Magnetic control intelligent traction system for endoscopic surgery and traction method thereof

Abstract

The invention discloses a magnetic control intelligent traction system for endoscopic surgery. The magnetic control intelligent traction system comprises an in-vivo magnetic clamp assembly, an in-vitro magnetic control device, a navigation positioning subsystem and a central processing unit. The in-vivo magnetic clamp assembly firmly grabs mucous membrane tissues through the shape memory effect and the surface spinule structure of a nickel-titanium alloy clamp arm; the in-vivo magnetic clamp assembly is integrated with an MEMS force sensor, actual traction force is detected in real time, and data of the force sensor are transmitted to the central processing unit through the wireless communication module. The in-vitro magnetic force control device is used for pulling the in-vivo magnetic clamp to move, receiving target magnetic field parameters of the central processing unit and outputting a PWM signal to adjust current; the navigation positioning subsystem receives the position information of the in-vivo magnetic clamp, tracks the position of an in-vivo magnetic clamp assembly, and distributes an optimal traction vector through a particle swarm optimization algorithm; the central processing unit receives navigation positioning data, feedback of the force sensor and an operator operation instruction, generates a magnetic field control signal and distributes traction force vectors of all clamps through a PSO algorithm. Accurate and stable output of traction force in the ESD operation is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical devices, and particularly relates to an endoscopic surgery magnetically controlled intelligent traction system and a traction method thereof. Background Art

[0002] Endoscopic submucosal dissection (ESD) is a new treatment method that has emerged in recent years. It has good clinical application prospects, enabling more early-stage digestive tract cancers to be completely resected endoscopically at one time. However, ESD has relatively long operation time, great operation difficulty and high operation risk, and poor exposure of the submucosal field of view is the most important reason for these difficulties. Currently, there is no particularly ideal traction method.

[0003] The existing traction tools such as dental floss and nylon rings used in endoscopic submucosal dissection (ESD) have the following defects: the traction force cannot be quantitatively controlled, relying on the operator's hand feeling, which is likely to cause tissue tearing or insufficient traction; the traditional metal clips cannot dynamically adjust the traction direction and force after fixation; there is a lack of real-time force feedback mechanism, posing a safety hazard.

[0004] In view of the above problems, it is necessary to design an endoscopic surgery magnetically controlled intelligent traction system with small trauma, fast response and accurate and stable output of traction force. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide an endoscopic surgery magnetically controlled intelligent traction system, which realizes accurate and stable output of traction force during ESD.

[0006] To solve the above technical problem, the present invention adopts the following technical solution: an endoscopic surgery magnetically controlled intelligent traction system, comprising an in-vivo magnetic clip assembly, an external magnetic force control device, a navigation and positioning subsystem and a central processor;

[0007] The in-vivo magnetic clip assembly firmly grasps the mucosal tissue through the shape memory effect of the nickel-titanium alloy clip arm and the surface micro-spike structure; the in-vivo magnetic clip assembly integrates a MEMS force sensor to detect the actual traction force in real time, and transmits the force sensor data to the central processor through a wireless communication module;

[0008] The external magnetic force control device is used to pull the in-vivo magnetic clip to move, receives the target magnetic field parameters of the central processor, and outputs a PWM signal to adjust the current;

[0009] The navigation and positioning subsystem receives the position information of the in-vivo magnetic clip and tracks the position of the in-vivo magnetic clip assembly, and allocates the optimal traction vector through the particle swarm optimization algorithm;

[0010] The central processor receives the navigation and positioning data, the force sensor feedback and the operator's operation instructions, generates a magnetic field control signal, and allocates the traction force vector of each clip through the PSO algorithm.

[0011] Furthermore, the in-vivo magnetic clip assembly includes a clip handle, nitinol clip arms, neodymium iron boron magnets, MEMS force sensors, a wireless transmission module, and a detachable unlocking buckle structure; the clip arms are connected to the clip handle through a nitinol hinge; the neodymium iron boron magnets are embedded in the inner grooves of the nitinol clip arms; the MEMS force sensors are welded onto a flexible PCB, the PCB is attached to the root of the clip arm, and is connected to the MCU inside the clip handle through micro-wires; the wireless transmission module and the MCU are integrated in the sealed cavity of the clip handle, and the cavity is encapsulated with medical silicone.

[0012] Furthermore, the closing dimension of the in-vivo magnetic clip assembly is ≤8 mm, the opening angle limit of the clip arms is 60°, and a temperature adaptive compensation circuit is configured.

[0013] Furthermore, the tips of the nitinol clip arms are embedded with ring neodymium iron boron magnets. An external pulsed magnetic field triggers the phase change of the nitinol, and the clip arms resume opening. When the external magnetic field is switched to the reverse direction, the clip arms start to close under the action of magnetic torque.

[0014] Furthermore, the external magnetic force control device consists of an 8×8 electromagnetic array composed of Helmholtz coils, is configured with a PID closed-loop control module, and supports a constant traction mode and a pulsed traction mode.

[0015] Furthermore, the control method of the PID closed-loop control module includes the following steps:

[0016] a) Receive the actual traction force fed back by the MEMS force sensor;

[0017] b) Calculate the deviation between the target force and the actual traction force;

[0018] c) Adjust the current of the electromagnetic array according to the traction force deviation;

[0019] d) Trigger magnetic field cut-off protection when the actual traction force ≥3 N.

[0020] Furthermore, the navigation and positioning subsystem includes an in-vivo positioning tag, an external signal receiving and processing module, and a fusion positioning and calculation module. The in-vivo positioning tag emits radio frequency signals; the external signal receiving and processing module calculates the time difference of arrival and phase difference of the signals; the fusion positioning and calculation module fuses the endoscopic optical image and the electromagnetic positioning data to generate an augmented reality interface with superimposed three-dimensional coordinates.

[0021] The present invention also provides an endoscopic surgery traction control method, including the following steps:

[0022] (1) Deploy the magnetic clip through the endoscopic channel;

[0023] (2) Obtain the tissue depth d by CT scan and input it into the compensation algorithm;

[0024] (3) Set the target traction force Ft;

[0025] (4) Generate a gradient magnetic field through the electromagnetic array to tow the magnetic clip to move;

[0026] (5) Monitor Fn in real time and dynamically adjust the magnetic field strength;

[0027] (6) After reaching the predetermined peeling range, unlock the recovery clip body.

[0028] Furthermore, the endoscopic surgery traction control method specifically includes the following steps:

[0029] (1) Deploy the magnetic clip through the endoscopic channel

[0030] Under direct vision of the endoscope, send the push catheter through the forceps channel to the vicinity of the target mucosa;

[0031] Rotate the push handle, and the buckle at the front end of the catheter releases the magnetic clip. The clip arms are triggered by body temperature to have a shape memory effect, but are still restricted by the lock in the closed state;

[0032] Confirm the relative position of the clip body and the lesion through the endoscopic image;

[0033] (2) Obtain the tissue depth d by CT scan and input it into the compensation algorithm

[0034] Adopt a thin-layer CT scan range to cover the target area and generate a three-dimensional model including the clip body position, mucosal layer thickness, and surrounding blood vessel distribution;

[0035] Depth calculation:

[0036] a. Algorithm input

[0037] The shortest distance d from the clip body to the body surface

[0038] Tissue density correction factor ε

[0039] b. Compensation formula

[0040]

[0041] Among them, k = 0.67, which is the calibration coefficient;

[0042] Input d and ε into the central processing unit to generate the initial magnetic field parameter B, and synchronize it to the electromagnetic control device through Ethernet;

[0043] (3) Set the target traction force Ft

[0044] Set Ft through the slider or numeric keypad or voice recognition module, supporting 0.1N step adjustment;

[0045] Select the traction mode;

[0046] (4) Generate a gradient magnetic field through the electromagnetic array to tow the magnetic clip to move

[0047] a. Magnetic field parameter calculation

[0048] Three-dimensional vector decomposition: Calculate the target magnetic field gradient direction according to the clip position provided by the navigation subsystem:

[0049]

[0050] The current of each coil unit in the 8×8 electromagnetic array ni is the coil normal vector,

[0051] The FPGA outputs a PWM waveform, which is amplified by the H-bridge to drive the coil, and the magnetic field uniformity is calibrated through the reference tag every 5 minutes;

[0052] The clip generates a magnetic force under the action of the magnetic field gradient:

[0053]

[0054] Among them, χ = 1.2 is the magnetic susceptibility of neodymium iron boron, V = 8mm 3 is the magnet volume;

[0055] The traction vector is displayed in real time and superimposed on the endoscopic image, and the operator can fine-tune the direction through the foot pedal;

[0056] (5) Monitor Fn in real time and dynamically adjust the magnetic field intensity

[0057] The MEMS force sensor collects the strain at the root of the clip arm and transmits it to the central processor;

[0058] PID closed-loop regulation, the algorithm is executed as follows:

[0059]

[0060] Among them, I O is the input current value of the coil in the electromagnetic array at the current moment or the previous control cycle, I N is the updated current value that needs to be applied at the next moment or the current control cycle after being calculated by the PID algorithm;

[0061] e = Ft - Fa, is the traction force error; the parameter Kp (proportional coefficient) = 0.8; Ki (integral coefficient) = 0.2; Kd (derivative coefficient) = 0.05; Δt is the time difference between the previous control action and the current control action;

[0062] The step response time < 100ms, the steady-state error < ±0.1N;

[0063] If Fa ≥ 2.8N, the system automatically reduces the gain by 50%; when the BLE signal loss > 500ms, switch to the preset safety magnetic field;

[0064] (6) After reaching the predetermined peeling range, unlock the recovery clamp body.

[0065] The beneficial effects of the present invention are as follows: Through magnetic force dynamic regulation and intelligent feedback control, the present invention realizes the precise and stable output of the traction force during ESD surgery. Compared with the prior art, it has outstanding advantages such as small trauma, fast response, and multi-physical field coordination, providing an innovative solution for the development of minimally invasive surgical instruments. Description of the Drawings

[0066] The following further describes the present invention in conjunction with the drawings and embodiments.

[0067] Figure 1 It is a flowchart of the present invention.

[0068] Figure 2 It is a schematic structural diagram of the in-vivo magnetic clip assembly of the present invention. Detailed Embodiments

[0069] The following will clearly and completely describe the technical solutions of the present invention through specific embodiments.

[0070] Embodiment 1

[0071] Refer to Figure 1 and Figure 2 , an endoscopic surgical magnetic control intelligent traction system of the present invention includes an in-vivo magnetic clip assembly, an external magnetic force control device, a navigation and positioning subsystem, and a central processor;

[0072] The in-vivo magnetic clip assembly firmly grasps the mucosal tissue through the shape memory effect of the nickel-titanium alloy clip arm and the surface micro-spike structure; the in-vivo magnetic clip assembly integrates a MEMS force sensor to detect the actual traction force in real time, and transmits the force sensor data to the central processor through a wireless communication module;

[0073] The external magnetic force control device is used to pull the in-vivo magnetic clip to move, receives the target magnetic field parameters of the central processor, and outputs a PWM signal to adjust the current;

[0074] The navigation and positioning subsystem receives the position information of the in-vivo magnetic clip and tracks the position of the in-vivo magnetic clip assembly, and allocates the optimal traction vector through the particle swarm optimization algorithm;

[0075] The central processor receives the navigation and positioning data, the force sensor feedback, and the operator's operation instructions, generates a magnetic field control signal, and allocates the traction force vector of each clip through the PSO algorithm.

[0076] The in-vivo magnetic clip assembly of this embodiment includes a clip handle 1, a nickel-titanium alloy clip arm 2, a neodymium-iron-boron magnet 3, a MEMS force sensor 4, and a wireless transmission module 5; the clip arm and the clip handle are connected by a nickel-titanium alloy hinge; the neodymium-iron-boron magnet is embedded in the inner groove of the nickel-titanium alloy clip arm; the MEMS force sensor is welded on a flexible PCB, the PCB is attached to the root of the clip arm, and is connected to the MCU in the clip handle through micro-wires.

[0077] The nickel-titanium alloy clip arm is made of super-elastic nickel-titanium alloy, with a thickness of 0.15 - 0.3 mm, formed by laser cutting, and using the shape memory characteristic, it automatically restores the preset shape at body temperature.

[0078] The miniature neodymium-iron-boron magnet uses an N52 grade magnet, with a diameter of 2 mm, and a Parylene-C coating (thickness 3 μm) on the surface to ensure biocompatibility. The miniature neodymium-iron-boron magnet is symmetrically embedded near the tip inside the clip arm, and is bonded to the reserved groove in the clip arm by medical epoxy resin glue to avoid displacement. The magnetic pole directions are arranged in the N-S same direction to generate the maximum magnetic moment with the external magnetic field.

[0079] The tip of the nickel-titanium alloy clip arm is embedded with a ring-shaped neodymium-iron-boron magnet. An external pulsed magnetic field is applied to trigger the phase change of the nickel-titanium alloy, and the clip arm restores to open. When the external magnetic field is switched to the reverse direction, the clip arm starts to close under the action of the magnetic torque.

[0080] The in-vivo magnetic clip assembly of the present invention is convenient to control its opening and closing.

[0081] The MEMS force sensor is a miniature piezoresistive sensor (Bosch BMI160), integrated at the hinge of the clip arm root, directly detecting the strain force when the clip arm opens and closes. After the sensor output is converted by analog-to-digital conversion (ADC), it is processed by an embedded microcontroller (MCU).

[0082] The wireless transmission module is a low-power Bluetooth (BLE) transmission module, using Nordic nRF52840, supporting the BLE 5.0 protocol, co-located with the MCU in the inner cavity of the clip handle, and the antenna is embedded in the non-metallic part of the clip handle to reduce signal shielding. The wireless transmission module wirelessly transmits the force sensor data to the external controller at a frequency of 10 Hz.

[0083] The in-vivo magnetic clip assembly also includes a temperature adaptive compensation circuit, including a thermistor (NTC), which real-time monitors the clip body temperature (0 - 60 °C) to compensate for the sensor drift caused by temperature changes.

[0084] The external magnetic force control device The external magnetic force control device consists of a Helmholtz coil group, configured with a PID closed-loop control module, supporting a constant traction mode and a pulsed traction mode.

[0085] The Helmholtz coil set consists of 8×8 independent Helmholtz coil units. Each unit contains a pair of coaxial toroidal coils (diameter 120 mm, spacing 60 mm), generating a uniform gradient magnetic field (maximum 200 mT) and supporting three-dimensional vector synthesis.

[0086] The coil winding is made of high-purity copper wire (diameter 2 mm, insulation layer withstand voltage 1 kV), and the skeleton is 3D printed nylon 12 (temperature resistance 120 °C).

[0087] Each coil corresponds to an H-bridge drive circuit (such as Infineon IGBT module), with a maximum output current of 10 A and a response time < 1 ms.

[0088] Closed-loop Hall effect sensors (accuracy ±0.5%) are used to monitor the coil current in real time.

[0089] The PID closed-loop control module implements a parallel PID algorithm based on an FPGA (Xilinx Artix-7).

[0090] The extracorporeal magnetic force control device also includes a human-machine interaction interface and a signal receiving module. The human-machine interaction interface includes a touch screen, a foot switch, and a voice control module. The touch screen is a 7-inch capacitive screen (resolution 1280×800), which displays the heat map of the magnetic field strength distribution and the real-time curve of the traction force. The foot switch uses a dual-channel Hall effect pedal (0-5N pressure sensing) and supports gradient adjustment (±0.2N / step). The voice control module integrates a microphone array and an NLP chip (such as TensorFlow Lite) to recognize the voice commands of the operator (such as "increase 0.5N").

[0091] The signal receiving module uses an nRF52840 chipset to receive the force / temperature data of the in-vivo magnetic clip (2.4 GHz, bandwidth 2 Mbps).

[0092] The operator imports the patient's CT / MRI data, and the system automatically extracts the tissue depth d and density parameter ε.

[0093] According to the formula Calculate the dehumidification magnetic field strength (k is the calibration coefficient);

[0094] The electromagnetic positioning system is aligned with the endoscopic image coordinates;

[0095] Set the target traction force through the touch screen or voice command, and select the constant traction mode (continuous force) or the pulse traction mode (0.5-2Hz periodic traction, simulating manual operation).

[0096] The central processor decomposes the magnetic field vector according to the target traction direction

[0097] Each coil unit in the 8×8 electromagnetic array receives an independent current instruction

[0098] N is the number of coil turns, R is the coil radius, ni is the coil normal vector, and μ0 is the vacuum permeability.

[0099] The signal receiving module receives the force feedback (Fa) and position data of the body clamp in real time and corrects the current using the PID algorithm:

[0100]

[0101] Update electromagnetic field output, response time <50ms.

[0102] The clamp position (x, y, z) is mapped to the electromagnetic coordinate system in real time, and the magnetic field gradient direction is automatically adjusted to ensure that the traction force is always along the tangent direction of mucosal peeling.

[0103] After the operator confirms that traction is completed, he / she turns off the magnetic field output through the foot switch or touch screen.

[0104] The navigation and positioning subsystem of this embodiment includes an in-vivo positioning tag, an in-vivo signal receiving and processing module, and a fusion positioning solution module.

[0105] The in-body positioning tag uses a miniature radio frequency tag with an operating frequency of 2.4GHz and a size of 2×2×0.5mm. The miniature radio frequency tag is embedded inside the handle of the magnetic clip and periodically transmits a coded radio frequency signal (50 times per second) containing a unique ID code and timestamp. The miniature radio frequency tag is powered by a button battery built into the magnetic clip.

[0106] A composite shielding layer of copper foil and ferrite can be used to cover the tag circuit to reduce the absorption and attenuation of radio frequency signals by body tissues. The thickness of the shielding layer is 0.1mm.

[0107] The extracorporeal signal reception and processing module includes four groups of receiving antennas, which are arranged in a regular tetrahedron with a spacing of 300mm, covering the surgical area. The RF front end uses the Nordic nRF52840 chipset, supports multi-channel parallel signal capture, and uses a GPS rubidium atomic clock as a high-precision clock synchronization module to ensure that the multi-antenna time synchronization error is less than 1ns.

[0108] The fusion positioning solution module implements the TDOA (time difference of arrival) and PDOA (phase difference of arrival) hybrid positioning algorithm based on Xilinx Zynq-7000;

[0109] The in-vivo RFID tag emits a pulse signal every 20 ms, which contains the ID code and the transmission timestamp t0. Four groups of antennas respectively record the signal arrival times t1, t2, t3, t4 and the phase information φ1, φ2, φ3, φ4. After the signal passes through the band-pass filter, the adaptive filtering algorithm (LMS algorithm) is used to eliminate the multipath effect interference.

[0110] Calculate the time difference Δt of the signal arriving at different antennas ij =t i -t j , and construct a hyperbolic equation system:

[0111]

[0112] where C is the speed of light, and (x i , y i , z i ) is the coordinate of the i-th antenna. The least squares method is used to solve the target coordinates (x, y, z) with an accuracy of ±1 mm.

[0113] Use the phase difference Δφij to correct the TDOA error and improve the depth (z-axis) resolution to ±0.3 mm;

[0114] Convert the electromagnetic positioning coordinate system (with the center of the receiving antenna array as the origin) to the endoscopic optical coordinate system (with the endoscopic lens as the origin).

[0115] In the endoscopic video, the position of the clip body (red mark), the traction direction (arrow vector) and the safety boundary (yellow dotted line) are rendered in real time;

[0116] When multiple clip bodies are deployed, the PSO algorithm is used to solve the optimal traction force distribution, and a calibration signal (known position reference tag) is automatically emitted every 5 minutes to correct the time base error caused by temperature drift.

[0117] If the signal of a certain antenna is lost, switch to the remaining 3-antenna positioning mode (the accuracy drops to

[0118] ±2 mm), and at the same time trigger an audible and visual alarm.

[0119] The central processor in this embodiment includes a multi-core processor, an FPGA acceleration module, a storage module and a signal processing module; the multi-core processor uses an ARM Cortex-A72 quad-core processor, equipped with the NEON SIMD instruction set to accelerate floating-point operations, and coordinates the communication and task scheduling of each subsystem.

[0120] The FPGA acceleration module uses a Zynq UltraScale+ MPSoC FPGA acceleration module, which integrates programmable logic units to process multi-sensor data streams in parallel.

[0121] The storage module is used to cache real-time data (such as force sensor sampling, positioning coordinates), surgical preset parameters, historical case database, and machine learning models, etc., including dynamic memory (DRAM) and non-volatile storage (NAND Flash).

[0122] The signal processing module includes an ADC / DAC module, which is used to collect MEMS force sensor signals and output magnetic field control voltage signals.

[0123] The central processor detects the status of the memory, FPGA, and communication interface. If an abnormality occurs, it triggers an LED alarm and calibrates the ADC / DAC reference voltage. Then it loads the RTOS kernel, algorithm library, and preset surgical templates (such as the default parameters for gastric ESD) from Flash, starts the database service, and loads historical surgical data (>10,000 cases) for the machine learning model to call.

[0124] During the operation, it receives in-vivo BLE sensor data (10Hz force / temperature), navigation positioning coordinates (50Hz), and endoscopic images (30fps), calculates real-time magnetic field compensation parameters, resolves the positions of multiple clips and performs coordinate transformation; predicts the optimal traction path (based on historical data similarity matching); uses the PSO algorithm to allocate the traction force weights of each clip, sends current commands to the electromagnetic array, and synchronously updates the AR navigation interface.

[0125] After the operation, it gradually reduces the current of the electromagnetic array to zero to avoid clip body jitter caused by magnetic field mutation, encrypts and stores the data of the entire surgical process (force curve, magnetic field parameters, operation log), and updates the machine learning model (incremental learning to improve the prediction accuracy of the next operation).

[0126] Embodiment 2

[0127] The endoscopic surgical traction control method of this embodiment includes the following steps:

[0128] Step 1: Deploy magnetic clips through the endoscopic channel

[0129] Under direct endoscopic vision, send the pushing catheter through the forceps channel to near the target mucosa (2-3mm away from the edge of the lesion);

[0130] Rotate the pushing handle, and the buckle at the front end of the catheter releases the magnetic clip. The clip arms trigger the shape memory effect due to body temperature (37°C), but are still restricted in the closed state by the lock;

[0131] Confirm the relative position of the clip body and the lesion through endoscopic images, and use methylene blue staining for marking if necessary.

[0132] Step 2: Obtain the tissue depth d by CT scan and input it into the compensation algorithm

[0133] Use thin-slice CT (slice thickness 0.625 mm, tube voltage 120 kV) to scan the target area; generate a three-dimensional model including the position of the clip body, the thickness of the mucosal layer, and the surrounding blood vessel distribution;

[0134] Depth calculation:

[0135] a. Algorithm input

[0136] The shortest distance d from the clip body to the body surface (measured by CT coordinates, accuracy ±1 mm)

[0137] Tissue density correction factor ε (converted according to CT values: ε = 0.1 for adipose tissue, ε = 0.3 for muscle, ε = 0.5 for fibrotic tissue)

[0138] b. Compensation formula

[0139]

[0140] Among them, k = 0.67, which is the calibration coefficient.

[0141] Input d and ε into the central processing unit to generate the initial magnetic field parameter B, and synchronize it to the electromagnetic control device through Ethernet.

[0142] Step three: Set the target traction force Ft

[0143] The operator sets Ft through the slider, numeric keypad, or voice recognition module, supporting 0.1 N step adjustment;

[0144] The traction mode can be selected as constant traction or pulsed traction; constant traction continuously and stably pulls, suitable for large-area peeling; pulsed traction has a period of 0.5 - 2 Hz, simulating the artificial pulling rhythm to reduce tissue fatigue.

[0145] The system automatically limits the maximum traction force ≤ 3 N, and triggers a three-level protection mechanism when exceeding the limit.

[0146] Step four: Generate a gradient magnetic field through the electromagnetic array to move the magnetic clip

[0147] a. Magnetic field parameter calculation

[0148] Three-dimensional vector decomposition: According to the position of the clip body (x, y, z) provided by the navigation subsystem, calculate the target magnetic field gradient direction:

[0149]

[0150] The current of each coil unit in the 8×8 electromagnetic array (ni is the coil normal vector).

[0151] The FPGA outputs a PWM waveform (frequency 20 kHz, duty cycle 0 - 100%), which is amplified by an H-bridge and then drives the coil. The magnetic field uniformity is calibrated every 5 minutes through a reference tag (error < ±5%).

[0152] The clamping body generates a magnetic force under the action of the magnetic field gradient:

[0153]

[0154] Among them, χ = 1.2 is the magnetic susceptibility of neodymium iron boron, and V = 8 mm 3 is the volume of the magnet.

[0155] The traction vector is superimposed on the endoscopic image in real time, and the operator can finely adjust the direction (±10°) by foot pedal.

[0156] (5) Monitor Fn in real time and dynamically adjust the magnetic field intensity

[0157] The MEMS force sensor collects the strain at the root of the clamping arm at a frequency of 1 kHz and transmits it to the central processor via BLE 5.0 (delay < 20 ms).

[0158] A fourth-order Butterworth low-pass filter (cutoff frequency 50 Hz) is used to eliminate the noise caused by breathing / heartbeat.

[0159] PID closed-loop regulation, and the algorithm is executed as follows:

[0160]

[0161] Among them, I O is the input current value of the coil in the electromagnetic array at the current moment (or the previous control cycle), and I N is the updated current value that needs to be applied at the next moment (the current control cycle) after being calculated by the PID algorithm;

[0162] e = Ft - Fa, which is the traction force error; the parameter Kp (proportional coefficient) = 0.8; Ki (integral coefficient) = 0.2; Kd (differential coefficient) = 0.05; Δt is the time difference between the previous control action and the current control action.

[0163] The step response time < 100 ms, and the steady-state error < ±0.1 N.

[0164] If Fa ≥ 2.8 N, the system automatically reduces the gain by 50%; when the BLE signal is lost > 500 ms, it switches to the preset safe magnetic field (maintaining 80% of the current value).

[0165] (6) After reaching the predetermined peeling range, unlock and recycle the clamping body.

[0166] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. Those skilled in the art can make various modifications or equivalent replacements to the present invention within the essence and protection scope of the present invention, and such modifications or equivalent replacements should also be regarded as falling within the protection scope of the technical solution of the present invention.

Claims

1. An endoscopic surgery magnetic control intelligent traction system, characterized in that: It includes an in-vivo magnetic clip assembly, an external magnetic force control device, a navigation and positioning subsystem, and a central processor; The in-vivo magnetic clip assembly firmly grasps mucosal tissue through the shape memory effect of nickel-titanium alloy clip arms and surface micro-spike structures; the in-vivo magnetic clip assembly integrates an MEMS force sensor to detect the actual traction force in real time, and transmits the force sensor data to the central processor through a wireless communication module; The external magnetic force control device is used to traction the in-vivo magnetic clip to move, receives the target magnetic field parameters of the central processor, and outputs a PWM signal to adjust the current; The navigation and positioning subsystem receives the position information of the in-vivo magnetic clip and tracks the position of the in-vivo magnetic clip assembly, and allocates the optimal traction vector through the particle swarm optimization algorithm; The central processor receives the navigation and positioning data, the force sensor feedback, and the operator's operation instructions, generates a magnetic field control signal, and allocates the traction force vector of each clip through the PSO algorithm.

2. The endoscopic surgical magnetic control intelligent traction system according to claim 1, characterized in that: The in-vivo magnetic clip assembly includes a clip handle, nickel-titanium alloy clip arms, neodymium iron boron magnets, an MEMS force sensor, and a wireless transmission module; the clip arms are connected to the clip handle through nickel-titanium alloy hinges; the neodymium iron boron magnets are embedded in the inner grooves of the nickel-titanium alloy clip arms; the MEMS force sensor is welded on a flexible PCB, the PCB fits the root of the clip arm, and is connected to the MCU in the clip handle through micro-wires; the wireless transmission module and the MCU are integrated in the sealed cavity of the clip handle, and the cavity is encapsulated by medical silicone.

3. The endoscopic surgical magnetic control intelligent traction system according to claim 2, characterized in that: The closing size of the in-vivo magnetic clip assembly is ≤8 mm, the opening angle limit of the clip arm is 60°, and a temperature adaptive compensation circuit is configured.

4. An endoscopic surgical magnetic control intelligent traction system according to claim 2, characterized in that: The tip of the nickel-titanium alloy clip arm is embedded with a ring-shaped neodymium iron boron magnet. An external pulsed magnetic field triggers the phase change of the nickel-titanium alloy, and the clip arm returns to its open state. When the external magnetic field is switched to the reverse magnetic field, the clip arm starts to close under the action of the magnetic torque.

5. The endoscopic surgical magnetic control intelligent traction system according to claim 1, characterized in that: The external magnetic force control device is an 8×8 electromagnetic array composed of Helmholtz coils, configured with a PID closed-loop control module, and supports a constant traction mode and a pulsed traction mode.

6. The endoscopic surgical magnetic control intelligent traction system according to claim 5, wherein: The control method of the PID closed-loop control module includes the following steps: a) Receive the actual traction force fed back by the MEMS force sensor; b) Calculate the deviation between the target force and the actual traction force; c) Adjust the current of the electromagnetic array according to the traction force deviation; d) Trigger the magnetic field cut-off protection when the actual traction force ≥3 N.

7. An endoscopic surgery magnetic control intelligent traction system according to claim 1, characterized in that: The navigation and positioning subsystem includes an in-vivo positioning tag, an external signal receiving and processing module, and a fusion positioning and calculation module, and emits a radio frequency signal through the in-vivo positioning tag; the external signal receiving and processing module calculates the time difference of arrival and phase difference of the signals; The fusion positioning and calculation module fuses the endoscopic optical image and the electromagnetic positioning data to generate an augmented reality interface with superimposed three-dimensional coordinates.

8. An endoscopic surgery traction control method using the traction system according to any one of claims 1-7, including the following steps: (1) Deploy the magnetic clip through the endoscopic channel; (2) Obtain the tissue depth d by CT scan and input it into the compensation algorithm; (3) Set the target traction force Ft, (4) Generate a gradient magnetic field through the electromagnetic array to traction the magnetic clip to move; (5) Monitor Fn in real time and dynamically adjust the magnetic field strength; (6) After reaching the predetermined peeling range, unlock and recover the clip body.

9. The endoscopic surgery traction control method according to claim 8, characterized in that: The endoscopic surgery traction control method of this embodiment includes the following steps: (1) Deploy magnetic clips through the endoscopic channel Under direct endoscopic vision, send the pushing catheter through the forceps channel to near the target mucosa; Rotate the pushing handle, and the buckle at the front end of the catheter releases the magnetic clip. The clip arms are triggered by body temperature to have a shape memory effect, but are still restricted by the lock in the closed state; Confirm the relative position of the clip body and the lesion through the endoscopic image; (2) Obtain the tissue depth d by CT scan and input it into the compensation algorithm Use thin-layer CT scanning to cover the target area and generate a three-dimensional model including the position of the clip body, the thickness of the mucosal layer, and the surrounding blood vessel distribution; Depth calculation: a. Algorithm input The shortest distance d from the clip body to the body surface Tissue density correction factor ε b. Compensation formula where k = 0.67, which is the calibration coefficient; Input d and ε into the central processing unit to generate the initial magnetic field parameter B, and synchronize it to the electromagnetic control device through Ethernet; (3) Set the target traction force Ft Set Ft through the slider or numeric keypad or voice recognition module, supporting 0.1N step adjustment; Select the traction mode; (4) Generate a gradient magnetic field through the electromagnetic array to pull the magnetic clip to move a. Magnetic field parameter calculation Three-dimensional vector decomposition: According to the position of the clip body provided by the navigation subsystem, calculate the target magnetic field gradient direction: Current of each coil unit in the 8×8 electromagnetic array ni is the normal vector of the coil The FPGA outputs a PWM waveform, which is amplified by the H-bridge and then drives the coil. Calibrate the magnetic field uniformity through the reference tag every 5 minutes; The clip body generates magnetic force under the action of the magnetic field gradient: Among them, χ = 1.2 is the magnetic susceptibility of neodymium iron boron, and V = 8mm 3 is the volume of the magnet; Realtime display the superposition of the traction vector and the endoscopic image, and the operator can fine-tune the direction through the foot pedal; (5) Monitor Fn in real time and dynamically adjust the magnetic field intensity The MEMS force sensor collects the strain at the root of the clip arm and transmits it to the central processing unit; PID closed-loop regulation, and the algorithm is executed as follows: Where I O is the input current value of the coil in the electromagnetic array at the current moment or the previous control period, I N is the updated current value to be applied at the next moment or the current control period after calculation by the PID algorithm; e = Ft - Fa, which is the traction force error; parameters Kp (proportional coefficient) = 0.8; Ki (integral coefficient) = 0.2; Kd (differential coefficient) = 0.05; Δt is the time difference between the previous control action and the current control action; The step response time < 100ms, and the steady-state error < ±0.1N; If Fa ≥ 2.8N, the system automatically reduces the gain by 50%; when the BLE signal loss > 500ms, switch to the preset safe magnetic field; (6) After reaching the predetermined stripping range, unlock and recover the clip body.

Images