Heart intervention catheter robot tracking control method and system based on data driving

Abstract

The invention discloses a heart intervention catheter robot tracking control method and system based on data driving. The method comprises the steps of data acquisition, system identification, controller design, visual feedback, closed-loop control and the like. Constructing a linear dynamic model of the catheter system through the collected sample data by utilizing a Koopman operator theory; a Chebyshev polynomial is adopted as an observation function, and a nonlinear system is linearized; designing a catheter robot tracking controller based on a linear quadratic regulator; using a key point detection algorithm based on deep learning to detect key point positions, including valve edges and blood vessel branches, of the operation target from the medical image, and taking the key point positions as visual feedback; a feedback control loop is designed, key point positions of visual feedback are input into a controller, the state of the catheter robot is adjusted in real time, and accurate tracking of an operation target is achieved. According to the method, accurate and automatic control over nonlinear dynamics of the catheter can be achieved without building an analytical model of the catheter.

Description

Technical Field

[0001] The present invention relates to the technical field of robot-assisted surgery systems, and more specifically, to a data-driven cardiac interventional catheter robot tracking control method and system. Background Art

[0002] Heart valve disease is a common and serious cardiovascular disease, mainly including valvular stenosis and valvular insufficiency. Valvular stenosis causes the opening of the heart valve to narrow, restricting blood flow and increasing the burden on the heart; while valvular insufficiency causes blood to flow back in the heart, affecting the normal function of the heart. These diseases usually lead to heart failure, angina pectoris, dyspnea and a significant decrease in quality of life, and can even be life-threatening in severe cases. Globally, the incidence of heart valve disease has increased year by year, becoming a major challenge in the field of public health.

[0003] Traditional treatment options usually involve surgical valve replacement. Although this method can effectively improve patients' symptoms, it is highly traumatic, has high surgical risks, and a long recovery period. Cardiac interventional surgery is a new technology for diagnosing and treating cardiovascular diseases. It is a diagnostic and treatment method that punctures surface blood vessels, inserts cardiac intervention catheters, and uses specific cardiac intervention catheter operation techniques to diagnose and treat heart disease. As a minimally invasive treatment method, transcatheter heart valve intervention (TAVI / TAVR) uses a catheter to deliver a folded artificial heart valve to a specific location in the heart, and precisely releases and implants it. The emergence of this technology has greatly reduced the surgical risks and recovery time of patients, and improved the treatment effect.

[0004] Despite this, the implementation of valve intervention surgery still faces many challenges. During the operation, the doctor needs to use a guidewire and catheter to accurately deliver the artificial valve to the target position and release it. This process requires the doctor to have extremely high operating accuracy, stability, and hand-eye coordination. Under the guidance of real-time fluoroscopic image feedback, the physician needs to constantly adjust the position and angle of the catheter to ensure the accurate implantation of the valve. Due to the narrow space inside the heart and the complex operating environment, any slight deviation may lead to surgical failure or complications. Therefore, this type of surgery currently relies mainly on the doctor's rich experience and superb operating skills. The operation takes a long time, is difficult, and the accuracy is difficult to guarantee.

[0005] In order to meet these challenges, robot-assisted catheter interventional surgery technology has gradually been introduced into clinical practice. Through the automatic control of the robot system, the catheter can be precisely manipulated, thereby improving the accuracy and stability of heart valve implantation. The robot system can provide higher operating accuracy in a narrow space, reduce the fatigue and burden of doctors, and help shorten the operation time, thereby improving the overall success rate and safety of the operation.

[0006] However, catheter robotic systems face challenges with highly nonlinear dynamic characteristics in practical applications. Due to the narrow space inside the heart and the certain compliance of the catheter material, traditional model-based control methods often have difficulty in effectively dealing with these nonlinear characteristics. The movement and operation of the catheter involve complex mechanical models and nonlinear dynamics, which make accurate modeling and control extremely difficult. Existing control algorithms and technologies often cannot meet the high requirements for accuracy and stability in actual surgery.

[0007] Therefore, it is urgent to study and develop new catheter robot control methods to overcome these nonlinear dynamic problems. Through innovative control strategies and algorithms, it is possible to automatically track and locate the heart valve, thereby significantly improving the accuracy and safety of valve interventional surgery. This will not only help improve the treatment effect of patients, but will also promote the further development and application of cardiac interventional surgery technology. Summary of the invention

[0008] In view of the difficulty of controlling the catheter robot in cardiac intervention surgery, the present invention provides a data-driven tracking control method and system for a cardiac intervention catheter robot, aiming to improve the accuracy and safety of the surgery.

[0009] To achieve the above object, on the one hand, the present invention provides a data-driven cardiac interventional catheter robot tracking control method, characterized in that it includes the following steps:

[0010] Data collection: In the cardiac interventional catheter robot system, sample data of system status including catheter position, angle, and posture and control input including motor speed and pressure are collected as the basis for training the model;

[0011] System identification: Using the Koopman operator theory, a linear dynamic model of the catheter system is constructed through the collected sample data; the Chebyshev polynomial is used as the observation function to linearize the nonlinear system;

[0012] Controller design: Design a catheter robot tracking controller based on a linear quadratic regulator and the identified linear dynamic model. The controller input is the desired state including the target position and angle, and the output is the control input parameters including motor speed and pressure.

[0013] Visual feedback: Using a deep learning-based key point detection algorithm, the key point locations of surgical targets, including valve edges and vascular branches, are detected from medical images as visual feedback;

[0014] Closed-loop control: Design a feedback control loop, input the key point positions of visual feedback into the controller, adjust the catheter robot state in real time, and achieve accurate tracking of the surgical target.

[0015] The present invention uses a data-driven method to achieve precise control of the nonlinear dynamics of the catheter without establishing an analytical model of the catheter. The Koopman operator theory can effectively capture the dynamic characteristics of nonlinear systems, while the Chebyshev polynomial can achieve approximate linearization of nonlinear systems, providing a basis for subsequent controller design. Combining visual feedback and designing feedback control loops can achieve precise tracking of cardiac interventional surgery targets (such as heart valves), improving surgical accuracy and safety.

[0016] Furthermore, data collection is performed under different surgical conditions, using a variety of sensors including but not limited to pressure sensors, position sensors, and speed sensors to record in detail the state of the catheter robot in the surgical environment; at the same time, the operator's control input to the catheter is recorded, including but not limited to force, direction, and speed parameters. This technical feature ensures that the data set covers a variety of possible operating scenarios and states to improve the generalization ability of the model.

[0017] Furthermore, in the system identification, the collected data is first subjected to denoising, normalization and data synchronization processing to ensure that the quality of the data meets the modeling requirements, and then the linear dynamic model of the catheter system is constructed using the Koopman operator theory.

[0018] Furthermore, in the system identification, the complexity and accuracy of the linear dynamic model are balanced by adjusting the order of the polynomial, ensuring that the model can capture the key dynamics of the system without overfitting.

[0019] Furthermore, when designing the controller, the cost function is set to comprehensively consider the tracking error and the smoothness of the control input. By adjusting the weight matrix of the LQR, the catheter robot has high accuracy and low oscillation when tracking the desired trajectory, so as to flexibly achieve the best balance between tracking accuracy and control smoothness.

[0020] Furthermore, the visual feedback includes developing and training a convolutional neural network model to adapt it to images under different lighting conditions, angles, and occlusions, to accurately identify and locate the key positions of surgical targets from medical images in real time; and to process the detected key point position information into visual feedback signals for the control system. This technical feature helps to achieve fast and accurate visual feedback to fine-tune the catheter machine status in real time.

[0021] On the other hand, the present invention provides a data-driven cardiac interventional catheter robot tracking control system, characterized in that it includes:

[0022] The data acquisition module collects sample data of system status including catheter position, angle, and posture, and control input including motor speed and pressure in the cardiac interventional catheter robot system as the basis for training the model;

[0023] The data-driven system identification module uses Chebyshev polynomials as observation functions to linearize nonlinear systems;

[0024] The visual feedback and optimal control module includes a linear quadratic regulator controller and a key point detection model based on deep learning; the controller guides the cardiac intervention catheter robot to collect real-time images for the key point detection model to track key points; the key point position of the visual feedback is input into the controller to adjust the state of the catheter robot in real time.

[0025] Compared with the prior art, the present invention has the following technical effects:

[0026] (1) The present invention uses a data-driven approach to achieve precise control of the nonlinear dynamics of the catheter without establishing an analytical model of the catheter.

[0027] (2) Improve surgical accuracy: The present invention can accurately place surgical instruments at the target location through precise tracking control, reduce surgical deviations, and improve surgical success rates. Closed-loop control can continuously modify control strategies based on real-time feedback information to improve control accuracy and robustness.

[0028] (3) Shorten the operation time: The automated surgical control provided by the present invention can reduce the doctor's operation time and improve the efficiency of the operation.

[0029] (4) Reduce surgical risks: Precise control can reduce damage to heart tissue and reduce the risk of surgical complications. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a flow chart of a cardiac interventional catheter robot control method according to an embodiment of the present invention;

[0031] Figure 2 This is a diagram of a key point detection network architecture in one embodiment of the present invention.

[0032] Figure 3 This is a schematic diagram of a cardiac interventional catheter robot control system in one embodiment of the present invention; DETAILED DESCRIPTION

[0033] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, but they are not intended to limit the present invention.

[0034] In the following detailed description, many specific details are set forth to provide a more thorough understanding of the present invention. However, it is obvious to those skilled in the art that well-known algorithms and models (such as Koopman operator, linear quadratic regulator, convolutional neural network) are not shown in detail to avoid obscuring the subject matter of the present invention.

[0035] In addition, the order of execution of actions, steps, etc. in the devices and methods shown in the claims, specifications and drawings can be implemented in any order as long as there is no special explicit limitation on the order and the output of the previous processing is not used in the subsequent processing.

[0036] Example 1

[0037] See also Figure 1 In this embodiment, the data-driven cardiac interventional catheter robot tracking control method includes the following steps:

[0038] Data collection: In the cardiac interventional catheter robot system, sample data of system status including catheter position, angle, and posture and control input including motor speed and pressure are collected as the basis for training the model. More specifically, under different surgical conditions, various sensors including but not limited to pressure sensors, position sensors, speed sensors, etc. are used to record in detail the various states of the catheter robot in the surgical environment. At the same time, the operator's control input to the catheter is recorded, including but not limited to parameters such as force, direction, and speed, to ensure that the data set covers a variety of possible operating scenarios and states to improve the generalization ability of the model.

[0039] System identification: Using the Koopman operator theory, a linear dynamic model of the catheter system is constructed through the collected sample data; more specifically, the Koopman operator theory is applied to map the collected data into a high-dimensional space so that the dynamic characteristics of the nonlinear system can be approximated by a linear model. Specifically, the Chebyshev polynomial is used as the observation function, which effectively linearizes the nonlinear behavior of the system with its superior performance in function approximation. The complexity and accuracy of the model are balanced by adjusting the order of the polynomial to ensure that the model can capture the key dynamics of the system without overfitting.

[0040] Controller design: A catheter robot tracking controller is designed based on a linear quadratic regulator (LQR) and the identified linear dynamic model. The controller input is the desired state including the target position and angle, and the output is the control input parameters including motor speed and pressure. More specifically, the cost function is set in detail, and the tracking error and the smoothness of the control input are comprehensively considered to ensure high accuracy and low oscillation of the catheter robot when tracking the desired trajectory. By adjusting the weight matrix of the LQR, the best balance between tracking accuracy and control smoothness can be flexibly achieved.

[0041] Visual feedback: Using a key point detection algorithm based on deep learning, the key point positions of the surgical target, including the valve edge and vascular branches, are detected from the medical image as visual feedback. As an example, a key point detection algorithm such as a convolutional neural network is used to accurately identify and locate the key positions of the surgical target from medical images in real time. Develop and train the network model to adapt to images under different lighting conditions, angles, and occlusions to improve the robustness of the detection. The detected key point position information is processed into a visual feedback signal for the control system to achieve real-time fine-tuning of the catheter machine state.

[0042] Closed-loop control: Design a feedback control loop to input the key point positions of visual feedback into the controller, calculate and adjust the catheter robot state (control input) in real time, and ensure that it can quickly respond to the movement and changes of the surgical target and maintain accurate tracking.

[0043] The following are some specific implementation examples of controlling catheter robots for heart valve interventional surgery based on the above control method. Advanced simulation software is used to build a virtual surgical environment that includes vascular structure and catheter robot dynamics, and detailed simulation experiments are conducted to evaluate the performance of the control strategy in different surgical scenarios, including but not limited to tracking accuracy, response time, and system stability.

[0044] 1. Data collection: In the catheter robot system, collect state variables such as catheter position and speed, as well as sample data of motor control input, with the data volume of no less than 50,000 groups.

[0045] Sensors and data acquisition devices are equipped in the catheter robot system. These devices include position sensors (such as laser displacement sensors, optical encoders), force sensors, accelerometers, etc.

[0046] The data acquisition system is set up to ensure that the position, velocity, angle of the catheter, and the applied control inputs (such as force and torque) are recorded in real time.

[0047] An example device configuration is:

[0048] - Position sensor: OptoNCDT 1420 laser displacement sensor, accuracy ±2μm

[0049] -Force sensor: ATI Nano17 six-axis force / torque sensor, resolution 0.003N

[0050] -Accelerometer: ADXL355 three-axis accelerometer, range ±2g

[0051] -Data acquisition system: National Instruments cDAQ-9178 chassis with NI-9215 analog input modules

[0052] Acquisition parameters:

[0053] - Sampling frequency: 1000Hz

[0054] - Collection time: 50 seconds / group

[0055] -Number of data sets: 1000 sets

[0056] Collection content:

[0057] - Catheter tip position (x,y,z)

[0058] -Catheter applied force (Fx, Fy, Fz)

[0059] -Catheter acceleration (ax,ay,az)

[0060] -Motor control input (voltage V1, V2, V3, V4)

[0061] Data Records:

[0062] During the operation of the catheter robot, sample data of system status and control inputs are collected. Data collection needs to be performed under different operating conditions to cover the possible working range.

[0063] The sample data should include the dynamic behavior data of the catheter under different states, such as the motion trajectory under different heart rhythms and valve positions.

[0064] 2. System Identification

[0065] Data preprocessing:

[0066] The collected data is preprocessed, including denoising, normalization and data synchronization, to ensure that the data quality meets the modeling requirements. For example, a Butterworth low-pass filter is used for denoising with a cutoff frequency of 50 Hz; normalization processing: all data are scaled to the range of [-1, 1].

[0067] Koopman operator theory application:

[0068] Using sample data and Koopman operator theory, the dynamic characteristics of the system are modeled, the dynamic behavior of the nonlinear system is mapped to a high-dimensional space, and the state space model of the system is obtained.

[0069] In this process, an appropriate observation function (such as Chebyshev polynomials) is selected to linearize the nonlinear system. The observation function example is a 5th-order Chebyshev polynomial, and the state space dimension is expanded from the original 13 dimensions to 78 dimensions.

[0070] Assume there is a nonlinear system whose state equation is:

[0071]

[0072] Where (x(t)) is the system state, (u(t)) is the control input, and (f) is the nonlinear dynamic function. Koopman operator theory maps the nonlinear behavior of the system into a high-dimensional space by introducing an observation function (φ), thereby obtaining a linear model.

[0073] First, define the observation function:

[0074]

[0075] Then, construct a linear dynamic system and use the Koopman operator Linearize the system:

[0076]

[0077] in, It is the matrix representation of the Koopman operator, which can make the dynamic behavior of the system linear in the observation function space.

[0078] 3. Controller Design

[0079] LQR controller design:

[0080] Based on the linear dynamic model, a linear quadratic regulator (LQR) controller is designed. The LQR controller controls the system state by optimizing the cost function. Determine the cost function, including the cost terms of the state error and the control input. Set the weight parameters to balance the system performance (accuracy) and the cost of the control input. Calculate the control gain matrix and generate the control strategy of the controller.

[0081] The goal of LQR controller design is to minimize a cost function, which is usually expressed as:

[0082] J = ∫0 ∞ (x T (t)Qx(t)+u T (t)Ru(t))dt

[0083] Where (x(t)) is the system state, (u(t)) is the control input, (Q) and (R) are weight matrices. The design of the LQR controller aims to find an optimal control input (u(t)) that minimizes the cost function (J).

[0084] The state feedback control law of the LQR controller can be expressed as:

[0085] [u(t)=-Kx(t)]

[0086] Where (K) is the control gain matrix, which is determined by the following matrix equation:

[0087] [A T P+PA-PBR -1 B T P+Q=0]

[0088] Among them, (P) is the solution matrix of the cost function, (A) is the system matrix, (B) is the control input matrix, (R) is the weight matrix of the control input, and (Q) is the weight matrix of the state.

[0089] The control gain matrix (K) is calculated as:

[0090] [K=R -1 B T P]

[0091] Controller implementation:

[0092] The LQR algorithm is implemented in the controller. Based on the desired state (such as the position and posture of the heart valve), the control input is calculated and applied to the catheter robotic system.

[0093] LQR parameter settings:

[0094] -State weight matrix Q: diagonal matrix, the main diagonal elements are [100,100,100,10,10,1,1,1,1,1,1,1]

[0095] -Control input weight matrix R: diagonal matrix, the main diagonal elements are [1,1,1,1]

[0096] 4. Visual Feedback

[0097] Image acquisition:

[0098] High-resolution medical imaging equipment is used to collect image data in real time during surgical scenarios.

[0099] Image acquisition equipment:

[0100] -X-ray imaging system: Siemens Artis zee, frame rate 30fps, resolution 1024x1024

[0101] Key point detection:

[0102] Apply deep learning models (such as convolutional neural networks (CNNs)) for key point detection. Train the model to identify and locate key points of surgical targets (such as the edge or center of a heart valve). Perform coordinate transformation on the detected key points and convert them into the visual feedback format required by the control system.

[0103] Assume that the keypoint position detected from the medical image is (pd) and the actual target position is (ptre). Visual feedback is used to adjust the state of the catheter robot system so that the catheter robot can track the target position.

[0104] Set the target position error e to:

[0105] [e=ptre-pd]

[0106] The controller is adjusted based on this error and the control input (u(t)) is calculated by the following formula:

[0107] [u(t)=-Ke]

[0108] Where (K) is the control gain matrix, which is used to adjust the control input according to the error to reduce the error and achieve accurate tracking.

[0109] UNet is a special CNN structure that is particularly suitable for image segmentation tasks, including key point detection. Figure 2 ,The UNet structure consists of an encoder path and a decoder path, which are connected by skip connections to preserve the ,detailed information of the image.

[0110] UNet network parameters:

[0111] - Input image size: 256x256x1

[0112] - Number of encoder channels: [64,128,256,512]

[0113] -Decoder channels: [512,256,128,64]

[0114] -Convolution kernel size: 3x3

[0115] -Pooling size: 2x2

[0116] -Activation function: ReLU

[0117] Training parameters:

[0118] -Batch size: 16

[0119] -Learning rate: 0.001, using Adam optimizer

[0120] - Number of training rounds: 200

[0121] -Loss function: Mean Squared Error (MSE)

[0122] 5. Closed-loop control

[0123] Feedback Integration:

[0124] The key point positions in the visual feedback are input into the LQR controller. The state of the catheter robot is adjusted using the real-time feedback data.

[0125] Implement closed-loop control to apply control input to the catheter robotic system to adjust the position and angle of the catheter to match the target position in the visual feedback.

[0126] Control cycle frequency: 100Hz

[0127] Control gain matrix K: calculated by MATLAB's lqr function.

[0128] Real-time adjustments:

[0129] During the operation, the status and target position of the catheter robot are monitored in real time. The closed-loop control system continuously adjusts the movement of the catheter robot to ensure that it accurately tracks the target.

[0130] 6. Simulation Verification

[0131] Simulation environment construction:

[0132] A virtual model of the catheter robot system is constructed in the simulation environment, including a model of the heart valve and a simulation of the surgical scene, to realize the simulation of the data-driven model and the LQR controller.

[0133] Simulation test:

[0134] Test the effectiveness of control methods in a simulation environment. Simulate different operating conditions and surgical scenarios to verify the tracking accuracy of the controller and the stability of the system.

[0135] Simulation environment:

[0136] -Use MATLAB Simulink to build simulation models

[0137] -Heart and blood vessel models: 3D models built based on CT scan data

[0138] -Catheter dynamics model: flexible body model based on finite element method

[0139] Simulation parameters:

[0140] -Simulation duration: 60 seconds

[0141] -Simulation step: 0.001 seconds

[0142] -Target trajectory: sine wave, amplitude 10mm, frequency 0.5Hz

[0143] The simulation evaluates the performance of the control method in actual operation to ensure that it meets the requirements of surgical accuracy and safety.

[0144] Evaluation Metrics:

[0145] -Mean Tracking Error (MTE)

[0146] -Maximum tracking error (MaxTE)

[0147] -Control stability (standard deviation)

[0148] In the simulation environment, verify the feasibility and tracking accuracy of the control method. Assume that the state equation in the simulation environment is:

[0149]

[0150] During the simulation, the control input is given by the LQR controller:

[0151] [u(t)=-Kx(t)]

[0152] Simulation validation includes evaluating system responses and errors:

[0153] [x sim (t) = Ax sim (t)+B(-Kx sim (t))]

[0154] The error is calculated as:

[0155] [Error(t)=|x target (t)-x sim (t)|]

[0156] Among them, (x target (t)) is the desired target state, (x sim (t)) is the state obtained by simulation.

[0157] Result analysis:

[0158] Analyze simulation results, adjust and optimize control strategies. Adjust controller parameters or UNet network structure according to test results to ensure the best performance of the control system in actual surgery.

[0159] Example 2

[0160] See also Figure 3 This embodiment provides a data-driven cardiac interventional catheter robot tracking control system, including:

[0161] The data acquisition module 100 collects sample data of system status including catheter position, angle, and posture and control input including motor speed and pressure in the cardiac interventional catheter robot system as the basis for training the model;

[0162] The data-driven system identification module 200 uses Chebyshev polynomials as observation functions to linearize the nonlinear system;

[0163] The visual feedback and optimal control module 300 includes a linear quadratic regulator controller and a key point detection model based on deep learning; the controller guides the cardiac intervention catheter robot to collect real-time images for the key point detection model to track key points; the key point position of the visual feedback is input into the controller to adjust the catheter robot state in real time.

[0164] If the data-driven cardiac interventional catheter robot tracking control method and system is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Therefore, the essence of this technical solution or the part that contributes to the prior art or the part of this technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for an electronic device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0165] In summary, the present invention provides a data-driven cardiac interventional catheter robot tracking control method and system, including data acquisition, system identification, controller design, visual feedback, closed-loop control and other steps. Using the Koopman operator theory, a linear dynamic model of the catheter system is constructed through the collected sample data; the Chebyshev polynomial is used as the observation function to linearize the nonlinear system; the catheter robot tracking controller is designed based on the linear quadratic regulator; the key point detection algorithm based on deep learning is used to detect the key point positions of the surgical target including the valve edge and vascular branches from the medical image as visual feedback; a feedback control loop is designed to input the key point positions of the visual feedback into the controller, adjust the state of the catheter robot in real time, and achieve accurate tracking of the surgical target. The present invention can achieve precise and automated control of its nonlinear dynamics without establishing an analytical model of the catheter.

[0166] Those skilled in the art should understand that those skilled in the art can implement variations by combining the prior art and the above embodiments, which will not be described in detail here. Such variations do not affect the essential content of the present invention, and will not be described in detail here.

[0167] The above describes the preferred embodiments of the present invention. It should be understood that the present invention is not limited to the above-mentioned specific embodiments, and the devices and structures that are not described in detail should be understood to be implemented in a common manner in the art; any technician familiar with the art can use the above-disclosed methods and technical contents to make many possible changes and modifications to the technical solutions of the present invention without departing from the scope of the technical solutions of the present invention, or modify them into equivalent embodiments of equivalent changes, which does not affect the essential content of the present invention. Therefore, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solutions of the present invention are still within the scope of protection of the technical solutions of the present invention.

[0168] Those skilled in the art will appreciate that the units, i.e., algorithm steps, of the various examples described in conjunction with this embodiment can be implemented in electronic hardware or in a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

Claims

1. A data-driven cardiac interventional catheter robot tracking control method, characterized in that: The following steps are involved: Data collection: In the cardiac interventional catheter robot system, sample data of system status including catheter position, angle, and posture and control input including motor speed and pressure are collected as the basis for training the model; System identification: Using the Koopman operator theory, a linear dynamic model of the catheter system is constructed through the collected sample data; the Chebyshev polynomial is used as the observation function to linearize the nonlinear system; Controller design: Design a catheter robot tracking controller based on a linear quadratic regulator and the identified linear dynamic model. The controller input is the desired state including the target position and angle, and the output is the control input parameters including motor speed and pressure. Visual feedback: Using a deep learning-based key point detection algorithm, the key point locations of surgical targets, including valve edges and vascular branches, are detected from medical images as visual feedback; Closed-loop control: Design a feedback control loop, input the key point positions of visual feedback into the controller, adjust the catheter robot state in real time, and achieve accurate tracking of the surgical target.

2. The cardiac interventional catheter robot tracking control method according to claim 1, characterized in that: Data is collected under different surgical conditions, using a variety of sensors including but not limited to pressure sensors, position sensors, and speed sensors to record in detail the status of the catheter robot in the surgical environment; at the same time, the operator's control input to the catheter is recorded, including but not limited to force, direction, and speed parameters.

3. The cardiac interventional catheter robot tracking control method according to claim 1, characterized in that: In the system identification, the collected data is first subjected to denoising, normalization and data synchronization processing, and then the linear dynamic model of the catheter system is constructed using the Koopman operator theory.

4. The cardiac interventional catheter robot tracking control method according to claim 1 or 3, characterized in that: In the system identification, the complexity and accuracy of the linear dynamic model are balanced by adjusting the order of the polynomial to ensure that the model can capture the key dynamics of the system without overfitting.

5. The cardiac interventional catheter robot tracking control method according to claim 1, characterized in that: When designing the controller, the cost function is set, and the tracking error and the smoothness of the control input are comprehensively considered. By adjusting the weight matrix of the LQR, the catheter robot has high accuracy and low oscillation when tracking the desired trajectory.

6. The cardiac interventional catheter robot tracking control method according to claim 1, characterized in that: The visual feedback includes developing and training a convolutional neural network model to adapt it to images under different lighting conditions, angles and occlusions, so as to accurately identify and locate the key positions of surgical targets from medical images in real time; and processing the detected key point position information into a visual feedback signal of the control system.

7. A data-driven cardiac interventional catheter robot tracking control system, characterized in that: include: The data acquisition module collects sample data of system status including catheter position, angle, and posture, and control input including motor speed and pressure in the cardiac interventional catheter robot system as the basis for training the model; The data-driven system identification module uses Chebyshev polynomials as observation functions to linearize nonlinear systems; The visual feedback and optimal control module includes a linear quadratic regulator controller and a key point detection model based on deep learning; the controller guides the cardiac intervention catheter robot to collect real-time images for the key point detection model to track key points; the key point position of the visual feedback is input into the controller to adjust the state of the catheter robot in real time.

Images