Medical endoscope operation trajectory intelligent navigation system

Abstract

The present application relates to the technical field of intelligent navigation of medical endoscopes, in particular to an intelligent navigation system for operation trajectory of medical endoscopes; the system comprises data acquisition, deviation processing, state evaluation, risk prediction and adaptive control modules; by acquiring the predicted trajectory and real-time positioning data, the navigation deviation is determined; the core is to calculate the human-machine trust decay index and brain tissue deformation index according to the deviation and key anatomical markers, and to establish a risk model to predict the future trust decay trend; when the predicted risk value exceeds the preset loss of credit threshold, the system will adaptively switch the navigation mode; the present application realizes the transformation from passive response to deviation to active prediction of risk, and can identify and manage the risk of human-machine collaborative rupture in advance.

Description

Technical Field

[0001] This invention relates to the field of intelligent navigation technology for medical endoscopes, specifically to an intelligent navigation system for the operation trajectory of a medical endoscope. Background Technology

[0002] In endoscopic surgeries that rely on preoperative planning, dynamic physiological changes such as brain tissue deformation during the procedure often lead to deviations between the preset navigation path and the actual anatomical structure. These deviations reduce the accuracy of navigation information, which in turn causes doctors to distrust the system's guidance. When doctors rely on their own experience to make significant corrections, it not only reduces the efficiency of human-machine collaboration but may also increase surgical risks due to navigation failure, posing a potential threat to safety. Therefore, how to assess and predict this risk in real time and adaptively adjust the navigation strategy before the human-machine trust relationship breaks down, ensuring the safety and efficiency of the surgery, is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0003] To solve the above-mentioned technical problems, the present invention provides an intelligent navigation system for medical endoscope operation trajectory. Specifically, the technical solution of the present invention includes:

[0004] The data acquisition module is used to acquire the predicted trajectory point sequence, real-time positioning point sequence, actual operation trajectory points, and coordinates of key anatomical landmarks;

[0005] The deviation processing module is used to determine the navigation deviation vector and operation correction vector based on the predicted trajectory point sequence, the real-time positioning point sequence, and the actual operation trajectory points.

[0006] The status assessment module is used to calculate the human-machine trust decay index and brain tissue deformation index based on the navigation deviation vector, operation correction vector, and coordinates of key anatomical landmarks.

[0007] The risk prediction module is used to predict the future trust decay index based on the human-machine trust decay index and the brain tissue deformation index.

[0008] The adaptive control module is used to determine the navigation mode control parameters based on the future trust decay index and the preset trust loss threshold, and adaptively switch the navigation mode according to the navigation mode control parameters.

[0009] Preferably, the status assessment module calculates the human-machine trust decay index, including:

[0010] Calculate the navigation deviation magnitude of the navigation deviation vector and the operation correction magnitude of the operation correction vector;

[0011] The first normalized deviation value is determined based on the navigation deviation modulus and the preset first reference deviation modulus.

[0012] The second normalized deviation value is determined based on the operational correction modulus and the preset second reference deviation modulus.

[0013] The human-machine trust decay index is generated by combining the first normalized deviation value and the second normalized deviation value, and by linearly weighting and summing according to the preset weight coefficients.

[0014] Preferably, the state assessment module calculates the brain tissue deformability index, including:

[0015] Obtain the coordinates of key anatomical landmarks at the current and previous time points;

[0016] Based on the coordinates of the current time and the previous time, determine the displacement of each key anatomical landmark;

[0017] Based on the displacement and the preset time step, the average rate of displacement of key anatomical landmarks is calculated, and the brain tissue deformation index is generated.

[0018] Preferably, the risk prediction module predicts the future trust decay index, including:

[0019] Obtain the current human-machine trust decay index and brain tissue deformation index;

[0020] Based on the brain tissue deformation index, the preset risk transmission coefficient, and the preset time step, the trust decay increment is determined.

[0021] The trust decay increment is summed with the current human-machine trust decay index to generate the future trust decay index.

[0022] Preferably, the adaptive control module determines the navigation mode control parameters, including:

[0023] The risk ratio is calculated based on the future trust decay index and the preset default threshold.

[0024] The initial control parameters are determined by subtracting the square of the risk ratio from 1;

[0025] Take the larger value between the initial control parameter and 0 to generate the navigation mode control parameter.

[0026] Preferably, when the navigation mode control parameter is greater than a preset first threshold, the adaptive control module switches the navigation mode to the cooperative guidance mode;

[0027] The collaborative guidance mode is characterized by displaying the predicted trajectory point sequence with high visual weight and displaying the real-time positioning point sequence with low visual weight.

[0028] Preferably, when the navigation mode control parameter is less than or equal to the first threshold and greater than 0, the adaptive control module switches the navigation mode to the decision support mode;

[0029] The decision support mode is characterized by reducing the visual weight of the predicted trajectory point sequence and highlighting the navigation deviation vector.

[0030] Preferably, when the navigation mode control parameter is equal to 0, the adaptive control module switches the navigation mode to the safety anchor point mode;

[0031] The safety anchor mode is characterized by removing the guidance information of the predicted trajectory point sequence and providing only a pure positioning view based on the real-time positioning point sequence.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] 1. By quantifying human-machine trust and establishing a risk prediction model, this system can directly link physical interference such as intraoperative brain tissue deformation with the doctor's level of trust in the system, realizing the transformation from passive response deviation to proactive risk prediction, and identifying and managing the risk of human-machine collaboration failure that may be caused by preoperative planning failure in advance.

[0034] 2. This system establishes an adaptive control closed loop based on risk prediction, which can proactively and smoothly adjust the navigation strategy based on the prediction of future risks before the trust between humans and machines is completely broken. It can downgrade from a strong guidance mode to a weak assistance mode or even a pure positioning mode, safely and unambiguously returning the surgical decision-making power to the doctor, thereby effectively avoiding surgical risks caused by unreliable navigation information and improving the safety of the system in complex dynamic environments.

[0035] 3. This system, through the design of a hierarchical navigation mode, provides the most appropriate assistance under different risk levels, offers efficient guidance when it is safe, provides decision support when risks occur, and provides safety anchors when danger arises. This greatly enhances the doctor's sense of trust and control over the system during surgery, realizes the transformation from human to machine to machine to human, and maximizes the efficiency of human-machine collaboration.

[0036] 4. This system can monitor the displacement of key anatomical landmarks in real time to quantify the severity of brain tissue deformation, solving the problem that traditional navigation systems rely too much on static preoperative planning and cannot cope with dynamic changes during surgery. This significantly enhances the robustness of the navigation system in real surgical environments and the reliability of its clinical applications. Attached Figure Description

[0037] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0038] Figure 1 This is a structural diagram of the system of the present invention. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0040] Example 1:

[0041] Please see Figure 1 The intelligent navigation system for medical endoscope operation trajectory includes:

[0042] The data acquisition module is used to acquire the predicted trajectory point sequence, real-time positioning point sequence, actual operation trajectory points, and coordinates of key anatomical landmarks;

[0043] The deviation processing module is used to determine the navigation deviation vector and operation correction vector based on the predicted trajectory point sequence, the real-time positioning point sequence, and the actual operation trajectory points.

[0044] The status assessment module is used to calculate the human-machine trust decay index and brain tissue deformation index based on the navigation deviation vector, operation correction vector, and coordinates of key anatomical landmarks.

[0045] The risk prediction module is used to predict the future trust decay index based on the human-machine trust decay index and the brain tissue deformation index.

[0046] The adaptive control module is used to determine the navigation mode control parameters based on the future trust decay index and the preset trust loss threshold, and adaptively switch the navigation mode according to the navigation mode control parameters.

[0047] This embodiment provides an intelligent navigation system for medical endoscope operation trajectories, aiming to solve the technical problem that preoperative planning fails due to dynamic changes during surgery, leading to doctors' distrust of the system guidance and reduced surgical safety. This system constructs a closed loop of quantitative assessment and risk prediction of human-machine trust, and adaptively adjusts the navigation strategy accordingly, thereby maximizing the efficiency of human-machine collaboration while ensuring surgical safety.

[0048] In this embodiment, the system specifically includes a data acquisition module, a deviation processing module, a state assessment module, a risk prediction module, and an adaptive control module;

[0049] The data acquisition module provides all the raw data inputs required for the system's status assessment and risk prediction; this module operates in a multi-channel parallel manner to acquire four types of core data: predicted trajectory point sequences. Real-time location point sequence Actual operation trajectory points and coordinates of key anatomical landmarks Predicted trajectory point sequence Based on preoperative 3D image data, this system generates a theoretically optimal safe surgical path point 3D coordinate sequence using a pre-trained Generative Adversarial Network (GAN), providing doctors with forward-looking intelligent path guidance. For example, the generator of this GAN can employ an encoder-decoder structure based on a 3D U-Net, taking preoperative 3D image data as input and outputting a 3D path point probability map. The discriminator uses a 3D convolutional neural network to distinguish between real expert surgical paths and generated paths. The training process uses Wasserstein distance as the loss function to enhance training stability. The training dataset contains preoperative images of at least 500 similar surgeries and optimal paths annotated by senior experts. Real-time localization point sequences are also included. It utilizes Simultaneous Localization and Mapping (SLAM) technology, fusing visual odometry information and electromagnetic sensor signals from the endoscope to output a real-time six-degree-of-freedom three-dimensional coordinate sequence of the endoscope's positioning points in the surgical space coordinate system. This sequence accurately reflects the endoscope's true position in physical space. The specific fusion algorithm can employ an extended Kalman filter. The system's state vector... This includes the endoscope's position, orientation, and velocity; in the prediction phase, high-frequency motion estimation provided by visual odometry is used to update the state; in the update phase, low-frequency but drift-free absolute position measurements provided by electromagnetic sensors are used to correct the state vector, thereby achieving robust and accurate positioning; actual operational trajectory points. It captures the doctor's true intentions by tracking the three-dimensional coordinate vectors of the endoscope tip, obtained from sensors on the endoscope's actuator; and the coordinates of key anatomical landmarks. It is the three-dimensional coordinate vector of N key anatomical structures continuously tracked in the endoscopic field of view through image recognition algorithms at the current time t, which provides a basis for quantifying dynamic physical interference during surgery;

[0050] The deviation processing module transforms the raw multi-source navigation data into core vector indicators that can directly characterize the human-computer interaction state; for ease of description, the predicted trajectory points mentioned below will be referred to as such. Real-time location points and actual operation trajectory points Unless otherwise specified, all times refer to the current moment. The specific three-dimensional coordinate vectors obtained from the corresponding point sequence; in this embodiment, the module is based on the predicted trajectory point sequence obtained by the data acquisition module. Real-time location sequence and actual operation trajectory points Real-time determination of navigation deviation vector With operation correction vector Navigation deviation vector From real-time location points Point to the predicted trajectory point The spatial vector, i.e. ; Operation correction vector From the predicted trajectory points Pointing to the doctor's actual operation trajectory point The spatial vector, i.e. Its function is to quantify the proactive interventions made by doctors based on their own professional judgments in response to the system guidance, and directly reflect the doctors' level of trust in the system guidance and their intention to correct their behavior.

[0051] The status assessment module calculates a scalar index that quantifies the current health status of the human-machine collaborative relationship and external risk sources based on the vector index output by the deviation processing module and dynamic interference data. In this embodiment, the module uses the navigation deviation vector... Operational correction vector and coordinates of key anatomical landmarks Calculate the human-machine trust decay index With brain tissue deformability index Human-Machine Trust Decline Index By comprehensively assessing both the system's inherent biases and the doctor's external behaviors, a single scalar index is generated to dynamically reflect the current state of the human-machine collaborative relationship; brain tissue deformability index. By monitoring the displacement rate of key anatomical markers, the severity of physical disturbances caused by factors such as brain tissue deformation can be quantified in real time, providing crucial external driving force data for risk prediction.

[0052] The risk prediction module enables proactive risk management, predicting future deterioration trends before the human-machine trust relationship completely breaks down. In this embodiment, the module assesses the current state, i.e., the human-machine trust decay index. With brain tissue deformability index Using a time-series prediction model, the future trust decay index is predicted. This design goes beyond the traditional passive response model, enabling the system to intervene in advance based on predictions of the future.

[0053] The adaptive control module transforms the predicted abstract risks into concrete, executable system behaviors, enabling intelligent degradation of the navigation strategy. In this embodiment, the module is based on the future trust decay index output by the risk prediction module. With a preset default threshold Determine a navigation mode control parameter Based on this control parameter The specific values ​​are adaptively and smoothly switched between multiple navigation modes, thereby dynamically adjusting the intensity and form of the system's intelligent guidance;

[0054] The system in this embodiment constructs a closed-loop technology system of data acquisition, deviation processing, state assessment, risk prediction, and adaptive control, which quantitatively models and proactively predicts the concept of trust between humans and machines. It can perceive in real time the risk of navigation information failure caused by intraoperative dynamic interference such as brain tissue deformation, and proactively and smoothly downgrade the navigation mode from strong guidance to weak assistance or even pure positioning mode before the doctor develops serious distrust of the system. Ultimately, it safely and unambiguously returns the surgical decision-making power to the doctor. This greatly improves the robustness of the navigation system in complex dynamic environments and the safety of clinical applications, effectively avoids surgical risks caused by human-machine discord, and realizes true human-machine collaborative intelligent navigation.

[0055] Example 2:

[0056] The status assessment module calculates the human-machine trust decay index, including:

[0057] Calculate the navigation deviation magnitude of the navigation deviation vector and the operation correction magnitude of the operation correction vector;

[0058] The first normalized deviation value is determined based on the navigation deviation modulus and the preset first reference deviation modulus.

[0059] The second normalized deviation value is determined based on the operational correction modulus and the preset second reference deviation modulus.

[0060] The human-machine trust decay index is generated by combining the first normalized deviation value and the second normalized deviation value, and by linearly weighting and summing according to the preset weight coefficients.

[0061] This embodiment calculates the human-machine trust decay index using the state assessment module in Embodiment 1. One specific implementation method; this method uses a standardized mathematical model to represent the deviation vectors of different physical dimensions and different sources, i.e., navigation deviations. and operation correction This is integrated into a unified, dimensionless trust metric.

[0062] The calculation process is as follows: calculate the navigation deviation magnitude of the navigation deviation vector. Operational correction modulus of the operation correction vector ; here The Euclidean norm of the pointer is used to obtain the absolute magnitude of the deviation; based on the navigation deviation magnitude. The first reference deviation modulus Determine the first normalized deviation value; first reference deviation modulus. This is a preset constant representing the upper limit of clinically acceptable navigation deviation tolerance. Its value is determined by: statistically analyzing a database of historical successful surgeries for a specific surgical procedure, extracting all navigation deviation data, and taking the 95th percentile as the baseline value for that procedure; then calculating the ratio... The navigation deviation magnitude, which has physical dimensions, is converted into a dimensionless relative deviation value; similarly, the operational correction magnitude is... Deviation modulus from the preset second reference Determine the second normalized deviation value; second reference deviation modulus. The source and determination method are the same as Similarly, all are based on statistical analysis of historical big data; finally, by combining the first and second normalized deviation values ​​and according to preset weighting coefficients, a human-machine trust decay index is generated through linear weighted summation. The calculation formula is as follows: ;

[0063] in, Let be the trust decay exponent at time t; and The weighting coefficients are dimensionless adjustable parameters assigned to navigation deviations and operational corrections, respectively, satisfying the following conditions: Its function is to adjust the contribution of system bias and physician intervention corrections to the final evaluation value. The specific value can be determined through expert scoring or regression analysis based on historical data. Specifically, when using regression analysis, a logistic regression model can be constructed, where the dependent variable is whether physician intervention occurred during the surgery, and the independent variables are the first and second normalized bias values. By fitting historical surgical data, the model coefficients, after normalization, can be used as weighting coefficients. and ;

[0064] Through the above implementation methods, this invention standardizes and normalizes two core deviations of different natures and introduces adjustable weighting coefficients. and This causes the human-machine trust decay index to decline. It is not only an objective quantitative indicator, but also a flexible assessment model that can be customized according to different surgical types. Compared with monitoring various deviations separately, this method provides a single indicator that can comprehensively, dynamically and sensitively reflect the health status of human-machine collaboration, providing highly condensed and effective input for subsequent risk prediction and adaptive control.

[0065] Example 3:

[0066] The state assessment module calculates the brain tissue deformability index, including:

[0067] Obtain the coordinates of key anatomical landmarks at the current and previous time points;

[0068] Based on the coordinates of the current time and the previous time, determine the displacement of each key anatomical landmark;

[0069] Based on the displacement and the preset time step, the average rate of displacement of key anatomical landmarks is calculated, and a brain tissue deformation index is generated.

[0070] This embodiment calculates the brain tissue deformation index using the state assessment module in Embodiment 1. One specific implementation method; this method uses a computationally efficient and physically meaningful approach to quantify in real time the dynamic physical disturbances caused by non-rigid deformations such as intraoperative brain tissue displacement;

[0071] The calculation process is as follows: obtain the value at the current time. Compared to the previous moment The system records the coordinates of key anatomical landmarks; it will also record the three-dimensional coordinate vectors of N landmarks at the current time t. And in The three-dimensional coordinate vector of the previous moment before time ; The time step is a preset system parameter; based on the coordinates of the current time and the previous time, the displacement of each key anatomical landmark is determined, that is, the displacement of each landmark i in time is calculated. Magnitude of the spatial displacement vector within the time window Based on the displacement and the preset time step The average rate of displacement of key anatomical landmarks was calculated, and a brain tissue deformability index was generated. The initial source of this calculation is the discrete approximation of instantaneous velocity in kinematics, which is achieved by averaging the displacements of all N markers and dividing by the time step. This yields a scalar exponent that characterizes the overall deformation velocity within the current field of view; its calculation formula is as follows: ;

[0072] in, Let be the brain tissue deformation index at time t, with velocity as the physical dimension; The total number of key anatomical landmarks continuously tracked by image recognition algorithms within the endoscopic field of view;

[0073] This implementation proposes a non-invasive, vision-tracking-based method for quantifying intraoperative dynamic interference. Compared to methods relying on complex biomechanical models or implanted sensors, this method has lower computational overhead, higher real-time performance, and can directly reflect the deformation of the most critical areas within the endoscopic field of view. It calculates the average rate of displacement of all markers... The index can effectively filter out the influence of tracking noise on individual markers, thereby accurately and robustly quantifying the core physical root cause of preoperative planning failure, namely the current severity of brain tissue deformation, providing crucial input characterizing external risk sources for subsequent prediction of trust risk evolution.

[0074] Example 4:

[0075] The risk prediction module predicts the future trust decay index, including:

[0076] Obtain the current human-machine trust decay index and brain tissue deformation index;

[0077] Based on the brain tissue deformation index, the preset risk transmission coefficient, and the preset time step, the trust decay increment is determined.

[0078] The trust decay increment is summed with the current human-machine trust decay index to generate the future trust decay index.

[0079] This embodiment predicts the future trust decay index using the risk prediction module in Embodiment 1. One specific implementation method; its underlying logic lies in establishing a connection between physical world interference caused by Representation and human-machine collaborative state by The time-series prediction model represents the system's risk management, thereby upgrading it from passive response to proactive prediction.

[0080] The prediction process involves obtaining the current moment, which is calculated in real time by the state assessment module. Human-Machine Trust Decline Index With brain tissue deformability index Based on brain tissue deformability index Preset risk transmission coefficient and preset time steps The step involves determining the increment of trust decay; the theoretical basis for this step is the assumption that the rate of change in trust decay is proportional to the rate of brain tissue deformation; risk transmission coefficient. It is a system parameter that quantifies the change in the trust index caused by a unit deformation rate per unit time; to ensure that the dimensions on both sides of the formula are consistent, Its physical dimension is the reciprocal of length, and its specific numerical value is determined by: performing regression analysis on historical surgical data and fitting... and The relationship between the rates of change was obtained; specifically, a series of data points were extracted from historical surgical data. The least squares method was used to perform linear regression analysis on these data points, and the slope of the resulting straight line is the risk transmission coefficient. As a more refined implementation method, and to more accurately characterize the nonlinear psychophysical effects in human-computer interaction, the risk transmission coefficient... It can be modeled as a function related to the current trust state, rather than being treated as a constant system parameter. For example, a monotonically increasing function form, such as an exponential function, can be designed. ,in It is the basic risk transmission coefficient. It is a positive regulating factor used to control the severity of nonlinear growth; and Calibration can be achieved by performing nonlinear least-squares fitting on historical data; for example, designing a random-sense least-squares fitting algorithm... An increasing and monotonically increasing function is used to characterize how, when the trust foundation is already relatively weak, the same degree of physical interference will cause a more severe decline in trust, thus making the risk prediction model more faithful; the increment of trust decline is... The increment of trust decay is compared with the current human-machine trust decay index. Sum to generate the future trust decay index. The initial source for this calculation is the Euler forward difference method, used in physics and control theory to describe the evolution of system states over time; its calculation formula is as follows: ;

[0081] This implementation establishes a risk evolution prediction model from physical interference to human-machine trust; by introducing a risk transmission coefficient... It will observable physical phenomena such as the rate of brain tissue deformation. Trust decay index, an abstract measure of human-computer relationship Direct correlation; this enables the system to no longer merely respond to deviations that have already occurred, but to proactively predict the extent to which human-machine trust will deteriorate based on the severity of current brain tissue deformation; this predictive ability is the logical prerequisite for achieving true adaptive control, greatly enhancing the system's initiative and safety in dealing with emergencies.

[0082] Example 5:

[0083] The adaptive control module determines the navigation mode control parameters, including:

[0084] The risk ratio is calculated based on the future trust decay index and the preset default threshold.

[0085] The initial control parameters are determined by subtracting the square of the risk ratio from 1;

[0086] Take the larger value between the initial control parameter and 0 to generate the navigation mode control parameter.

[0087] This embodiment describes how the adaptive control module in Embodiment 1 determines navigation mode control parameters. One specific implementation method; in order to achieve nonlinear control that keeps the system stable when the predicted risk is low, and can quickly and decisively reduce the guiding weight when the risk approaches the critical point, the following control law is designed in this embodiment;

[0088] The established process is based on the future trust decay index output by the risk prediction module. Compared with the preset default threshold Calculate the risk ratio; default threshold. The underlying logic is to perform offline analysis on a large amount of screen recording data from simulated surgeries, and to statistically analyze the moments of human-machine incoordination marked by expert physicians before critical operational errors occur. The value, thus defining a critical value; risk ratio. This reflects how close the predicted future risk is to the default threshold; the initial control parameters are determined by subtracting the square of the risk ratio from 1, where the square term... This is used to achieve nonlinear control effects; finally, the larger value between the initial control parameters and 0 is taken to generate the navigation mode control parameters. This ensures that its value range is strictly limited to the [0,1] interval;

[0089] In summary, the complete calculation formula for the navigation mode control parameters is as follows: ;

[0090] The nonlinear control law designed in this embodiment makes the adaptive adjustment behavior of the navigation system more intelligent. When the risk is low, it can maintain high-intensity intelligent guidance; when the risk gradually increases, it will smoothly transition; and when it is predicted that the risk is about to reach the threshold, it can decisively return control to the doctor. This risk prediction-based, nonlinear, forward-looking adjustment mechanism balances the efficiency of intelligent guidance with the safety in extreme cases, and is significantly better than the traditional control method that makes linear or step-like adjustments based only on the current deviation.

[0091] Example 6:

[0092] When the navigation mode control parameter is greater than the preset first threshold, the adaptive control module will switch the navigation mode to the cooperative guidance mode.

[0093] The collaborative guidance mode is characterized by displaying the predicted trajectory point sequence with high visual weight and displaying the real-time positioning point sequence with low visual weight.

[0094] When the navigation mode control parameter is less than or equal to the first threshold and greater than 0, the adaptive control module will switch the navigation mode to the decision support mode.

[0095] The decision support mode is characterized by reducing the visual weight of the predicted trajectory point sequence and highlighting the navigation deviation vector.

[0096] When the navigation mode control parameter is equal to 0, the adaptive control module will switch the navigation mode to the safety anchor point mode;

[0097] The safety anchor mode is characterized by removing the guidance information of the predicted trajectory point sequence and providing only a pure positioning view based on the real-time positioning point sequence.

[0098] This embodiment is a specific implementation of the hierarchical adaptive switching of navigation strategy in Embodiment 5; the system controls the parameters according to the navigation mode that takes effect in real time. The value automatically switches between collaborative guidance mode, decision support mode, and safety anchor mode, forming a smooth degradation spectrum from close collaboration to complete autonomy. In this embodiment, the first threshold for mode switching is set to 0.5. The determination of this threshold is based on a simulated surgical user experience study of 10 senior neurosurgeons. The results show that when the threshold is set to 0.5, the system can achieve the best balance between providing effective warnings and avoiding excessive interference, which best matches the clinical decision-making habits of doctors.

[0099] When navigation mode control parameters Greater than the first threshold When the adaptive control module switches the navigation mode to the cooperative guidance mode, the system judges that the predicted future risk is low. The navigation interface displays the predicted trajectory point sequence with high visual weight and overlays the real-time positioning point sequence with low visual weight.

[0100] When navigation mode control parameters Less than or equal to the first threshold and greater than 0 At this point, the adaptive control module switches the navigation mode to decision support mode; in this mode, the system determines that the predicted future risk has entered the observation zone; the system reduces the visual weight of the predicted trajectory point sequence and highlights the navigation deviation vector. Actively warn doctors of discrepancies between predictions and reality;

[0101] When navigation mode control parameters When the value equals 0, the adaptive control module switches the navigation mode to the safety anchor point mode; in this mode, the system determines that the predicted future risk has exceeded or is about to exceed the default threshold. The system removes guidance information from the predicted trajectory point sequence and only provides a pure positioning view based on the real-time positioning point sequence, downgrading the system function to an absolutely honest positioning tool and clearly and unambiguously returning all decision-making power to the doctor.

[0102] Through these three progressively degrading navigation modes, this invention establishes a complete adaptive human-computer interaction strategy based on quantitative risk prediction; it abstracts control parameters. This translates into a concrete, intuitive interface that aligns with the doctor's operational intuition. This tiered correction strategy ensures that the system can provide the most appropriate assistance at different risk levels: providing efficient guidance when it is safe, providing decision support when risks arise, and providing safety anchors when danger occurs. This intelligent and smooth mode switching mechanism greatly enhances the doctor's trust and sense of control over the system throughout the entire surgical process, realizing the transformation from human adaptation to machine to machine adaptation to human.

[0103] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A medical endoscope operation trajectory intelligent navigation system, characterized in that, include: The data acquisition module is used to acquire the predicted trajectory point sequence, real-time positioning point sequence, actual operation trajectory points, and coordinates of key anatomical landmarks; The deviation processing module is used to determine the navigation deviation vector and operation correction vector based on the predicted trajectory point sequence, the real-time positioning point sequence, and the actual operation trajectory points. The status assessment module is used to calculate the human-machine trust decay index based on the navigation deviation vector and the operation correction vector, and to calculate the brain tissue deformation index based on the coordinates of key anatomical landmarks. The risk prediction module is used to predict the future trust decay index based on the human-machine trust decay index and the brain tissue deformation index. The adaptive control module is used to determine the navigation mode control parameters based on the future trust decay index and the preset trust loss threshold, and adaptively switch the navigation mode according to the navigation mode control parameters. The status assessment module calculates the human-machine trust decay index, including: Calculate the navigation deviation magnitude of the navigation deviation vector and the operation correction magnitude of the operation correction vector; The first normalized deviation value is determined based on the navigation deviation modulus and the preset first reference deviation modulus. The second normalized deviation value is determined based on the operational correction modulus and the preset second reference deviation modulus. The human-machine trust decay index is generated by combining the first normalized deviation value and the second normalized deviation value, and by linearly weighting and summing according to the preset weight coefficients. The state assessment module calculates the brain tissue deformability index, including: Obtain the coordinates of key anatomical landmarks at the current and previous time points; Based on the coordinates of the current time and the previous time, determine the displacement of each key anatomical landmark; Based on the displacement and the preset time step, the average rate of displacement of key anatomical landmarks is calculated, and a brain tissue deformation index is generated.

2. The intelligent navigation system for medical endoscope operation trajectory according to claim 1, characterized in that, The risk prediction module predicts the future trust decay index, including: Obtain the current human-machine trust decay index and brain tissue deformation index; Based on the brain tissue deformation index, the preset risk transmission coefficient, and the preset time step, the trust decay increment is determined. The trust decay increment is summed with the current human-machine trust decay index to generate the future trust decay index.

3. The intelligent navigation system for medical endoscope operation trajectory according to claim 1, characterized in that, The adaptive control module determines the navigation mode control parameters, including: The risk ratio is calculated based on the future trust decay index and the preset default threshold. The initial control parameters are determined by subtracting the square of the risk ratio from 1; Take the larger value between the initial control parameter and 0 to generate the navigation mode control parameter.

4. The intelligent navigation system for medical endoscope operation trajectory according to claim 3, characterized in that, When the navigation mode control parameter is greater than the preset first threshold, the adaptive control module will switch the navigation mode to the cooperative guidance mode. The collaborative guidance mode is characterized by displaying the predicted trajectory point sequence with high visual weight and displaying the real-time positioning point sequence with low visual weight.

5. The intelligent navigation system for medical endoscope operation trajectory according to claim 3, characterized in that, When the navigation mode control parameter is less than or equal to the first threshold and greater than 0, the adaptive control module will switch the navigation mode to the decision support mode. The decision support mode is characterized by reducing the visual weight of the predicted trajectory point sequence and highlighting the navigation deviation vector.

6. The intelligent navigation system for medical endoscope operation trajectory according to claim 3, characterized in that, When the navigation mode control parameter is equal to 0, the adaptive control module will switch the navigation mode to the safety anchor point mode; The safety anchor mode is characterized by removing the guidance information of the predicted trajectory point sequence and providing only a pure positioning view based on the real-time positioning point sequence.

Images