Spinal surgery scheme decision judgment system and method based on artificial intelligence

Abstract

The invention belongs to the technical field of medical information processing, and particularly relates to a spinal surgery scheme decision judgment system and method based on artificial intelligence. Comprising a real-time data acquisition module used for synchronously acquiring multi-modal data used for representing an operation state from an operation site; the digital twinborn state updating module is used for calculating a geometric configuration and an internal stress field so as to generate a digital twinborn model; the dynamic risk assessment module is used for calculating a prospective stress risk and a proximity risk; generating a dynamic risk comprehensive index; the self-adaptive decision re-planning module is used for generating a new optimal operation path; and the man-machine interaction and instruction output module is used for visually presenting the new optimal operation path and the risk map. According to the method, the digital twinborn model synchronous with the physical world is constructed in real time, so that the biomechanical stress risk possibly caused on the future path of the surgical instrument and the collision risk of the surgical instrument and the key anatomical structure can be calculated and quantified in a prospective manner.

Description

Technical Field

[0001] The present invention belongs to the field of medical information processing technology, and specifically relates to a spinal surgery plan decision-making system and method based on artificial intelligence. Background Art

[0002] Traditional spinal surgery planning relies heavily on static preoperative imaging data, such as CT or MRI scans. Doctors use these static images to plan the surgical path and determine a theoretically optimal approach. However, this approach inherently lacks synergy: offline preoperative planning is disconnected from the real-time dynamics of surgery.

[0003] During actual surgery, the surgeon's manipulation, the forces applied by instruments, and the traction of tissues all cause continuous, dynamic changes in the patient's spinal geometry and internal stresses. These changes cannot be predicted by preoperative static imaging. Consequently, the accuracy and safety of pre-established optimal surgical plans will gradually decrease, or even fail, in the complex and ever-changing real-world intraoperative environment. Existing technologies generally lack a closed-loop system that can perceive intraoperative changes in real time, dynamically update models, and adaptively adjust the surgical path. This system fails to address the core technical challenge of the progressive failure of preoperative static plans in a dynamic surgical environment, posing potential risks to the safety and accuracy of surgery. Summary of the Invention

[0004] The purpose of the present invention is to provide a spinal surgery plan decision-making system and method based on artificial intelligence, which solves the problems existing in the background technology.

[0005] To solve the above technical problems, the present invention provides an artificial intelligence-based spinal surgery decision-making system, comprising: A real-time data acquisition module is used to synchronously acquire multimodal data representing the surgical status from the surgical site; wherein the multimodal data includes intraoperative imaging data, instrument three-dimensional spatial coordinates and posture data streams, and external force data; a digital twin state update module, configured to receive the multimodal data and calculate a geometric configuration based on a preset patient-specific biomechanical model; the digital twin state update module is further configured to solve an internal stress field based on the updated geometric configuration, and combine the geometric configuration and the internal stress field to generate a digital twin model; a dynamic risk assessment module for calculating forward-looking stress risk and proximity risk for future path points of the device based on the digital twin model; the dynamic risk assessment module is further configured to perform a weighted fusion of the calculated forward-looking stress risk and proximity risk to generate a dynamic risk composite index; an adaptive decision-making replanning module, configured to generate a new optimal surgical path by minimizing a cost function when the dynamic risk comprehensive index exceeds a preset risk threshold; The human-computer interaction and instruction output module is used to visualize the new optimal surgical path and the risk map generated according to the dynamic risk comprehensive index.

[0006] Preferably, the calculation of the geometric configuration is to use the intraoperative imaging data and the three-dimensional spatial coordinates and posture data stream of the instrument to provide displacement boundary conditions for the biomechanical model, and at the same time apply the external force data as a load to the biomechanical model, and then solve the overall deformation field to update the geometric configuration.

[0007] Preferably, the forward-looking stress risk is calculated as follows: assuming that the instrument moves to the future path point and applies a preset standard operating force, the biomechanical model is called to calculate the predicted stress field, and the maximum value of the predicted stress field in the preset key area is compared with the safety threshold and then normalized.

[0008] Preferably, the proximity risk is calculated by calculating the Euclidean distance between the future path point and the critical hazardous anatomical structure surface extracted from the current geometric configuration, and converting the Euclidean distance into a risk value using an inverse function.

[0009] Preferably, the weighted fusion of the dynamic risk comprehensive index adopts a preset weight coefficient; wherein the weight coefficient is set by an expert system according to the surgical stage.

[0010] Preferably, the cost function of the adaptive decision replanning module includes a path average risk term for characterizing path risk and a target deviation term for characterizing surgical effect.

[0011] Preferably, the target deviation term is determined by comparing the difference between the final spinal geometric configuration predicted to be achieved after adopting the new optimal surgical path and the ultimate surgical target configuration set before the operation.

[0012] Preferably, when the dynamic risk comprehensive index does not exceed the preset risk threshold, the adaptive decision replanning module does not generate the new optimal surgical path.

[0013] Preferably, the artificial intelligence-based spinal surgery decision-making method includes: The real-time data acquisition module synchronously acquires intraoperative imaging data, instrument 3D spatial coordinates and posture data stream, and external force data from the surgical site; The digital twin state update module calculates the geometric configuration and the internal stress field, and generates a digital twin model by combining the geometric configuration and the internal stress field; The dynamic risk assessment module calculates the forward-looking stress risk and proximity risk, performs weighted fusion on the risks, and generates a dynamic risk composite index; Determining whether the dynamic risk comprehensive index exceeds a preset risk threshold; If the dynamic risk comprehensive index exceeds the preset risk threshold, the adaptive decision replanning module generates a new optimal surgical path, and the human-computer interaction and instruction output module visualizes the new optimal surgical path and the risk map.

[0014] The artificial intelligence-based spinal surgery decision-making system and method provided by the present invention have the following beneficial effects: 1. By constructing a digital twin model synchronized with the physical world in real time, this system proactively calculates and quantifies the biomechanical stress risks and collision risks with critical anatomical structures that may be induced by the surgical instrument's future path. When the combined risk index exceeds a safety threshold, the system proactively generates and recommends an optimal path to avoid the risk, guiding the surgeon to avoid high-risk procedures and effectively preventing damage to vital tissues such as the nerves and spinal cord due to stress overload or accidental contact.

[0015] 2. This invention establishes a complete closed-loop information flow of "perception-modeling-decision-execution," transforming surgery from a linear process based on static preplans to an intelligent process that continuously self-adjusts based on real-time intraoperative dynamics. The system intelligently adjusts the weighting of different risks based on the characteristics of different surgical stages, enabling more context-aware decision-making. 3. When replanning the path, the present invention sets a cost function that not only considers safety but also includes an assessment of the final surgical outcome. By calculating the deviation between the predicted corrective effect achieved by the new path and the ideal preoperative target, the system can minimize risks while ensuring that the surgery proceeds toward the intended treatment goal, resolving the potential conflict between safety and efficacy and ensuring surgical accuracy. 4. This invention establishes a risk activation threshold, avoiding frequent interventions due to minor fluctuations. The system intervenes only when a high risk is identified. It then presents the risk map and recommended path in an intuitive, visual format to the physician as a decision aid. This operating model provides critical safety assurance while fully respecting the surgeon's operational autonomy, maintaining a coherent and smooth surgical process. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention, and those skilled in the art can derive other drawings based on these drawings without inventive effort.

[0017] Figure 1 This is a logic block diagram of the spinal surgery plan decision-making system based on artificial intelligence of the present invention; DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0019] Example 1 See also Figure 1 , an artificial intelligence-based spinal surgery decision-making system, including: A real-time data acquisition module is used to synchronously acquire multimodal data from the surgical site to characterize the surgical status; the multimodal data includes intraoperative imaging data, instrument 3D spatial coordinates and posture data streams, and external force data; A digital twin state update module is used to receive multimodal data and calculate the geometric configuration based on a preset patient-specific biomechanical model. The digital twin state update module is also used to solve the internal stress field based on the updated geometric configuration and combine the geometric configuration and internal stress field to generate a digital twin model. A dynamic risk assessment module is used to calculate forward-looking stress risk and proximity risk for future path points of the device based on the digital twin model. The dynamic risk assessment module is also used to perform a weighted fusion of the calculated forward-looking stress risk and proximity risk to generate a dynamic risk composite index. An adaptive decision-making replanning module is used to generate a new optimal surgical path by minimizing the cost function when the dynamic risk comprehensive index exceeds the preset risk threshold; The human-computer interaction and command output module is used to visualize the new optimal surgical path and the risk map generated based on the dynamic risk comprehensive index; The artificial intelligence-based spinal surgery decision-making system provided by the embodiment of the present invention constructs a complete "perception-modeling-decision-execution" closed-loop information flow through its internal interconnected functional modules; The core technical contribution of this system lies in its architectural design, which aims to overcome the inherent flaw of existing technologies in decoupling preoperative offline planning from intraoperative online execution. Through the real-time data acquisition module's precise perception of dynamic changes during surgery, the digital twin state update module's real-time modeling of the patient's spinal status, the dynamic risk assessment module's decision-making analysis of potential risks, and the adaptive decision replanning module's generation of the optimal path when necessary, which is then output by the human-computer interaction module to guide execution, this system tightly couples the previously separate static planning and dynamic execution processes. The establishment of this mechanism enables the system to track in real time the changes in spinal morphology and stress caused by the surgeon's operation, solving the core technical problem of the progressive failure of the preoperative static optimal plan in a dynamic surgical environment, and realizing a technical paradigm shift from static image-based planning to adaptive guidance based on real-time dynamic models, thereby significantly improving the safety of the operation and the ultimate correction accuracy in complex spinal surgery.

[0020] Example 2 The calculation of geometric configuration uses intraoperative imaging data and the instrument's three-dimensional spatial coordinates and posture data stream to provide displacement boundary conditions for the biomechanical model. At the same time, external force data is applied to the biomechanical model as a load, and the overall deformation field is calculated to update the geometric configuration. In the embodiment of the present invention, the calculation process of the geometric configuration is a digital and accurate reproduction of the physical world, and its calculation logic is defined by the following formula:

[0021] This formula draws mathematical inspiration from state-space equations in dynamic system modeling and innovatively applies them to the description of the geometric evolution of biological tissue. Its technical motivation is to accurately capture and simulate the geometric evolution of the patient's spinal tissue caused by surgical operations in real time. In the composition of this formula, and The three-dimensional geometric configurations of the spine at the current moment and the previous moment, respectively, exist in the form of a three-dimensional coordinate matrix of each key anatomical landmark, and its physical dimension is length; is the geometric configuration change obtained through calculation, that is, the deformation field composed of the displacement vectors of each landmark point, and its dimension is also length; the input parameters of the formula come from the following sources: Represents the intraoperative imaging data sequence acquired by the C-arm X-ray machine and other equipment in the real-time data acquisition module; It is also the instrument three-dimensional spatial coordinate and posture data stream provided by the surgical navigation system which is also the output of the real-time data acquisition module; The real-time external force data is installed on the surgical instrument and collected by the real-time data acquisition module; It is a biomechanical finite element model constructed by the digital twin state update module based on the patient's preoperative high-precision CT / MRI data, and has completed personalized settings of parameters such as Young's modulus and Poisson's ratio; Indicates timestamp, ensuring data synchronization; At the application level, the inherent logic of the geometric configuration calculation process is to use real-time images and tool coordinates , through image registration and point cloud alignment technology, for the biomechanical model Apply precise displacement boundary conditions; at the same time, the external forces measured in real time are Applied as external loads to the model; biomechanical models Under the dual drive of displacement constraints and external force loads, nonlinear solution is performed to output the overall deformation field that is completely consistent with the physical laws. ; This calculation mechanism ensures that the geometric configuration in the digital twin model It is not a simple morphological simulation, but a precise state update that integrates multimodal real-time data and follows the laws of biomechanics, thus providing a high-fidelity geometric foundation for subsequent stress analysis and risk assessment. Its accuracy far exceeds that of traditional methods that rely on static preoperative data.

[0022] Example 3 The forward-looking stress risk is calculated by assuming that the device moves to a future path point and applies a preset standard operating force. The biomechanical model is used to calculate the predicted stress field. The maximum value of the predicted stress field in the preset critical area is compared with the safety threshold and then normalized. In the embodiment of the present invention, the calculation of forward-looking stress risk is intended to quantify the biomechanical damage that may be caused by future operations. Its core is a dimensionless forward-looking stress risk function ; The establishment of this function can predict and evaluate the stress overlimit risk that may be caused by dangerous operations before they actually occur; The system assumes that the surgical instrument moves to a potential waypoint in the immediate future , and apply a preset standard operating force at this position The standard operating force can be pre-set based on the typical instrument load data in the biomechanics literature, or obtained based on the preoperative calibration of the operator's operating force to simulate a typical and representative surgical operation. The system calls the biomechanical model to perform a rapid virtual loading calculation to obtain a predicted stress field. ; The system extracts the predicted stress field Maximum stress values ​​at pre-defined critical anatomical areas and biomechanical safety threshold For comparison and normalization, the normalization calculation can be done using the formula: To ensure the feasibility of this step, the safety threshold The determination is based on public experimental data in the field of biomechanics and authoritative clinical safety consensus, ensuring its scientificity and reliability; The application of this calculation method gives the system foresight capabilities; by calculating the stress risk index for each potential surgical operation point, the system can identify in advance those dangerous operations that may cause excessive compression of the spinal cord or nerve roots; this forward-looking risk quantification provides key decision-making input for subsequent adaptive path replanning, which can guide surgeons to actively avoid high-risk operations, thereby greatly improving the safety of the operation.

[0023] Example 4 The proximity risk is calculated by calculating the Euclidean distance between the future waypoint and the critical hazardous anatomical structure surface extracted from the current geometric configuration, and converting the Euclidean distance into a risk value using an inverse function; In the embodiment of the present invention, the calculation of proximity risk is used to assess the collision risk between surgical instruments and critical anatomical structures. The core of the calculation is a dimensionless proximity risk function. ; The fundamental technical motivation for establishing this function is to convert the physical quantity of distance in geometric space into a risk metric that can be understood and used by decision-making systems, thereby preventing surgical instruments from causing accidental physical damage in complex and narrow anatomical spaces. The system updates the geometry from the current Extract the surface model of key dangerous anatomical structures , and then calculate the future instrument path points With the surface The shortest Euclidean distance between ; The system uses an inverse function to Converted into risk value, for example: when the distance Less than or equal to the preset warning distance When the risk value When the distance Greater than the warning distance When the risk value is 0, to ensure the feasibility of this step, the warning distance The value of the risk surge index is determined based on statistical analysis of a large number of clinical surgical cases and expert experience to strike a balance between providing effective early warning and avoiding excessive interference. It is an adjustable parameter used to control the steepness of the risk curve. Its default value can be set based on expert experience, and the surgeon can make real-time fine-tuning through the human-computer interface to adapt to the sensitivity requirements of different surgical areas; The application of this calculation method is equivalent to establishing a digital safety boundary around all critical dangerous anatomical structures; when the planned path of the surgical instrument invades this boundary, the proximity risk value The calculation of proximity risk provides the decision-making system with a clear and sensitive quantitative indicator of geometric collision risk, ensuring that subsequent path planning can effectively avoid these high-risk areas, thereby significantly improving the accuracy and safety of surgical operations in complex anatomical environments.

[0024] Example 5 The weighted fusion of the dynamic risk comprehensive index uses a preset weight coefficient; wherein the weight coefficient is set by the expert system according to the surgical stage; In the embodiment of the present invention, the weighted fusion of the dynamic risk comprehensive index is the core step of integrating the aforementioned risk sources of different natures to form a unified decision-making basis; this fusion process is achieved through the following customized heuristic risk model formula:

[0025] Real surgical risks are multidimensional, and any single risk assessment is one-sided. Therefore, a comprehensive assessment model must be established that integrates biomechanical risk and geometric proximity risk to provide a global, quantifiable decision-making indicator. In the composition of this formula, It is the dimensionless dynamic risk comprehensive index outputted at the end; and They are the forward-looking stress risk and proximity risk calculated above, both of which are dimensionless, ensuring the consistency of the dimensions on both sides of the formula; the core lies in the dimensionless weight coefficient and , they satisfy These weight coefficients are automatically set by the built-in expert system according to the current surgical stage; for example, during the pedicle screw implantation stage, the risk of the instrument approaching the nerve root is the main contradiction, and the system will automatically increase the proximity risk weight. During the rotation phase of the orthotic rod, the overall stress distribution of the spine is the main consideration, and the system will increase the forward-looking stress risk weight. The value of The application of this weighted fusion mechanism gives the system a high degree of adaptability; it can dynamically adjust the focus on different risk sources according to the specific context of the surgery and the individual characteristics of the patient; and produce a dynamic risk composite index that is more comprehensive and accurate than any single risk metric. This index can more realistically reflect the overall risk level of future operations, thereby providing the most critical and reliable basis for judging whether the subsequent decision-making replanning module should be triggered and how to plan.

[0026] Example 6 The cost function of the adaptive decision replanning module includes a path average risk term used to characterize path risk and a target deviation term used to characterize surgical effectiveness. The target deviation term is determined by comparing the difference between the final spinal geometric configuration predicted to be achieved after adopting the new optimal surgical approach and the ultimate surgical target configuration set before surgery; In the embodiment of the present invention, the core of the adaptive decision-making replanning module is a module for solving the optimal surgical path. The cost function of , its optimization objective can be expressed as:

[0027] The framework of this cost function originates from control theory and robot path planning, but its internal structure is a fundamental innovation designed to solve the specific technical problems of this invention. The technical motivation is that when the system determines that the current path is too risky, it must generate a new path that balances safety and efficiency. The cost function consists of two logically related core parts; the first is the path average risk term , whose input is the output of the dynamic risk assessment module; by following the new path Dimensionless risk score Perform line integration, including the spinal surgery decision-making system based on artificial intelligence is the arc length variable along the path and is divided by the total length of the path , the dimensionless processing of the overall risk of the path is completed, and a dimensionless average risk value is obtained to characterize the safety of the entire path; the second item is the target deviation item , which is directly related to the final effect of the operation; among them, A biomechanical model Predicted, fully executed new path The final spinal geometry that can be achieved after It is the ultimate target configuration set during preoperative planning and represents the most ideal surgical outcome; function Responsible for calculating the difference between the two configurations. For example, a quantitative, dimensionless difference value can be obtained by calculating the root mean square deviation between the key anatomical landmarks preset on the two models, thereby quantifying the degree of deviation between the new path and the ultimate goal of the surgery. The output is also dimensionless. It is a dimensionless hyperparameter used to balance the weights of the above two items. Its value can be preset by the expert system according to the type of surgery or dynamically adjusted by the surgeon based on clinical judgment during the surgery. Specifically, for cervical spine surgery with higher risks, a higher value can be preset. A value (such as 0.8) can force the path planning to strictly avoid risks; for routine lumbar degeneration surgery, a moderate The value (such as 0.5) is used to balance safety and correction efficiency.

[0028] This cost function provides a clear and computable mathematical goal for the path optimization problem; the adaptive decision replanning module can algorithmically find a new optimal path by minimizing this cost function. This path can not only effectively bypass high-risk areas, that is, minimize the average risk item of the path, but also ensure that the surgical process moves towards the preset final goal to the greatest extent, that is, minimize the target deviation item; this mechanism resolves the inherent conflict between safety and effectiveness during surgery, and ensures that the decision recommendations provided by the system are always safe and of high clinical value.

[0029] Example 7 When the dynamic risk comprehensive index does not exceed the preset risk threshold, the adaptive decision replanning module does not generate a new optimal surgical path; In the embodiment of the present invention, the activation of the adaptive decision-making replanning module is subject to a key condition constraint: only when the dynamic risk comprehensive index calculated by the dynamic risk assessment module is The adaptive decision replanning module will be triggered only when the instantaneous value or the predicted value on the future path exceeds a pre-set risk threshold; on the contrary, when When the threshold is always below the safety threshold, the adaptive decision-making replanning module remains silent and does not intervene. This operating logic is designed to ensure the stability of the system and the harmony of human-computer interaction; it avoids the system from frequently proposing path modification suggestions due to minor risk fluctuations; by establishing clear trigger thresholds, the system can accurately maintain a monitoring status during routine and safe operation stages, and only actively intervene and provide decision support when risks beyond the safety range do occur; this working mode enables the system to operate efficiently as a decision-making support tool, while respecting and maintaining the surgeon's operational autonomy and the smoothness of the surgical process to the greatest extent possible while ensuring safety.

[0030] Example 8 The artificial intelligence-based spinal surgery plan decision-making method includes: a real-time data acquisition module synchronously acquires intraoperative imaging data, instrument three-dimensional spatial coordinates and posture data streams, and external force data from the surgical site; a digital twin state update module calculates the geometric configuration and internal stress field, and combines the geometric configuration and internal stress field to generate a digital twin model; a dynamic risk assessment module calculates the forward-looking stress risk and proximity risk, performs weighted fusion on the risks, and generates a dynamic risk comprehensive index; determines whether the dynamic risk comprehensive index exceeds a preset risk threshold; if the dynamic risk comprehensive index exceeds the preset risk threshold, the adaptive decision replanning module generates a new optimal surgical path, and the human-computer interaction and command output module visualizes the new optimal surgical path and the risk map; The artificial intelligence-based spinal surgery decision-making method provided by the embodiment of the present invention is presented as a continuous closed-loop processing flow; The method consists of a real-time data acquisition module that synchronously acquires multimodal data streams from the surgical site. The digital twin state update module receives this data stream and updates the spine's geometric configuration and internal stress field based on the patient's personalized biomechanical model. Together, these two modules form a digital twin model that is synchronized with the physical world in real time. Based on this high-fidelity model, the dynamic risk assessment module proactively calculates stress risk and proximity risk for the instrument's future path, weighting these two risks with adjustable weights to generate a single, comprehensive dynamic risk index. The system continuously assesses this index, and once its value exceeds a preset safety threshold, the adaptive decision-making replanning module is immediately activated to generate a new optimal surgical path. To ensure the feasibility of this assessment step, the preset risk threshold is a key parameter. Its specific value can be pre-set by a team of clinical experts based on the complexity of the surgery or obtained as a benchmark through statistical analysis of historical safe surgical data. The human-computer interaction and command output module presents this new path and a highlighted risk map rendered based on the risk index in a highly intuitive visual format. The surgeon's subsequent actions are captured again by the real-time data acquisition module, initiating a new cycle. This method transforms the entire surgical process from a linear task of executing a static plan to an adaptive, intelligent process of continuous "perception-assessment-decision-feedback", thereby improving the safety, accuracy and predictability of complex spinal surgery.

[0031] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. The artificial intelligence-based spinal surgery decision-making system is characterized by: include: A real-time data acquisition module is used to synchronously acquire multimodal data representing the surgical status from the surgical site; wherein the multimodal data includes intraoperative imaging data, instrument three-dimensional spatial coordinates and posture data streams, and external force data; a digital twin state update module, configured to receive the multimodal data and calculate a geometric configuration based on a preset patient-specific biomechanical model; the digital twin state update module is further configured to solve an internal stress field based on the updated geometric configuration, and combine the geometric configuration and the internal stress field to generate a digital twin model; a dynamic risk assessment module for calculating forward-looking stress risk and proximity risk for future path points of the device based on the digital twin model; the dynamic risk assessment module is further configured to perform a weighted fusion of the calculated forward-looking stress risk and proximity risk to generate a dynamic risk composite index; an adaptive decision-making replanning module, configured to generate a new optimal surgical path by minimizing a cost function when the dynamic risk comprehensive index exceeds a preset risk threshold; The human-computer interaction and instruction output module is used to visualize the new optimal surgical path and the risk map generated according to the dynamic risk comprehensive index.

2. The artificial intelligence-based spinal surgery decision-making system according to claim 1, characterized in that: The calculation of the geometric configuration uses the intraoperative imaging data and the three-dimensional spatial coordinates and posture data stream of the instrument to provide displacement boundary conditions for the biomechanical model, and at the same time applies the external force data as a load to the biomechanical model, and then solves the overall deformation field to update the geometric configuration.

3. The artificial intelligence-based spinal surgery decision-making system according to claim 1, characterized in that: The forward-looking stress risk is calculated as follows: assuming that the instrument moves to the future path point and applies a preset standard operating force, the biomechanical model is called to calculate the predicted stress field, and the maximum value of the predicted stress field in the preset key area is compared with the safety threshold and then normalized.

4. The artificial intelligence-based spinal surgery decision-making system according to claim 1, characterized in that: The proximity risk is calculated by calculating the Euclidean distance between the future path point and the critical hazardous anatomical structure surface extracted from the current geometric configuration, and converting the Euclidean distance into a risk value using an inverse function.

5. The artificial intelligence-based spinal surgery decision-making system according to claim 1, characterized in that: The weighted fusion of the dynamic risk comprehensive index adopts a preset weight coefficient; wherein the weight coefficient is set by an expert system according to the operation stage.

6. The artificial intelligence-based spinal surgery decision-making system according to claim 1, characterized in that: The cost function of the adaptive decision replanning module includes a path average risk term for characterizing path risk and a target deviation term for characterizing surgical effect.

7. The artificial intelligence-based spinal surgery decision-making system according to claim 6, characterized in that: The target deviation term is determined by comparing the difference between the final spinal geometric configuration predicted to be achieved after adopting the new optimal surgical path and the ultimate surgical target configuration set before the operation.

8. The artificial intelligence-based spinal surgery decision-making system according to claim 1, characterized in that: When the dynamic risk comprehensive index does not exceed the preset risk threshold, the adaptive decision replanning module does not generate the new optimal surgical path.

9. A method for decision-making and judging a spinal surgery plan based on artificial intelligence, applied to a spinal surgery plan decision-making and judging system based on artificial intelligence according to any one of claims 1 to 8, characterized in that: include: The real-time data acquisition module synchronously acquires intraoperative imaging data, instrument 3D spatial coordinates and posture data stream, and external force data from the surgical site; The digital twin state update module calculates the geometric configuration and the internal stress field, and generates a digital twin model by combining the geometric configuration and the internal stress field; The dynamic risk assessment module calculates the forward-looking stress risk and proximity risk, performs weighted fusion on the risks, and generates a dynamic risk composite index; Determining whether the dynamic risk comprehensive index exceeds a preset risk threshold; If the dynamic risk comprehensive index exceeds the preset risk threshold, the adaptive decision replanning module generates a new optimal surgical path, and the human-computer interaction and instruction output module visualizes the new optimal surgical path and the risk map.

Images