Minimally invasive surgery system based on artificial intelligence

Abstract

The invention relates to the technical field of medical instruments, and discloses a minimally invasive surgery system based on artificial intelligence. Comprising a target identification and positioning module, a multi-mode sensing module, a self-adaptive control module, a man-machine interaction module and a scene adaptation module. The target recognition and positioning module positions an operation target in real time through image recognition and a deep learning algorithm; the multi-mode sensing module collects the stress and movement track of an instrument and physiological signals of a patient. The self-adaptive control module adjusts movement parameters of the instrument according to the data; the man-machine interaction module displays data and provides an operation interface; the scene adaptation module adjusts the instrument form and function according to the operation type. In addition, the system also comprises an operational stability evaluation module, an instrument life prediction module and a biological signal fusion module. The system can achieve precise positioning, real-time monitoring and intelligent control, and the precision, safety and operation convenience of minimally invasive surgery are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical devices, and in particular to a minimally invasive surgery system based on artificial intelligence. Background Art

[0002] Minimally invasive surgery has been widely used in modern medicine due to its advantages of minimal trauma and rapid recovery. However, current minimally invasive surgery technology still faces many challenges, which limit its further development and improvement of clinical effectiveness.

[0003] When it comes to surgical target positioning, existing positioning technologies lack accuracy. Traditional methods rely primarily on the doctor's experience and preoperative imaging data. However, during surgery, the position of the patient's organs will move due to physiological activities such as breathing and heartbeat. Preoperative imaging is unable to reflect these changes in real time, resulting in deviations in the positioning of surgical instruments. For example, during liver surgery, the position and shape of the liver will shift significantly with breathing. If the lesion site cannot be accurately located, surgical errors may occur, damaging surrounding normal tissues, affecting surgical outcomes, and increasing the risk of postoperative complications for patients. Moreover, traditional positioning technology lacks a real-time dynamic adjustment mechanism. Once positioning deviates, it is difficult to correct it in a timely manner. Doctors can only make rough adjustments based on their experience, which places extremely high demands on the doctor's technical level and increases the uncertainty of the surgery.

[0004] There are also defects in the operational control of surgical instruments. On the one hand, existing surgical instruments are unable to fully perceive various information during the operation. For example, insufficient force feedback makes it difficult for doctors to accurately grasp the force between the instrument and the tissue. Excessive force may damage the tissue, and too little force may make the operation impossible to complete. When performing delicate vascular suture surgery, if the doctor cannot accurately perceive the force between the suture needle and the vascular tissue, it is easy to cause the blood vessel to tear or the suture to be loose, affecting the success rate of the operation. On the other hand, the motion control of the instrument is not flexible and precise enough, and the motion trajectory is difficult to strictly follow the preset path. The operational stability is poor, affecting the precision and quality of the surgery.

[0005] The ability to fuse and process multimodal information is weak. During surgery, a large amount of different types of information is generated, such as imaging, physiological signals, and instrument motion data. However, existing surgical systems struggle to efficiently fuse and analyze this information. This prevents doctors from fully understanding the surgical status and making accurate decisions. For example, during complex tumor resection surgery, doctors need to simultaneously monitor the tumor's location, the physiological state of surrounding tissues, and the instrument's operation. Due to the lack of an effective information fusion mechanism, doctors may not be able to detect potential risks in a timely manner, delaying surgery.

[0006] Furthermore, the function and form of surgical instruments cannot be flexibly adapted to different surgical scenarios. Different surgical types require significantly different instruments, and existing general-purpose instruments struggle to meet diverse surgical needs. Ophthalmic and orthopedic surgeries require vastly different instrument precision, length, and bending angles. Using single-function instruments limits the operating space and reduces surgical effectiveness. Summary of the Invention

[0007] The purpose of the present invention is to provide a minimally invasive surgery system based on artificial intelligence to solve the problems raised in the above background technology.

[0008] To achieve the above objectives, the present invention provides the following technical solutions: an artificial intelligence-based minimally invasive surgery system, the system comprising a target recognition and positioning module, a multimodal sensing module, an adaptive control module, a human-computer interaction module, and a scene adaptation module;

[0009] The target recognition and positioning module uses image recognition and deep learning algorithms to perform real-time recognition and three-dimensional positioning of the surgical target area, generates target coordinate data, and sends it to the adaptive control module;

[0010] The multimodal sensing module integrates force feedback sensors, position tracking sensors, and bioelectric signal sensors to collect real-time force data, motion trajectory data, and patient physiological signal data of surgical instruments;

[0011] The adaptive control module adjusts the motion parameters of the surgical instrument according to the target coordinate data and the real-time sensor data, and generates a motion trajectory deviation signal if it is detected that the deviation value between the motion trajectory data and the preset path exceeds a first threshold value;

[0012] The human-computer interaction module displays surgical parameters, sensor data, and deviation signals, and provides a touch operation interface and voice command interface;

[0013] The scene adaptation module dynamically adjusts the instrument shape and functional mode according to the type of surgery, including bending angle adjustment, telescopic length control and tool head switching.

[0014] Preferably, the specific operation process of the target identification and positioning module is as follows:

[0015] A three-dimensional model of the patient's anatomical structure is constructed through preoperative medical images, and an endoscopic video stream is obtained in real time during the operation. Each frame of the video stream is aligned with the three-dimensional model, and the real-time spatial coordinates of the target area are calculated; the alignment error value and coordinate offset are collected. If the alignment error value exceeds the second threshold or the coordinate offset exceeds the third threshold, a positioning abnormality signal is generated; the positioning abnormality signal is sent to the human-computer interaction module for warning, and the adaptive control module is triggered to suspend the movement of the instrument until the abnormality is resolved.

[0016] Preferably, the data processing process of the multimodal sensing module includes:

[0017] The raw data of the force feedback sensor is converted into the force vector at the end of the instrument, and the difference between it and the preset safety force threshold is calculated and marked as the force segregation value; the raw data of the position tracking sensor is fitted with the preset motion path, and the trajectory curvature change rate and speed fluctuation value are extracted, and marked as the curvature analysis value and the fluctuation analysis value respectively; if the force segregation value, curvature analysis value or fluctuation analysis value exceeds the corresponding fourth threshold, an abnormal instrument operation signal is generated, and the abnormal signal type and level are sent to the adaptive control module.

[0018] Preferably, the control logic of the adaptive control module is as follows:

[0019] When a motion trajectory deviation signal is received, a compensatory motion instruction is generated based on the reinforcement learning algorithm to adjust the velocity vector and direction angle of the end of the device; during the compensation process, the decay rate of the trajectory deviation value is continuously monitored. If the decay rate is lower than the fifth threshold, it switches to the preset safety mode, limits the maximum movement speed of the device and narrows the operating range; if the deviation value returns to zero within the preset time, the current control is recorded as a valid control event.

[0020] Preferably, the system also includes an operational stability evaluation module, which collects the total number of control events and the proportion of effective control events generated by the adaptive control module, compares them with historical data to calculate a stability index; if the stability index is lower than a sixth threshold, an instrument control degradation signal is generated, and the scene adaptation module is triggered to switch to a redundant control mode and enable the backup drive unit.

[0021] Preferably, the operation stability evaluation module further analyzes the time distribution characteristics of the control events, extracts high-frequency abnormal periods and low-frequency abnormal periods, and marks the duration ratio of the high-frequency abnormal period as the period abnormal frequency value; if the period abnormal frequency value exceeds the seventh threshold, an operating environment interference signal is generated and pushed to the human-computer interaction module to prompt the doctor to check the connection status of the external device.

[0022] Preferably, the function switching process of the scene adaptation module includes:

[0023] The preset morphology database is called according to the type of surgery to match the bending angle parameters, telescopic length parameters and tool head type of the instrument; during the switching process, the response delay value and the in-place error value of each joint are collected. If the response delay value exceeds the eighth threshold or the in-place error value exceeds the ninth threshold, a morphology adaptation abnormality signal is generated, and the adaptive control module is triggered to perform a reverse calibration action.

[0024] Preferably, the system also includes an instrument life prediction module, which collects the number of form switching times, joint wear coefficient and drive unit load peak, and obtains the instrument health score through weighted calculation; if the health score is lower than the tenth threshold, an instrument maintenance warning signal is generated, and the scene adaptation module is prohibited from performing high-risk form switching operations.

[0025] Preferably, it also includes a biological signal fusion module, which performs noise reduction and feature extraction on the physiological signals collected by the multimodal sensing module, and identifies the patient's heart rate variability characteristics and muscle tension index; if the heart rate variability characteristics exceed the eleventh threshold or the muscle tension index exceeds the twelfth threshold, a patient stress signal is generated, and the adaptive control module is triggered to reduce the movement speed of the device to a safe level.

[0026] Preferably, the biosignal fusion module further analyzes the trigger frequency and duration of the stress signal, and marks the number of triggers per unit time as the stress frequency value; if the stress frequency value exceeds the thirteenth threshold, a surgery suspension recommendation signal is generated and pushed to the human-computer interaction module to request the doctor to confirm the subsequent operation strategy.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] In terms of precise positioning, the target recognition and positioning module constructs a three-dimensional model before surgery and combines it with intraoperative endoscopic video stream registration to obtain the precise coordinates of the surgical target area in real time. This process not only utilizes deep learning algorithms to improve recognition accuracy but also monitors registration errors and coordinate offsets in real time. If an anomaly is detected, instrument movement is immediately paused and an alert is issued, preventing surgical errors caused by positioning deviations. For example, in kidney surgery, precise positioning allows surgical instruments to accurately reach the lesion site, minimizing damage to surrounding healthy tissue, reducing surgical risks, and improving surgical success rates, while also shortening the patient's postoperative recovery time and pain.

[0029] The multimodal sensing module integrates multiple sensors to comprehensively collect data on the force and motion trajectory of surgical instruments, as well as the patient's physiological signals. By converting the raw data from the force feedback sensor into a force vector and comparing it with a safety threshold, as well as performing trajectory fitting analysis on the position tracking sensor data, abnormal instrument operation can be detected promptly. For example, during a tissue cutting operation, if the force segregation value exceeds the threshold, indicating that the instrument may encounter abnormal resistance, the system will promptly issue a signal to prevent the instrument from excessively applying force and damaging the tissue. At the same time, monitoring of the patient's physiological signals, such as heart rate variability characteristics and muscle tension indicators, can reflect the patient's physical condition in real time. If patient stress is detected, the adaptive control module can reduce the instrument's movement speed to ensure patient safety.

[0030] Based on a reinforcement learning algorithm, the adaptive control module generates precise compensatory motion commands when deviations occur in the instrument's trajectory. By continuously monitoring the decay rate of trajectory deviations, it promptly adjusts the control strategy to ensure the instrument accurately returns to the preset path. During complex brain surgeries, this module effectively addresses minor deviations in instrument movement, ensuring that surgical instruments remain within a safe and precise operating range. If the deviation returns to zero within a preset time, it is recorded as a valid control event and used to optimize the control algorithm, further improving the system's operational stability and accuracy.

[0031] The human-computer interaction module provides an intuitive display interface and convenient operation interface. Doctors can view surgical parameters, sensor data, and deviation signals in real time, providing a comprehensive understanding of the surgical progress. The touch-screen interface and voice command interface allow doctors to conveniently control the surgical system, improving operational convenience and efficiency. In emergency situations, doctors can quickly adjust instrument parameters through voice commands, saving time and avoiding the risks associated with untimely manual operation.

[0032] The scenario adaptation module dynamically adjusts the instrument's form and functional mode based on the type of surgery. It calls upon a database of preset forms to precisely adjust the instrument's bending angle, telescopic length, and tool head. During laparoscopic surgery, the instrument's form can be rapidly adjusted based on the depth and angle requirements of the surgical site, improving the flexibility and adaptability of the surgical procedure. Furthermore, the module monitors joint response delays and positioning errors during the switching process. If any anomalies are detected, the adaptive control module is triggered to perform reverse calibration to ensure the accuracy and reliability of instrument form adjustments.

[0033] Furthermore, the Operational Stability Assessment Module calculates a stability index by collecting control event data, promptly identifying any degradation in instrument control and switching to redundant control mode to activate the backup drive unit, ensuring surgical continuity and safety. The Instrument Life Prediction Module calculates the instrument's health score based on the number of configuration switches, joint wear coefficient, and peak drive unit load. When the instrument's health score is low, it prohibits high-risk configuration switches and issues maintenance alerts, extending the instrument's lifespan and reducing medical costs. The Biosignal Fusion Module, through in-depth analysis of patient physiological signals, not only reduces instrument speed when the patient is stressed but also recommends pausing the procedure based on the stress frequency, providing an additional layer of patient safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 This is a diagram showing the working principle of the minimally invasive surgery system based on artificial intelligence according to the present invention;

[0035] Figure 2 A flowchart of the operation stability assessment and redundant control triggering;

[0036] Figure 3Workflow diagram for interference signal generation in the operating environment;

[0037] Figure 4 Flowchart for patient stress signal processing. DETAILED DESCRIPTION

[0038] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0039] See also Figure 1-4 The present invention provides a minimally invasive surgery system based on artificial intelligence, and its overall implementation scheme is as follows:

[0040] The target recognition and positioning module uses image recognition and deep learning algorithms to perform real-time recognition and three-dimensional positioning of the surgical target area. Its specific operation process is: before the operation, by obtaining the patient's medical images (such as CT, MRI, etc.), a three-dimensional model of the patient's anatomical structure is constructed. During the operation, the endoscopic video stream is obtained in real time, and each frame image in the video stream is aligned with the three-dimensional model, and then the real-time spatial coordinates of the target area are calculated. At the same time, the registration error value and the coordinate offset are collected. Once the registration error value exceeds the preset second threshold or the coordinate offset exceeds the third threshold, a positioning abnormality signal is generated. The signal will be sent to the human-computer interaction module for warning, and at the same time, the adaptive control module will be triggered to suspend the movement of the instrument until the abnormality is resolved. Finally, the generated target coordinate data will be sent to the adaptive control module to provide a basis for the precise positioning of subsequent surgical instruments.

[0041] The multimodal sensing module integrates a force feedback sensor, a position tracking sensor, and a bioelectric signal sensor. The force feedback sensor collects the force data of the surgical instrument in real time. Its raw data will be converted into the force vector of the instrument end, and the difference between the vector and the preset safety force threshold will be calculated and marked as the force segregation value. The position tracking sensor collects the motion trajectory data of the surgical instrument. Its raw data is fitted with the preset motion path, and the trajectory curvature change rate and speed fluctuation value are extracted, which are marked as the curvature analysis value and the fluctuation analysis value respectively. The bioelectric signal sensor collects the patient's physiological signal data. If the force segregation value, curvature analysis value, or fluctuation analysis value exceeds the corresponding fourth threshold, an abnormal instrument operation signal is generated, and the abnormal signal type and level are sent to the adaptive control module.

[0042] The adaptive control module receives the target coordinate data sent by the target recognition and positioning module and the real-time sensing data sent by the multimodal sensing module, and adjusts the motion parameters of the surgical instrument accordingly. When it is detected that the deviation value of the motion trajectory data from the preset path exceeds the first threshold, a motion trajectory deviation signal is generated. Upon receiving this signal, a compensation motion instruction is generated based on the reinforcement learning algorithm to adjust the velocity vector and direction angle of the end of the instrument. During the compensation process, the attenuation rate of the trajectory deviation value is continuously monitored. If the attenuation rate is lower than the fifth threshold, it switches to the preset safety mode, limits the maximum movement speed of the instrument and narrows the operating range. If the deviation value returns to zero within the preset time length, this regulation is recorded as a valid regulation event.

[0043] The human-computer interaction module is primarily responsible for displaying surgical parameters, sensor data, and deviation signals, providing doctors with intuitive surgical information. It also provides a touch interface and voice command interface, facilitating doctors' operation of the surgical system and achieving efficient human-computer interaction.

[0044] The scenario adaptation module dynamically adjusts the instrument's morphology and functional mode based on the type of surgery. This involves calling corresponding bending angle parameters, telescopic length parameters, and tool head types from a preset morphology database to implement bending angle adjustment, telescopic length control, and tool head switching. During the switching process, the response delay and in-place error values of each joint are collected. If the response delay exceeds the eighth threshold or the in-place error exceeds the ninth threshold, a morphology adaptation anomaly signal is generated, triggering the adaptive control module to perform a reverse calibration.

[0045] The implementation of the present invention will be further described below with reference to Examples 1 to 6.

[0046] Example 1:

[0047] The target recognition and positioning module is the basis for the precise operation of the minimally invasive surgical system, and its accuracy directly affects the surgical effect and safety.

[0048] Before surgery begins, the medical team obtains detailed medical imaging data from the patient. This data is analyzed and processed using specialized medical image processing software. Using a 3D reconstruction algorithm, the 2D imaging data is converted into a 3D model of the patient's anatomy, accurately representing the position, shape, and spatial relationships of each organ and tissue within the body.

[0049] During surgery, the endoscope captures a real-time video stream of the surgical area. Each frame in the video stream is transmitted to the target recognition and positioning module. The module's image registration algorithm begins to operate, based on feature point matching and spatial transformation. First, feature points are extracted from the endoscopic image and the 3D model, such as vascular bifurcations and organ edge features. A specific algorithm then calculates the spatial transformation relationship between these feature points to achieve registration between the image and the 3D model. During the registration process, the real-time spatial coordinates of the target area are continuously calculated.

[0050] In order to ensure the accuracy of positioning, the system collects the registration error value and coordinate offset in real time. The registration error value is obtained by calculating the deviation between the actual registration feature point and the ideal registration position, and the coordinate offset reflects the difference between the actual coordinates of the target area and the preset coordinates. Once the registration error value exceeds the preset second threshold or the coordinate offset exceeds the third threshold, it indicates that the positioning is abnormal. At this time, the module generates a positioning abnormality signal. On the one hand, this signal is sent to the human-computer interaction module, and the doctor is reminded of the positioning problem with a striking color and warning mark on the display screen; on the other hand, it triggers the adaptive control module to suspend the movement of the surgical instrument to avoid misoperation of the instrument due to inaccurate positioning, which may cause harm to the patient. The surgical instrument will not continue to move until the abnormal problem is resolved, such as readjusting the endoscope position or optimizing the registration algorithm parameters, and the positioning returns to normal.

[0051] Example 2:

[0052] The force feedback sensors in the multimodal sensing module use high-precision strain gauge sensors and are distributed at key force-bearing locations on surgical instruments, such as the clamping area at the end of the instrument and joint connections. When the surgical instrument comes into contact with tissue, the force feedback sensor generates corresponding electrical signals, which serve as the raw data. The raw data is amplified and filtered by the signal conditioning circuit, and then converted into a force vector at the end of the instrument based on the sensor's calibration parameters. This vector contains information about the magnitude and direction of the force. The difference between this force vector and the preset safety force threshold is then calculated to obtain the force segregation value. The preset safety force threshold is determined based on a large amount of clinical experimental data and the mechanical properties of the surgical instrument. It represents the maximum safe force that the instrument can withstand during normal surgical operations.

[0053] Position tracking sensors typically use optical tracking technology or electromagnetic tracking technology. Taking optical tracking technology as an example, reflective markers are installed on the surgical instrument, and multiple cameras are used to capture the markers from different angles. The raw data collected by the position tracking sensor is the image coordinate information of each marker at different times. Using the principles of triangulation and spatial geometric relationships, these image coordinates are converted into actual spatial coordinates to obtain the motion trajectory of the surgical instrument. This motion trajectory is then fitted to the preset motion path, with the least squares method being a common fitting method. The parametric equation of the trajectory is obtained through fitting, and the trajectory curvature change rate and velocity fluctuation value are then extracted, labeled as the curvature analysis value and fluctuation analysis value, respectively. The trajectory curvature change rate reflects the change in the curvature of the instrument's motion trajectory, while the velocity fluctuation value reflects the stability of the instrument's motion speed.

[0054] If the force segregation value, curvature segregation value, or fluctuation segregation value exceeds their respective fourth thresholds, it indicates that the surgical instrument is operating abnormally. For example, a force segregation value that is too large may mean that the instrument is experiencing excessive resistance, possibly clamping important tissue or causing the instrument to become stuck; a curvature segregation value that is too large indicates that the instrument's motion trajectory is too tortuous and does not meet normal surgical operation requirements; and a fluctuation segregation value that is too large indicates that the instrument's motion speed is unstable. At this point, the multimodal sensing module generates an instrument operation abnormality signal and sends the abnormal signal type (such as force abnormality, trajectory abnormality, etc.) and level (based on the degree of exceeding the threshold, such as mild, moderate, or severe) to the adaptive control module so that appropriate measures can be taken in a timely manner.

[0055] Example 3:

[0056] After receiving target coordinate data from the target recognition and positioning module and real-time sensor data from the multimodal sensing module, the adaptive control module continuously adjusts the motion parameters of the surgical instrument. If the deviation between the detected motion trajectory data and the preset path exceeds a first threshold, indicating that the surgical instrument's motion has deviated from the intended trajectory, the adaptive control module generates compensatory motion instructions based on a reinforcement learning algorithm.

[0057] The core concept of reinforcement learning algorithms is to continuously learn optimal behavioral strategies through the interaction between an intelligent agent (in this system, the adaptive control module) and the environment (the surgical instrument's motion environment). In this system, the adaptive control module takes the current state of the surgical instrument (such as position, velocity, and force) as input and, through the reinforcement learning algorithm's neural network model, outputs instructions for adjusting the velocity vector and azimuth angle of the instrument's end. This process is like allowing the control module to learn through continuous trial and error how to better adjust the instrument's motion to achieve the target path.

[0058] During the compensation process, the adaptive control module continuously monitors the decay rate of the trajectory deviation value. The decay rate of the trajectory deviation value is obtained by calculating the ratio of the change in the trajectory deviation value at adjacent moments to the time interval. If the decay rate is lower than the fifth threshold, it means that the current compensation strategy is not effective, which may cause the surgical instrument to deviate from the target path for a long time, posing a greater risk. At this time, the adaptive control module switches to the preset safety mode, limiting the maximum movement speed of the instrument and narrowing the operating range. For example, the maximum movement speed of the instrument is reduced to half of the normal speed, and the operating range is reduced to one-third of the original speed to reduce harm to the patient caused by improper movement of the instrument.

[0059] If the deviation returns to zero within a preset time, the control is successful, and the adaptive control module records this control as a valid control event. Data from these valid control events is stored for subsequent evaluation of system performance and optimization of the control algorithm. By continuously accumulating data from valid control events, the reinforcement learning algorithm can learn more optimal control strategies and improve overall system performance.

[0060] Example 4:

[0061] The operation stability assessment module can not only evaluate the stability of instrument control, but also analyze the impact of the operating environment on the operation, providing multi-faceted guarantees for the smooth progress of the operation.

[0062] The operational stability evaluation module collects the total number of control events and the proportion of effective control events generated by the adaptive control module. The total number of control events reflects how frequently the system adjusts the movement of the instrument during surgery, and the proportion of effective control events reflects the effectiveness of these adjustments. The collected data is compared with the historical data to calculate the stability index. The stability index can be calculated using a weighted average method. For example, the weight of the proportion of effective control events is set to 0.7, and the weight of the total number of control events is set to 0.3. In this way, by comprehensively considering the effectiveness and frequency of control, an indicator that can accurately reflect the operational stability of the system is obtained.

[0063] If the stability index is lower than the sixth threshold, it indicates that the instrument control has deteriorated, which may affect the accuracy and safety of the operation. At this time, the operation stability assessment module generates an instrument control degradation signal and triggers the scene adaptation module to switch to redundant control mode. In redundant control mode, the backup drive unit is enabled. The backup drive unit can take over the work of the main drive unit when it fails or its performance degrades, ensuring that the surgical instrument can continue to move normally. For example, in some complex minimally invasive surgeries, the movement of the instrument requires high-precision drive control. If the main drive unit has problems such as unstable motor speed, the backup drive unit can intervene in time to ensure that the instrument moves according to the preset path.

[0064] The operation stability evaluation module further analyzes the time distribution characteristics of the control events. By recording and analyzing the time when the control events occur, high-frequency abnormal periods and low-frequency abnormal periods are extracted. High-frequency abnormal periods refer to time periods in which control events frequently occur within a period of time, while low-frequency abnormal periods are the opposite. The duration ratio of the high-frequency abnormal period is marked as the period heterogeneity value. If the period heterogeneity value exceeds the seventh threshold, it indicates that there may be interference factors in the surgical operation environment, which affects the normal control of surgical instruments. For example, electromagnetic interference in the operating room, signal conflicts between devices, etc. may cause this situation. At this time, the operation stability evaluation module generates an operating environment interference signal and pushes it to the human-computer interaction module to prompt the doctor to check the connection status of the external device and troubleshoot the interference source to ensure the smooth progress of the operation.

[0065] Example 5:

[0066] This embodiment describes the function switching process of the scenario adaptation module and the impact of the instrument life prediction module on it. These two modules work together to ensure that surgical instruments can operate safely and reliably in different surgical scenarios.

[0067] The scenario adaptation module accesses a database of preset morphologies based on the type of surgery. This database stores information such as instrument bending angle parameters, telescopic length parameters, and tool head types for different surgical procedures. For example, in a cholecystectomy, a specific instrument bending angle is required to better reach the gallbladder, while a suitable tool head is required for tissue separation and cutting. In a heart bypass surgery, the instrument's telescopic length and bending angle must be adjusted based on the position of the heart and the needs of the surgical procedure.

[0068] In the process of switching the morphology and functional mode of the device, the scene adaptation module collects the response delay value and the in-place error value of each joint. The response delay value reflects the time difference from the issuance of the switching command to the actual start of the joint movement, and the in-place error value indicates the deviation between the joint and the ideal position when it moves to the target position. If the response delay value exceeds the eighth threshold or the in-place error value exceeds the ninth threshold, it indicates that the morphology adaptation process is abnormal. At this time, the scene adaptation module generates a morphology adaptation abnormality signal and triggers the adaptive control module to perform a reverse calibration action. The reverse calibration action adjusts the position of the joint to make it close to the ideal position by sending a reverse control signal to the joint to correct the deviation in the morphology adaptation process.

[0069] The instrument life prediction module collects data such as the number of morphological switching times, joint wear coefficient, and drive unit load peak. The number of morphological switching times reflects the frequency of use of the instrument in different surgical scenarios. The joint wear coefficient is obtained through a comprehensive analysis of factors such as joint material properties, usage time, and movement frequency. The drive unit load peak reflects the maximum load borne by the drive unit during operation. The instrument health score is obtained through weighted calculation. For example, the weight of the number of morphological switching times is set to 0.3, the weight of the joint wear coefficient is set to 0.4, and the weight of the drive unit load peak is set to 0.3. If the health score is lower than the tenth threshold, it indicates that the health of the instrument is poor and there is a high risk of failure. At this time, the instrument life prediction module generates an instrument maintenance warning signal and prohibits the scene adaptation module from performing high-risk morphological switching operations to avoid surgical failure or harm to the patient due to instrument failure.

[0070] Example 6:

[0071] This example focuses on the functions of the biosignal fusion module and its key role in ensuring patient safety. By monitoring and analyzing the patient's physiological signals, the biosignal fusion module can promptly detect the patient's stress response and take appropriate measures to ensure patient safety during surgery.

[0072] The biosignal fusion module performs noise reduction and feature extraction on the physiological signals collected by the multimodal sensing module. Physiological signals are often subject to interference from various noise sources, such as environmental noise and instrument noise. Digital filtering techniques, such as low-pass and high-pass filtering, remove noise signals while retaining useful physiological signal components. A feature extraction algorithm then identifies the patient's heart rate variability and muscle tension index. Heart rate variability reflects the activity and regulatory capacity of the cardiac autonomic nervous system, while muscle tension index reflects the patient's muscle tension.

[0073] If the heart rate variability characteristic exceeds the eleventh threshold or the muscle tension index exceeds the twelfth threshold, it indicates that the patient is experiencing a stress response. At this point, the biosignal fusion module generates a patient stress signal and triggers the adaptive control module to reduce the device's motion speed to a safe level. For example, the speed may be reduced to one-third of the normal speed to reduce irritation to the patient's tissues and alleviate the patient's stress.

[0074] The biosignal fusion module further analyzes the trigger frequency and duration of the stress signal, and marks the number of triggers per unit time as the stress frequency value. If the stress frequency value exceeds the thirteenth threshold, it means that the patient's stress response is more frequent and severe, and continuing the operation may pose a threat to the patient's life safety. At this time, the biosignal fusion module generates a surgery suspension recommendation signal and pushes it to the human-computer interaction module to request the doctor to confirm the subsequent operation strategy. The doctor can decide whether to suspend the operation based on the patient's specific situation and take appropriate measures to alleviate the patient's stress response, such as adjusting the anesthesia depth, giving drug intervention, etc., and continue the operation after the patient's condition stabilizes.

[0075] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "includes," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0076] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A minimally invasive surgery system based on artificial intelligence, characterized in that: Including target recognition and positioning module, multimodal sensing module, adaptive control module, human-computer interaction module and scene adaptation module; The target recognition and positioning module uses image recognition and deep learning algorithms to perform real-time recognition and three-dimensional positioning of the surgical target area, generates target coordinate data, and sends it to the adaptive control module; The multimodal sensing module integrates force feedback sensors, position tracking sensors, and bioelectric signal sensors to collect real-time force data, motion trajectory data, and patient physiological signal data of surgical instruments; The adaptive control module adjusts the motion parameters of the surgical instrument according to the target coordinate data and the real-time sensor data, and generates a motion trajectory deviation signal if it is detected that the deviation value between the motion trajectory data and the preset path exceeds a first threshold value; The human-computer interaction module displays surgical parameters, sensor data, and deviation signals, and provides a touch operation interface and voice command interface; The scene adaptation module dynamically adjusts the instrument shape and functional mode according to the type of surgery, including bending angle adjustment, telescopic length control and tool head switching.

2. The minimally invasive surgery system based on artificial intelligence according to claim 1, characterized in that: The specific operation process of the target recognition and positioning module is as follows: A three-dimensional model of the patient's anatomical structure is constructed through preoperative medical images, and an endoscopic video stream is obtained in real time during the operation. Each frame of the video stream is aligned with the three-dimensional model, and the real-time spatial coordinates of the target area are calculated; the alignment error value and coordinate offset are collected. If the alignment error value exceeds the second threshold or the coordinate offset exceeds the third threshold, a positioning abnormality signal is generated; the positioning abnormality signal is sent to the human-computer interaction module for warning, and the adaptive control module is triggered to suspend the movement of the instrument until the abnormality is resolved.

3. The minimally invasive surgery system based on artificial intelligence according to claim 1, characterized in that: The data processing process of the multimodal sensing module includes: The raw data of the force feedback sensor is converted into the force vector at the end of the instrument, and the difference between it and the preset safety force threshold is calculated and marked as the force segregation value; the raw data of the position tracking sensor is fitted with the preset motion path, and the trajectory curvature change rate and speed fluctuation value are extracted, and marked as the curvature analysis value and the fluctuation analysis value respectively; if the force segregation value, curvature analysis value or fluctuation analysis value exceeds the corresponding fourth threshold, an abnormal instrument operation signal is generated, and the abnormal signal type and level are sent to the adaptive control module.

4. The minimally invasive surgery system based on artificial intelligence according to claim 1, characterized in that: The control logic of the adaptive control module is as follows: When a motion trajectory deviation signal is received, a compensatory motion instruction is generated based on the reinforcement learning algorithm to adjust the velocity vector and direction angle of the end of the device; during the compensation process, the decay rate of the trajectory deviation value is continuously monitored. If the decay rate is lower than the fifth threshold, it switches to the preset safety mode, limits the maximum movement speed of the device and narrows the operating range; if the deviation value returns to zero within the preset time, the current control is recorded as a valid control event.

5. The minimally invasive surgery system based on artificial intelligence according to claim 3, characterized in that: The system also includes an operational stability evaluation module, which collects the total number of control events and the proportion of effective control events generated by the adaptive control module, compares them with historical data to calculate a stability index; if the stability index is lower than a sixth threshold, an instrument control degradation signal is generated, and the scene adaptation module is triggered to switch to a redundant control mode and enable a backup drive unit.

6. The minimally invasive surgery system based on artificial intelligence according to claim 5, characterized in that: The operational stability assessment module further analyzes the temporal distribution characteristics of control events, extracts high-frequency abnormal periods and low-frequency abnormal periods, and marks the duration ratio of high-frequency abnormal periods as the period abnormal frequency value; If the time period frequency difference value exceeds the seventh threshold, an operating environment interference signal is generated and pushed to the human-computer interaction module to prompt the doctor to check the connection status of the external device.

7. The minimally invasive surgery system based on artificial intelligence according to claim 1, characterized in that: The function switching process of the scene adaptation module includes: The preset morphology database is called according to the type of surgery to match the bending angle parameters, telescopic length parameters and tool head type of the instrument; during the switching process, the response delay value and the in-place error value of each joint are collected. If the response delay value exceeds the eighth threshold or the in-place error value exceeds the ninth threshold, a morphology adaptation abnormality signal is generated, and the adaptive control module is triggered to perform a reverse calibration action.

8. The minimally invasive surgery system based on artificial intelligence according to claim 7, characterized in that: The system also includes an instrument life prediction module, which collects the number of form switching times, joint wear coefficient and drive unit load peak, and obtains the instrument health score through weighted calculation; if the health score is lower than the tenth threshold, an instrument maintenance warning signal is generated, and the scene adaptation module is prohibited from performing high-risk form switching operations.

9. The minimally invasive surgery system based on artificial intelligence according to claim 1, characterized in that: It also includes a biological signal fusion module, which performs noise reduction and feature extraction on the physiological signals collected by the multimodal sensing module to identify the patient's heart rate variability characteristics and muscle tension index; If the heart rate variability characteristic exceeds the eleventh threshold or the muscle tension index exceeds the twelfth threshold, a patient stress signal is generated and the adaptive control module is triggered to reduce the movement speed of the device to a safe level.

10. The minimally invasive surgery system based on artificial intelligence according to claim 9, characterized in that: The biosignal fusion module further analyzes the trigger frequency and duration of the stress signal, and marks the number of triggers per unit time as the stress frequency value; if the stress frequency value exceeds the thirteenth threshold, a surgery suspension recommendation signal is generated and pushed to the human-computer interaction module to request the doctor to confirm the subsequent operation strategy.

Images