Noninvasive pelvic bone tracking system for hip joint surgery

Abstract

The invention discloses a non-invasive pelvic bone tracking system for hip joint surgery, and belongs to the technical field of intelligent surgical navigation and medical imaging. A non-invasive pelvic bone tracking system for hip joint surgery comprises a multi-modal sensing module, a dynamic resolution optimization module, an AI correction module and a multi-region parallel tracking engine, and the multi-modal sensing module comprises a 3D ultrasonic unit, a near-infrared optical unit and an inertial measurement unit. The problems of positioning error and limitation of scanning speed and coverage range caused by fat interference in the prior art are solved. According to the method, low-frequency ultrasonic, near-infrared optical and IMU data are fused, scanning modes are intelligently switched based on fat thickness, the speed and precision are balanced, ultrasonic skeleton features and optical body surface topology are jointly analyzed, fat attenuation dynamic compensation is achieved, multi-region GPU parallel tracking supports intra-operative real-time multi-target collaborative navigation is achieved, and multi-target collaborative navigation is achieved. Millimeter-level real-time tracking of obese patients during operation is realized.

Description

Technical Field

[0001] The present invention relates to the field of intelligent surgical navigation and medical imaging technology, and specifically to a non-invasive pelvic tracking system for hip joint surgery. Background Art

[0002] With the increasing demand for precision in hip joint replacement surgery, non-invasive pelvic tracking technology has gradually replaced the traditional invasive bone nail marking method. However, the existing technology still has the following significant defects:

[0003] I. Positioning error caused by fat interference

[0004] The current mainstream non-invasive tracking systems mainly rely on optical navigation or electromagnetic tracking technology. For patients with a subcutaneous fat layer thickness ≥ 3 cm (commonly found in obese or specific body posture populations), the optical signal attenuates due to scattering by adipose tissue, resulting in a distorted mapping relationship between the body surface landmark points and the deep bones; the electromagnetic tracking system is affected by the difference in fat conductivity and is prone to magnetic field distortion. Clinical data shows that the intraoperative bone positioning error of such patients is generally > 3 mm, seriously affecting the control of the acetabular cup implantation angle (the target error needs to be < 1°), and even causing complications such as postoperative joint dislocation and prosthesis loosening.

[0005] II. Limitations of scanning speed and coverage

[0006] Although the existing image enhancement technologies (such as intraoperative 3D ultrasound or CT) can provide bone structure information, there are serious efficiency bottlenecks: Conventional 3D ultrasound takes more than 30 seconds to complete high-precision scanning of a single area and cannot track intraoperative dynamic displacements in real time (such as the slight bone movements caused by pelvic rotation or instrument operation); due to hardware computing power limitations, traditional systems can only continuously scan a single area such as the acetabulum or anterior superior iliac spine, and cannot synchronously obtain the associated motion data of key pelvic anatomical landmark points. For example, during the process of acetabular reaming, the surgeon needs to alternately observe the depth of the acetabulum and the position of the pubic symphysis, and the existing system causes data tomograms due to insufficient refresh rate (usually ≤ 2 Hz), forcing the surgery to pause and wait for the scanning results, extending the surgery time by more than 40%. The above defects jointly restrict the clinical applicability of the non-invasive pelvic tracking system. Especially in complex cases (such as obesity and pelvic deformity), the existing technology is difficult to meet the core requirements of real-time performance, accuracy, and multi-target tracking. Summary of the Invention

[0007] The object of the present invention is to provide a non-invasive pelvic tracking system for hip joint surgery. By fusing low-frequency ultrasound, near-infrared optics and IMU data, it intelligently switches the scanning mode based on fat thickness to balance speed and accuracy. Through the joint analysis of ultrasonic bone features and optical body surface topology, it realizes dynamic compensation for fat attenuation. The multi-region GPU parallel tracking supports intraoperative real-time multi-target collaborative navigation, achieving millimeter-level real-time tracking of obese patients during surgery, and solving the problems raised in the above-mentioned background technology.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A non-invasive pelvic tracking system for hip joint surgery, including a multi-modal sensing module, a dynamic resolution optimization module, an AI correction module and a multi-region parallel tracking engine. Among them, the multi-modal sensing module includes a 3D ultrasound unit, a near-infrared optical unit and an inertial measurement unit (IMU);

[0010] The 3D ultrasound unit is configured to emit 5-10 MHz ultrasonic waves to penetrate the fat layer to obtain deep bone structure data; mark the anatomical landmark points of the acetabulum, anterior superior iliac spine and pubic symphysis, and establish a three-dimensional pelvic model of the patient;

[0011] The near-infrared optical unit is configured to capture the contour information of the body surface landmarks;

[0012] The inertial measurement unit is configured to monitor the patient's body position movement through a 9-axis sensor;

[0013] The dynamic resolution optimization module is configured to adaptively switch the scanning mode according to the fat layer thickness measured before surgery:

[0014] When the fat layer thickness < 3 cm, the fast mode is enabled, using a 32x32x32 resolution, and the scanning speed is increased to 10 seconds / region;

[0015] When the fat layer thickness ≥ 3 cm, the high-precision mode is enabled, using a 64x64x64 resolution, the resolution is doubled and the accuracy is < 1 mm after error compensation;

[0016] The AI correction module is configured to be a fat layer signal attenuation compensation model based on deep learning, and generate attenuation compensation parameters through multi-modal registration training of preoperative CT data and intraoperative ultrasound data;

[0017] The multi-region parallel tracking engine is configured to synchronously process the tracking data of the acetabulum, anterior superior iliac spine and pubic symphysis through a GPU acceleration algorithm, achieving a data refresh rate of 5 Hz.

[0018] Preferably, the multi-modal sensing module further includes:

[0019] An ultrasonic signal preprocessing unit, configured to perform adaptive filtering on the original data of the deep bone structure collected by the 3D ultrasonic unit to eliminate motion artifacts;

[0020] An optical-IMU fusion unit, configured to perform Kalman filter fusion on the body surface contour changes and the IMU pose data to generate a real-time body position offset correction amount.

[0021] Preferably, the working process of the adaptive switching scanning mode includes:

[0022] Scanning and measuring the fat layer thickness in the area of the anterior superior iliac spine through the 3D ultrasonic unit;

[0023] Selecting resolution parameters based on the scanning results of the fat layer thickness;

[0024] Synchronously activating the corresponding compensation coefficients of the AI correction module during mode switching.

[0025] Preferably, the deep learning model of the AI correction module processes the ultrasonic original signal and the near-infrared optical signal, extracts the bone edge features through a 3D convolutional layer, analyzes the signal attenuation mode caused by the fat layer, extracts the topological relationship of the body surface landmark points through a convolutional network, and analyzes the correlation between the body surface deformation and the bone displacement;

[0026] Combining the outputs of the two parts with the pose data and time stamps output by the IMU unit, establishing a mapping relationship between the data through a cross-attention mechanism, and performing spatio-temporal alignment, outputting the influence weight and compensation coefficient of the fat layer deformation on the bone positioning, and finally generating the three-dimensional coordinates of the bone after error compensation.

[0027] Preferably, the multi-region parallel tracking engine realizes synchronous processing in the following ways:

[0028] Allocating independent GPU independent computing threads for each tracking region;

[0029] Adopting an asynchronous data flow pipeline to transmit the sensing data of each region;

[0030] Processing the ultrasonic data, optical data and IMU data in regions;

[0031] Stitching the data of each region into a complete pelvic dynamic model.

[0032] Preferably, the system further includes a real-time navigation module, configured to match the bone data collected during the operation with the preoperative three-dimensional model, and generate a real-time navigation guidance for the surgical instrument through the fusion of the optical contour, IMU pose and ultrasonic bone data;

[0033] Preferably, the real-time navigation module further includes:

[0034] Install an optical reflection sphere and an IMU detection device on the surgical instrument to obtain the position of the instrument tip in the three-dimensional bone coordinate system, as well as the pitch angle and deviation distance of the instrument in real time;

[0035] Map it to the CT model coordinates through transformation, calculate the deviation distance between the instrument tip and the planned path, and start the self-calibration program when the deviation is greater than 2 mm;

[0036] Calculate the deviation angle between the actual posture and the ideal posture of the instrument. When the instrument contacts the bone, the IMU detection device is triggered to compare the contact point coordinates with the navigation target coordinates, and visual path guidance is performed through the display.

[0037] Preferably, the self-calibration program includes:

[0038] Recalculate the mapping relationship between the fat layer thickness distribution map and the three-dimensional pelvic model;

[0039] Update the compensation coefficient of the AI correction module to the current surgical stage parameters.

[0040] Preferably, the system further includes a data synchronization management module, configured to:

[0041] Add millisecond-level timestamps to the ultrasound, optical, and IMU data streams;

[0042] Align the time series of multi-modal data through a sliding window caching mechanism;

[0043] Preferably, the timestamp error is controlled within 1 ms.

[0044] Compared with the prior art, the beneficial effects of the present invention are:

[0045] 1. The present invention directly penetrates the fat layer through 5-10 MHz low-frequency ultrasound to obtain bone data. The AI correction module extracts bone edge features through a 3D convolutional layer, analyzes the signal attenuation pattern caused by the fat layer, and dynamically analyzes the attenuation law of the fat layer deformation on the ultrasound signal and the associated influence on the body surface contour; the cross-attention mechanism calculates the contribution weights of the ultrasound and optical data in real time, eliminates the dependence on a single sensor, and effectively solves the problem of fat interference error.

[0046] 2. The present invention intelligently switches the scanning mode according to the fat thickness through the dynamic resolution optimization module, combined with the GPU asynchronous data stream processing of the multi-region parallel tracking engine, shortening the single-region scanning time from 30 seconds of traditional ultrasound to less than 10 seconds, and increasing the data refresh rate of the synchronous tracking of the three regions of the acetabulum, anterior superior iliac spine, and pubic symphysis to more than 8 Hz, meeting the real-time navigation requirements during the operation; through correlation analysis, the movement laws of multiple regions of the pelvis are calculated in real time, realizing multi-target collaborative tracking during the operation, effectively shortening the response delay of the surgical instrument path correction, and thus breaking through the limitations of the scanning speed and coverage range.

[0047] 3. From preoperative fat layer mapping to intraoperative dynamic verification, the system autonomously updates compensation parameters to avoid the static limitations of traditional devices that rely on preoperative calibration data. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a schematic diagram of the system working process of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0049] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0050] To solve the problems of positioning errors caused by fat interference and the limitations of scanning speed and coverage in the prior art, please refer to Figure 1 , the following technical solutions are provided in this embodiment:

[0051] A non-invasive pelvic tracking system for hip surgery, the system includes a multi-modal sensing module, a dynamic resolution optimization module, an AI correction module, and a multi-region parallel tracking engine. Among them, the multi-modal sensing module includes a 3D ultrasound unit, a near-infrared optical unit, and an inertial measurement unit (IMU);

[0052] Perform CT detection tomography on the pelvic position of the patient before surgery, generate a three-dimensional model of the pelvic structure through computer processing, and mark the skeletal landmark points of the acetabular center, the vertex of the anterior superior iliac spine, and the midpoint of the pubic symphysis;

[0053] Before surgery, a multi-modal sensing module is set above the patient's body surface. The multi-modal sensing module uses a 3D ultrasound probe, a near-infrared optical camera, and an IMU sensor, and ensures that the 3D ultrasound probe, the near-infrared optical camera, and the IMU sensor cover the pelvic target area;

[0054] The ultrasound provides bone depth data. The bone signal is directly obtained by the ultrasound penetrating the fat layer, avoiding the positioning drift caused by the sliding of the surface marker due to fat. The optical unit captures the body surface deformation, and the IMU monitors the overall body position shift. The three are fused to eliminate the limitations of a single sensor, realize multi-modal data complementarity, and solve the problems that traditional optical navigation is easily interfered by fat layer deformation, has large bone positioning errors, and single-modal data is easily affected by motion artifacts. When an abnormality occurs in a certain sensor (such as the optical unit fails due to blood contamination), the IMU and ultrasound can still maintain the basic tracking function to ensure the continuity of the surgery.

[0055] The intraoperative tracking process needs to go through three sub - processes: data acquisition, data processing, and real - time navigation.

[0056] Data acquisition:

[0057] First, pre - scan the ultrasonic waves to collect fat thickness data and real - time pelvic bone data. The specific process is as follows:

[0058] Emit low - frequency ultrasonic waves of 5 - 10 MHz to the pelvis, penetrate the skin and fat layer to reach the pelvic bone and receive the echo signal. Calculate the penetration distance from the skin surface to the bone using the time difference of the ultrasonic echo signal, and continuously obtain real - time data of the deep bone structure;

[0059] Receive ultrasonic echo signal data, perform real - time processing on the ultrasonic image, identify the skin layer, fat layer, and muscle layer, extract the characteristics of the skin - fat connection interface and the fat - muscle connection interface, identify the fat layer thickness, and automatically generate a fat layer thickness distribution map, mark the fat landmark points corresponding to the center of the acetabulum, the vertex of the anterior superior iliac spine, and the mid - point of the symphysis pubis, and complete the identification of the fat layer interface and the detection of the fat layer thickness;

[0060] Align the bone landmark points and fat landmark points, map the fat layer thickness distribution map to the outside of the three - dimensional model of the pelvic structure, and lock the positional relationship between the fat layer and the pelvic structure;

[0061] A thicker fat layer requires higher resolution to resolve bone details, but high resolution will reduce the scanning speed. To balance the scanning speed and accuracy requirements, an adaptive switching scanning mode is set here. Select the resolution parameter according to the intraoperative fat layer thickness scanning result. Specifically:

[0062] When the fat layer thickness < 3 cm, enable the fast scanning mode, use a resolution of 32x32x32, and the scanning speed is increased to 10 seconds / area, giving priority to improving the data acquisition speed to meet the intraoperative fast positioning requirements;

[0063] When the fat layer thickness ≥ 3 cm, enable the high - precision mode, use a resolution of 64x64x64, the resolution is doubled, and the accuracy is < 1 mm after error compensation, enhancing the ability to resolve bone details and solving the problem of ultrasonic signal attenuation caused by the thick fat layer;

[0064] It can be seen that enabling the fast mode for patients with a low - fat layer can avoid unnecessary precision redundancy, and specifically improve the resolution for patients with a high - fat layer to avoid repeated scanning.

[0065] While the low - frequency ultrasonic waves complete the identification of the fat layer interface and thickness detection, simultaneously collect the real - time dynamic data of the pelvic bone position;

[0066] The near-infrared optical unit captures the contour information and contour changes of body surface landmarks, and generates sub-millimeter coordinates of anatomical landmark points; the inertial measurement unit monitors the patient's body position movement through a 9-axis sensor;

[0067] The multi-modal sensing module further includes:

[0068] An ultrasonic signal preprocessing unit, configured to perform adaptive filtering on the raw data of the deep bone structure collected by the 3D ultrasonic unit, eliminate motion artifacts, and improve measurement accuracy;

[0069] An optical-IMU fusion unit, configured to perform Kalman filter fusion on the body surface contour changes captured by the near-infrared optical unit and the pose data collected by the IMU unit, and generate a real-time body position offset correction amount.

[0070] Data processing:

[0071] The AI correction module fuses multi-modal data collected by the 3D ultrasonic probe, near-infrared optical camera, and IMU sensor to correct the fat interference error; forms a fat layer signal attenuation compensation model through deep learning, jointly analyzes the ultrasonic signal and body surface deformation, generates bone position compensation parameters, and eliminates the influence of the fat layer on the imaging accuracy.

[0072] Specifically, the deep learning model of the AI correction module processes the ultrasonic raw signal and near-infrared optical signal, extracts the bone edge features through a 3D convolutional layer, analyzes the signal attenuation pattern caused by the fat layer, extracts the topological relationship of body surface landmark points through a convolutional network, analyzes the correlation between body surface deformation and bone displacement, and finally combines the outputs of the two parts with the pose data and timestamp output by the IMU unit, establishes a mapping relationship between data through a cross-attention mechanism, and performs spatio-temporal alignment, outputs the influence weight and compensation coefficient of fat layer deformation on bone positioning, and finally generates the three-dimensional coordinates of the bone after error compensation.

[0073] For patients with thick fat layers, the AI correction module can compensate for the 3-5mm original error caused by acoustic wave attenuation, significantly improve the positioning accuracy, and meet the intraoperative tracking requirements.

[0074] Real-time navigation:

[0075] After matching the bone data collected during the operation with the preoperative three-dimensional model, generate real-time navigation guidance for the surgical instrument through the fusion of optical contour, IMU pose, and ultrasonic bone data;

[0076] Specifically, an optical reflector ball and an IMU detection device are installed on the surgical instrument to obtain the position of the instrument tip in the three-dimensional coordinate system of the bone, as well as the pitch angle and deviation distance of the instrument in real time. The deviation distance between the instrument tip and the planned path is calculated by transforming and mapping it into the CT model coordinates. When the deviation is greater than 2 mm, the self-calibration program is started, and the deviation angle between the actual posture and the ideal posture of the instrument is calculated. When the instrument contacts the bone, the IMU detection device is triggered to compare the contact point coordinates with the navigation target coordinates, and the visual path guidance is carried out through the display;

[0077] Among them, the self-calibration program includes recalculating the mapping relationship between the fat layer thickness distribution map and the three-dimensional pelvic model and updating the compensation coefficient of the AI correction module to the current surgical stage parameters.

[0078] In order to ensure that the positions of all parts can be accurately synchronized and the navigation information is not stuck or delayed, the system is set with a multi-region parallel tracking engine to synchronously process the data of each part;

[0079] Since the traditional CPU serial processing cannot meet the refresh rate requirements of multiple regions ≥ 5 Hz, and the pelvis is a rigid connection structure, single-region tracking may miss the overall displacement. It is necessary to quickly integrate the data of multiple regions through parallel computing. To solve the above problems, the system is set with a multi-region parallel tracking engine, which is configured to process the tracking data of the acetabulum, anterior superior iliac spine and symphysis pubis in regions through GPU acceleration algorithms, that is, the pelvis is divided into several key tracking regions, and each region is assigned a dedicated GPU independent computing thread. The sensing data of each region is transmitted through an asynchronous data flow pipeline to avoid data blocking. The ultrasonic data, optical data and IMU data are processed in regions. After the data processing of each region is completed, the multi-region parallel tracking engine stitches the data into a complete dynamic pelvic model and refreshes the navigation information at a frequency of more than 5 times per second.

[0080] Through GPU parallel acceleration, the multi-region data processing delay is < 120 ms, and through cross-region motion correlation analysis, the risk of joint dislocation caused by body position offset can be pre-warned.

[0081] The system also includes a data synchronization management module, which is configured to: add millisecond-level timestamps to the ultrasonic, optical and IMU data streams, and control the timestamp error within 1 ms; align the time series of multi-modal data through a sliding window caching mechanism.

[0082] Working principle: The 3D ultrasound unit emits low-frequency ultrasonic waves of 5-10 MHz to penetrate the fat layer, calculates the bone depth using the echo time difference, and generates real-time three-dimensional coordinates of key points such as the acetabulum and anterior superior iliac spine. The near-infrared optical unit captures the contours of body surface landmark points, constructs a body surface deformation model through sub-millimeter-level optical positioning. The IMU sensor monitors the overall body position offset of the patient and corrects the global coordinate system in real time through 9-axis data. Before surgery, the fat layer thickness is measured by ultrasonic scanning and the scanning mode is switched. The AI correction module performs multi-modal data fusion, separately processes the ultrasonic bone edge features and the optical body surface topological relationship. The cross-attention mechanism dynamically allocates the fusion weights of ultrasonic and optical data, combines the IMU pose data for spatio-temporal alignment, and outputs the bone coordinates after error compensation. The GPU acceleration engine assigns independent thread blocks to the acetabulum, anterior superior iliac spine, and pubic symphysis, realizes the synchronous processing of three-region data through an asynchronous data flow pipeline, obtains true three-dimensional bone data by ultrasonic penetration of the fat layer, combines dynamic resolution switching and GPU parallel computing, breaks through the accuracy and speed bottlenecks of traditional optical / electromagnetic navigation, and realizes millimeter-level real-time tracking during surgery for obese patients.

[0083] It should be noted that in this article, relational terms such as first and second are only used 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 "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or elements inherent to such process, method, article or device.

[0084] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention.

Claims

1. A non-invasive pelvic tracking system for hip joint surgery, characterized in that, It includes a multi-modal sensing module, a dynamic resolution optimization module, an AI correction module, and a multi-region parallel tracking engine. Among them, the multi-modal sensing module includes a 3D ultrasound unit, a near-infrared optical unit, and an inertial measurement unit; The 3D ultrasound unit is configured to emit 5-10 MHz ultrasonic waves to penetrate the fat layer to obtain deep bone structure data; mark the anatomical landmark points of the acetabulum, anterior superior iliac spine, and symphysis pubis, and establish a three-dimensional pelvic model of the patient; The near-infrared optical unit is configured to capture the contour information of the body surface landmarks; The inertial measurement unit is configured to monitor the patient's body position movement through a 9-axis sensor; The dynamic resolution optimization module is configured to adaptively switch the scanning mode according to the fat layer thickness measured before surgery: When the fat layer thickness < 3 cm, the fast mode is enabled, using a 32x32x32 resolution, and the scanning speed is increased to 10 seconds / region; When the fat layer thickness ≥ 3 cm, the high-precision mode is enabled, using a 64x64x64 resolution, the resolution is doubled, and the accuracy is < 1 mm after error compensation; The AI correction module is configured to generate attenuation compensation parameters through multi-modal registration training of preoperative CT data and intraoperative ultrasound data based on a deep learning-based fat layer signal attenuation compensation model; The multi-region parallel tracking engine is configured to synchronously process the tracking data of the acetabulum, anterior superior iliac spine, and symphysis pubis through a GPU acceleration algorithm to achieve a data refresh rate of 5 Hz.

2. The non-invasive pelvic tracking system for hip joint surgery according to claim 1, characterized in that, The multi-modal sensing module further includes: An ultrasonic signal preprocessing unit configured to perform adaptive filtering on the original deep bone structure data collected by the 3D ultrasound unit to eliminate motion artifacts; An optical-IMU fusion unit configured to perform Kalman filter fusion on the body surface contour changes and IMU pose data to generate a real-time body position offset correction amount.

3. The non-invasive pelvic tracking system for hip joint surgery according to claim 1, characterized in that, The workflow of the adaptive switching scanning mode includes: Scanning and measuring the fat layer thickness in the anterior superior iliac spine area through the 3D ultrasound unit; Selecting resolution parameters based on the fat layer thickness scanning result; Synchronously activating the corresponding compensation coefficient of the AI correction module during mode switching.

4. The non-invasive pelvic tracking system for hip joint surgery according to claim 1, wherein The deep learning model of the AI correction module processes the ultrasonic original signal and the near-infrared optical signal, extracts bone edge features through a 3D convolutional layer, analyzes the signal attenuation mode caused by the fat layer, extracts the topological relationship of the body surface landmark points through a convolutional network, and analyzes the correlation between body surface deformation and bone displacement; Combining the outputs of the two parts with the pose data and timestamp output by the IMU unit, establishing a mapping relationship between the data through a cross-attention mechanism, and performing spatio-temporal alignment, outputting the influence weight and compensation coefficient of fat layer deformation on bone positioning, and finally generating the error-compensated three-dimensional coordinates of the bone.

5. The non-invasive pelvic tracking system for hip joint surgery according to claim 1, characterized in that, The multi-region parallel tracking engine achieves synchronous processing in the following ways: Allocating independent GPU independent computing threads for each tracking region; Adopting an asynchronous data flow pipeline to transmit the sensing data of each region; Processing the ultrasound data, optical data, and IMU data in regions; Stitching the data of each region into a complete pelvic dynamic model.

6. The non-invasive pelvic tracking system for hip joint surgery according to claim 1, characterized in that, It further includes a real-time navigation module configured to generate real-time navigation guidance for the surgical instrument through the fusion of intraoperative acquired bone data and preoperative three-dimensional model, optical contour, IMU pose, and ultrasonic bone data after matching.

7. The non-invasive pelvic tracking system for hip joint surgery according to claim 6, characterized in that, The real-time navigation module includes: Install an optical reflector ball and an IMU detection device on the surgical instrument to obtain the position of the instrument tip in the bone three-dimensional coordinate system, as well as the pitch angle and deviation distance of the instrument in real time; Map it to the CT model coordinates through transformation, calculate the deviation distance between the instrument tip and the planned path, and start the self-calibration program when the deviation is greater than 2 mm; Calculate the deviation angle between the actual pose and the ideal pose of the instrument. When the instrument contacts the bone, the IMU detection device is triggered to compare the contact point coordinates with the navigation target coordinates, and visual path guidance is performed through the display.

8. The non-invasive pelvic tracking system for hip joint surgery according to claim 7, wherein The self-calibration program includes: Recalculate the mapping relationship between the fat layer thickness distribution map and the pelvic three-dimensional model; Update the compensation coefficient of the AI correction module to the current surgical stage parameters.

9. The non-invasive pelvic tracking system for hip joint surgery according to claim 1, characterized in that, It further includes a data synchronization management module configured to: Add millisecond-level timestamps to the ultrasonic, optical, and IMU data streams; Align the time series of multimodal data through a sliding window caching mechanism.

10. The non-invasive pelvic tracking system for hip joint surgery according to claim 9, characterized in that, The timestamp error is controlled within 1 ms.

Images