown.

[0110] For example, the computer program 521 can be divided into one or more modules / units, which are stored in the memory 520 and executed by the processor 510 to complete this application. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, which can be used to describe the execution process of the computer program 521 in the computer device 500. For example, the computer program 521 can be divided into an initialization module, a fitness evaluation module, a genetic control module, and a target population generation module, with the specific functions of each module as follows: An initialization module is used for population initialization, wherein the population includes multiple individuals, and any individual is defined as including a first screw and a second screw, wherein the first screw and the second screw each have multiple screw placement information; The fitness evaluation module is used to evaluate the fitness of each individual in the population according to multiple fitness functions to obtain the fitness score of each individual. The fitness functions are constructed based on multiple pinning information. A genetic control module is used to genetically control the population based on the fitness score to generate the next generation population; The target population generation module is used to iteratively perform the genetic control multiple times to obtain a target population, which includes multiple target individuals, and any one of the target individuals represents a pedicle screw planning scheme.

[0111] The computer device 500 may be a device capable of implementing the relevant functions in the foregoing method embodiments. The computer device 500 may be a desktop computer, a cloud server, or other similar devices. The computer device 500 may include, but is not limited to, a processor 510 and a memory 520. Those skilled in the art will understand that... Figure 5 This is merely one example of computer device 500 and does not constitute a limitation on computer device 500. It may include more or fewer components than shown, or combine certain components, or different components. For example, computer device 500 may also include input / output devices, network access devices, buses, etc.

[0112] The processor 510 can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0113] The memory 520 can be an internal storage unit of the computer device 500, such as a hard disk or memory of the computer device 500. The memory 520 can also be an external storage device of the computer device 500, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, etc., equipped on the computer device 500. Furthermore, the memory 520 can include both internal and external storage units of the computer device 500. The memory 520 is used to store the computer program 521 and other programs and data required by the computer device 500. The memory 520 can also be used to temporarily store data that has been output or will be output.

[0114] This application also discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the methods described in the foregoing embodiments.

[0115] This application also discloses a computer-readable storage medium storing a computer program that, when executed by a computer, implements the methods described in the foregoing embodiments.

[0116] This application also discloses a computer program product, including a computer program that, when run on a computer, causes the computer to perform the methods described in the foregoing embodiments.

[0117] The embodiments described above are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A pedicle screw planning method, characterized by, include: Based on the spatial relationship between the vertebral body and the screw, multiple fitness functions are constructed to evaluate the placement plan of the corresponding screw. The multiple fitness functions include a reward function and a penalty function. The fitness score of each individual in the population is obtained by evaluating multiple fitness functions, wherein any individual in the population is defined as including the first screw and the second screw of the vertebral body to be inserted. Based on the fitness score, the population is subjected to multiple genetic control iterations to obtain the target population; Multiple target individuals are identified from the target population, and these multiple target individuals represent multiple executable pedicle screw planning schemes.

2. The method of claim 1, wherein, The first screw and the second screw each have multiple screw placement information, including screw entry point position, screw direction, screw diameter, and screw length; The reward function includes any one or more of the following functions: The region reward function is used to represent the ratio of the volume of the screw in the overlapping region of the cone to the total volume of the screw. A symmetric reward function is used to represent the angular difference between the first and second screws and the vertebral midline. The bone mineral density reward function is used to represent the ratio of the cumulative bone mineral density of the screw within the vertebral body to the overall bone mineral density of the vertebral body. The stability reward function is used to represent the ratio of screw diameter to maximum diameter, and the ratio of screw length to maximum length; The penalty function includes any one or more of the following functions: The collision penalty function is used to represent the collision state between the first screw and the second screw; The penetration penalty function is used to represent the positional relationship between the screw stop point and the vertebral body.

3. The method of claim 2, wherein, If the first screw collides with the second screw, the reward / penalty value corresponding to the collision penalty function is -1; if the first screw and the second screw do not collide, the reward / penalty value corresponding to the collision penalty function is 1. If the stop point of the first screw or the second screw penetrates outside the vertebral body, the reward / penalty value corresponding to the penetration penalty function is -1; if the stop points of both the first screw and the second screw do not penetrate outside the vertebral body, the reward / penalty value corresponding to the penetration penalty function is 1.

4. The method according to any one of claims 1 to 3, characterized in that, The process of iteratively performing genetic control on the population based on the fitness score to obtain the target population includes: Based on the fitness score, multiple strategies are used to genetically control the population to generate the next generation population. The genetic control is performed iteratively on the population multiple times to obtain the target population.

5. The method of claim 4, wherein, The various strategies include elite strategies and roulette wheel strategies; the genetic control of the population based on the fitness score, using multiple strategies simultaneously to generate the next generation population, includes: An elite strategy is employed to directly replicate a predetermined number of individuals with the highest fitness scores from the current population into the next generation; and, An individual from any region is randomly selected as the parent using a roulette wheel strategy. By performing crossover and mutation on the parent individuals, the corresponding offspring individuals are generated and placed into the next generation population.

6. The method of claim 5, wherein, The process of randomly selecting individuals from any region as parents using a roulette wheel strategy, and generating corresponding offspring individuals by performing crossover and mutation on the parent individuals, includes: Determine the first and second parent generations from the population; Using the first parent as the base sequence and the second parent as the cross sequence, the base sequence is traversed sequentially; wherein, each sequence value in the base sequence and the cross sequence is a value corresponding to multiple pinning information; When any sequence value in the base sequence satisfies the crossover probability, the corresponding sequence value in the crossover sequence is replaced in the base sequence to obtain the offspring sequence; Iterate through the child sequence in sequence; When any sequence value in the offspring sequence satisfies the mutation probability, the sequence value of the corresponding part in the offspring sequence is modified to obtain an offspring individual, which is then placed into the next generation population.

7. The method according to any one of claims 1 to 3 or 5 to 6, characterized in that, The fitness score of each individual in the population is obtained by evaluating multiple fitness functions, including: Determine the weights for each of the fitness functions; Based on the weights, the reward and penalty values ​​corresponding to multiple fitness functions are weighted and summed to obtain the fitness score of each individual.

8. The method of claim 7, wherein, The step of identifying multiple target individuals from the target population includes: Sort multiple individuals in the target population according to the fitness score; A predetermined proportion of individuals with the highest ranking are retained as target individuals representing executable pedicle screw planning schemes.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it causes the computer device to implement the method as described in any one of claims 1 to 8.

10. A computer program product comprising a computer program, characterized in that, When the computer program is run, the method as described in any one of claims 1 to 8 is performed.

Images