Search

CN-121971171-A - Three-dimensional registration-based orthopedic surgery navigation positioning method and system

CN121971171ACN 121971171 ACN121971171 ACN 121971171ACN-121971171-A

Abstract

The invention discloses an orthopedic surgery navigation positioning method and system based on three-dimensional registration, and belongs to the technical field of orthopedic surgery navigation. The system corresponds to the method, and comprises an image acquisition module, a three-dimensional reconstruction module, a registration calculation module, a navigation guide module, an error monitoring module and a control module, wherein the modules work cooperatively, the navigation positioning accuracy is ensured to be less than or equal to 0.3mm, the registration time is less than or equal to 3.5s, and the intra-operation dynamic error compensation response time is less than or equal to 50ms. The invention does not need to rely on physical markers, does not need to manually intervene in the registration process by doctors, has strong anti-interference capability, is suitable for various orthopedic surgery scenes, effectively reduces surgery risks, and improves surgery efficiency and standardization level.

Inventors

  • CUI MANYI

Assignees

  • 淮安市淮安医院（淮安市肿瘤医院）

Dates

Publication Date
20260505
Application Date
20260403

Claims (10)

  1. 1. A three-dimensional registration-based bone surgery navigation positioning method and system are characterized by comprising the following steps: Step 1, acquiring CT images, MRI images and ultrasonic images of a patient operation area, wherein the thickness of a CT image scanning layer is set to be 0.1-0.2mm, the scanning resolution is 512 multiplied by 512 pixels, the MRI images adopt T1 weighting sequences, the scanning time is less than or equal to 8s, the ultrasonic image sampling frequency is 10-15Hz, the acquisition range covers the operation area and the peripheral 2-3cm tissues, preprocessing the acquired multi-mode images, sequentially performing gray scale normalization, noise filtering, edge enhancement and redundant information removing operation, wherein the noise filtering adopts an improved Gaussian filtering algorithm, the size of a filtering kernel is set to be 3 multiplied by 3, the edge enhancement adopts an edge detection algorithm based on gradient lifting, and soft tissue images and background noise irrelevant to the operation area are removed to obtain a preprocessed multi-mode image data set; Step 2, reconstructing a preoperative three-dimensional bone model, namely reconstructing a three-dimensional model of a bone of an operation area by fusing bone density information of CT images, soft tissue boundary information of MRI images and real-time morphological information of ultrasonic images based on a preprocessed multi-mode image dataset and adopting a multi-mode fusion reconstruction algorithm, defining the model as a preoperative reference model, setting the voxel size to be 0.05x0.05x0.05mm3 in the reconstruction process, optimizing the surface smoothness of the model by triangular surface patches fitting, controlling the number of the triangular surface patches to be 10000-15000, ensuring that the morphological errors of the model and the actual bone are less than or equal to 0.1mm, marking an operation target point, an operation path and a dangerous area on the preoperative reference model, and establishing a preoperative coordinate system Wherein As the geometric center of the bone in the operation area, The axis is parallel to the long axis of the bone, The axis is perpendicular to the coronal plane of the bone, The axis being perpendicular to A plane; Step 3, collecting intraoperative dynamic point cloud, namely collecting dynamic point cloud data of the bone surface of an operation area in real time by adopting structural light three-dimensional scanning equipment, setting the scanning frequency to 15-20Hz, setting the point cloud density to be more than or equal to 100 points/mm < 2>, avoiding interference of operation light and instrument reflection on a collection result in the collecting process by adopting a self-adaptive exposure adjustment mechanism, removing outliers corresponding to collected bone surface blood stains and soft tissue residues, removing the outliers by adopting a radius filtering algorithm, setting the filtering radius to be 0.1mm to obtain a purified intraoperative dynamic point cloud data set, and defining an intraoperative coordinate system Wherein Is the real-time geometric center of the bone in the operation, A shaft(s), A shaft(s), Axes respectively with preoperative coordinate system Is parallel to the coordinate axis of the frame; and 4, three-dimensional registration calculation, namely adopting an improved hybrid registration algorithm to realize the accurate registration of a preoperative reference model and an intraoperative dynamic point cloud data set, wherein the algorithm combines global registration and local registration, firstly completes global coarse registration through an improved SAC-IA algorithm, and then completes local fine registration through an improved ICP algorithm, and the specific process is as follows: 4.1. global coarse registration, namely extracting a bone characteristic point set from a preoperative reference model Extraction of bone feature point set from intraoperative dynamic point cloud data set Wherein 、 Are all positive integers, the total number of the two is equal to the positive integer, Calculating feature point set And (3) with Adopts the improved FPFH characteristic descriptor by introducing bone curvature weight factors Optimizing the discrimination of the feature descriptors, The calculation of (2) is shown in the formula (1): Wherein, the Is the first The curvature values of the individual feature points, Is a feature point set The maximum curvature value of all the feature points in (a), Obtaining an initial matching point pair based on similarity matching of feature descriptors, removing error matching point pairs through a random sampling consistency algorithm, setting an error matching point removing threshold value to be 0.2mm, and obtaining a coarse registration transformation matrix T 1 ; 4.2. local fine registration by coarse registration transformation matrix As an initial value, adopting an improved ICP algorithm, and introducing a dynamic weight factor Optimizing a registration error function, wherein the error function is shown in a formula (2): Wherein, the For the fine registration of the transformation matrix, In order to effectively match the pair number of points, Is the first The dynamic weighting factors of the pairs of matching points, Dynamically adjusting according to the curvature similarity of the matching point pairs, , As a feature point of the pre-operative reference model, As a feature point of the intra-operative dynamic point cloud, Iterative optimization of error function for Euclidean distance by gradient descent method The iteration termination condition is that the error variation of two adjacent iterations is less than or equal to 0.001mm, and the iteration times are controlled between 20 and 30 times, so as to obtain the final fine registration transformation matrix ; Step 5, navigation positioning parameter calculation based on the final fine registration transformation matrix Coordinate system of preoperative reference model Conversion to an intraoperative coordinate system Acquiring the position information of the surgical instrument in real time, and obtaining the real-time coordinate of the surgical instrument in the intraoperative coordinate system through coordinate conversion Calculating deviation value of surgical instrument and surgical path The deviation value is calculated as shown in formula (3): wherein, the method comprises the following steps of , , ) For the coordinates of any point on the surgical path, Directional vectors of the surgical path in the intraoperative coordinate system 、 、 A component on the axis; step 6, guiding and error compensating the real-time navigation, and real-time coordinates and deviation values of the surgical instrument And the operation path parameters are fed back to the navigation display module in real time, and when the deviation value is calculated When the length of the surgical instrument is more than 0.3mm, automatically triggering a dynamic error compensation mechanism, and adjusting the position guiding parameters of the surgical instrument until the deviation value Simultaneously, the micro deformation and the body position deviation of the bone in the operation are monitored in real time, the dynamic point cloud data in the operation is updated once every 0.5s, and the steps 4-5 are repeated, so that the dynamic update of registration and positioning parameters is realized, and the whole-course navigation positioning accuracy of the operation is ensured to meet the requirements; And 7, after the operation is finished, recording whole course navigation positioning data including registration accuracy, deviation value change, error compensation times and operation time consumption, generating an operation navigation report, and providing data support for postoperative disc recovery and follow-up operation optimization.
  2. 2. The orthopedic surgery navigation positioning method and system based on three-dimensional registration according to claim 1, wherein the system comprises an image acquisition module, a three-dimensional reconstruction module, a registration calculation module, a navigation guidance module, an error monitoring module, a control module and a data storage module, wherein the modules are connected through a high-speed data bus, the data transmission rate is more than or equal to 1000Mbps, the real-time performance and stability of data transmission are ensured, and the specific module structure is as follows: The image acquisition module comprises a CT scanner, an MRI scanner and ultrasonic acquisition equipment, wherein the scanning layer thickness of the CT scanner can be adjusted between 0.1mm and 0.2mm, the scanning resolution is not lower than 512 multiplied by 512 pixels, and the scanning speed is more than or equal to 10 layers/s, the MRI scanner adopts a T1 weighting sequence, the scanning time is less than or equal to 8s, the magnetic field intensity is 1.5T-3.0T, the sampling frequency of the ultrasonic acquisition equipment is 10-15Hz, the detection depth is 5-10cm, the image acquisition module has the self-adaptive exposure adjusting function and can automatically avoid interference of surgical light and instrument reflection, and is used for acquiring multi-mode image data of a patient operation area, transmitting the acquired image data to the three-dimensional reconstruction module in real time, and simultaneously carrying out preliminary format conversion on the image data acquired by different equipment to uniformly convert the image data acquired by different equipment into a DICOM format; The three-dimensional reconstruction module is connected with the image acquisition module and is used for receiving the preprocessed multi-mode image dataset transmitted by the image acquisition module, adopts a multi-mode fusion reconstruction algorithm, fuses bone density information of CT images, soft tissue boundary information of MRI images and real-time morphological information of ultrasonic images, reconstructs a three-dimensional preoperative reference model of bones of an operation area, is internally provided with a voxelization processing unit, a triangular patch fitting unit and a model optimizing unit, wherein the voxelization processing unit can be set to 0.05X0.05X0.05mm3, the triangular patch fitting unit can control the number of triangular patches to 10000-15000, the model optimizing unit is used for optimizing the surface smoothness of the model and ensuring that the morphological errors of the model and the actual bones are less than or equal to 0.1mm, and simultaneously has a preoperative marking function, can mark an operation target point, an operation path and a dangerous area on the preoperative reference model, and establishes a preoperative coordinate system Transmitting the pre-operation reference model and the marking information to a registration calculation module; The registration calculation module is respectively connected with the three-dimensional reconstruction module and the error monitoring module and is used for receiving the preoperative reference model and the marking information transmitted by the three-dimensional reconstruction module, receiving the intraoperative dynamic point cloud data set transmitted by the error monitoring module, adopting an improved hybrid registration algorithm to finish the accurate registration of the preoperative reference model and the intraoperative dynamic point cloud data set, and internally arranging a feature extraction unit, a global coarse registration unit, a local fine registration unit and a transformation matrix optimization unit in the module, wherein the feature extraction unit is used for extracting the feature point set of the preoperative reference model and the intraoperative dynamic point cloud and an improved FPFH feature descriptor, and the global coarse registration unit adopts an improved SAC-IA algorithm to finish coarse registration to obtain a coarse registration transformation matrix The local fine registration unit adopts an improved ICP algorithm, and the error function based on the formula (2) is subjected to iterative optimization to obtain a final fine registration transformation matrix The transformation matrix optimizing unit is used for carrying out smoothing treatment on the transformation matrix, eliminating noise interference and ensuring the stability of the transformation matrix; the registration calculation module calculates the final fine registration transformation matrix The registration error data are transmitted to a navigation guiding module and an error monitoring module; the navigation guiding module is connected with the registration calculating module and the control module and is used for receiving the fine registration transformation matrix T transmitted by the registration calculating module, converting a coordinate system of the preoperative reference model into an intraoperative coordinate system, calculating coordinate parameters of an operation target point and an operation path under the intraoperative coordinate system, collecting position information of the operation instrument in real time, obtaining real-time coordinates of the operation instrument through coordinate conversion, and calculating a deviation value delta of the operation instrument and the operation path based on a formula (3), wherein the navigation guiding module comprises a navigation display unit and an instrument guiding unit, the navigation display unit adopts a high-definition crystal display screen, the resolution is not lower than 1920×1080 pixels, and the preoperative reference model, the intraoperative dynamic point cloud, the position of the operation instrument and the deviation value can be displayed in real time The instrument guiding unit adopts electromagnetic navigation technology, the guiding precision is less than or equal to 0.3mm, the guiding parameters can be automatically adjusted according to the deviation value delta, and the instrument guiding unit can automatically adjust the guiding parameters when the deviation value is equal to or less than 0.1s When the distance is more than 0.3mm, sending out an acousto-optic prompt, outputting an adjusting instruction, and guiding the surgical instrument to be aligned to the surgical path; The error monitoring module is respectively connected with the image acquisition module, the registration calculation module and the navigation guide module and is used for monitoring the quality of the intra-operative dynamic point cloud acquisition in real time, eliminating outliers, ensuring the purity of the intra-operative dynamic point cloud data set, and simultaneously monitoring the registration precision and the deviation value of the surgical instrument in real time The module is internally provided with an outlier removing unit, an error calculating unit and a deformation monitoring unit, wherein the outlier removing unit adopts a radius filtering algorithm, the filtering radius is set to be 0.1mm, and the error calculating unit is used for calculating registration errors and deviation values The deformation monitoring unit is used for monitoring tiny deformation of bones in operation, and the deformation monitoring precision is less than or equal to 0.05mm; The control module is used as a core control unit of the system and is respectively connected with the image acquisition module, the three-dimensional reconstruction module, the registration calculation module, the navigation guide module, the error monitoring module and the data storage module, adopts an ARM Cortex-A9 processor, has a main frequency of more than or equal to 1.5GHz, has a memory of more than or equal to 4GB and is used for controlling the cooperative work of the modules, receives data transmitted by the modules, analyzes and processes the data and issues a control instruction; The data storage module is connected with the control module, adopts a solid state disk, has the storage capacity of more than or equal to 1TB, is used for storing preoperative multi-mode image data, preoperative reference model data, intraoperative dynamic point cloud data, registration parameters, navigation positioning data and operation navigation reports, supports data encryption storage, adopts an AES-256 encryption algorithm to prevent data leakage, supports a data export function, can export the stored data into a general format, and is convenient for postoperative duplication and data sharing.
  3. 3. The three-dimensional registration-based bone surgery navigation positioning method and system according to claim 1, wherein the gray scale normalization operation in step 1 maps the gray scale value of the image to the [0,255] interval by adopting a linear normalization algorithm, and the normalization formula is as follows: Wherein the method comprises the steps of Is the gray value of the original image, Is the minimum gray value of the original image, Is the maximum gray value of the original image, And the edge enhancement operation is based on a gradient lifting algorithm, calculates the gradient value of the image, adopts a Sobel operator for calculating the gradient value, has the operator size of 3 multiplied by 3, and highlights the bone edge contour and suppresses the background noise by enhancing the region with larger gradient value.
  4. 4. The three-dimensional registration-based bone surgery navigation positioning method and system according to claim 1, wherein the multi-mode fusion reconstruction algorithm in the step 2 adopts a weighted fusion strategy, the weight coefficient of a CT image is set to 0.6, the weight coefficient of an MRI image is set to 0.25, the weight coefficient of an ultrasound image is set to 0.15, and the fusion formula is as follows: Wherein the method comprises the steps of Is the voxel value of the CT image, For the voxel values of the MRI image, Is the voxel value of the ultrasound image, In the fusion process, a self-adaptive threshold segmentation algorithm is adopted to segment a bone region and a soft tissue region, the threshold range is set to be [150,255], the accurate extraction of the bone region is ensured, and the interference of the soft tissue on the reconstruction of the three-dimensional model is avoided.
  5. 5. The three-dimensional registration-based bone surgery navigation positioning method and system as claimed in claim 1, wherein the method is characterized in that: the structural light three-dimensional scanning equipment in the step 3 adopts a binocular structural light scanning principle, the resolution of two cameras is 2048 multiplied by 1536 pixels, the baseline distance is 100-150mm, the scanning range is 50 multiplied by 50mm3-200 multiplied by 200mm3, and the scanning precision is less than or equal to 0.05mm; the self-adaptive exposure adjustment mechanism automatically adjusts the exposure time of the camera according to the illumination intensity of an operation area, the exposure time range is 10-50ms, meanwhile, a polarization filtering technology is adopted to filter stray light generated by the light reflection of the operation light and the instrument, the acquired point cloud data are clear and accurate, the radius filtering algorithm eliminates the points with the distance larger than the filtering radius (0.1 mm) by calculating the distance between each point and surrounding points, the purity of an intraoperative dynamic point cloud data set is ensured, and the rejection rate is controlled between 5% and 10%.
  6. 6. The three-dimensional registration-based bone surgery navigation positioning method and system as claimed in claim 1, wherein the improved SAC-IA algorithm in step 4 is characterized in that a feature point curvature similarity screening mechanism is introduced on the basis of the traditional SAC-IA algorithm, feature point pairs with curvature similarity not less than 0.8 are screened out as initial matching point pairs, the number of false matching point pairs is reduced, meanwhile, the number of random sampling is optimized, the sampling number is set to 1000-1500 times, the sampling iteration termination condition is that the accuracy of the matching point pairs is not less than 95%, the precision and efficiency of coarse registration are ensured, the coarse registration error is not more than 0.5mm, the coarse registration time is not more than 1.5s, and the improved ICP algorithm is divided by introducing dynamic weight factors Besides, an iteration step length self-adaptive adjustment mechanism is adopted, the iteration step length is dynamically adjusted according to the error change quantity, when the error change quantity is large, the step length is set to be 0.01mm, and when the error change quantity is small, the step length is set to be 0.001mm, so that the precision and convergence speed of fine registration are further improved, the fine registration error is less than or equal to 0.1mm, and the fine registration time consumption is less than or equal to 2s.
  7. 7. The orthopedic surgery navigation positioning method and system based on three-dimensional registration according to claim 1, wherein the surgical instrument position acquisition in the step 5 is characterized in that an electromagnetic positioning sensor is adopted, the positioning precision is less than or equal to 0.05mm, the sampling frequency is 20-30Hz, the sensor is fixed at the tail end of the surgical instrument, the position coordinates of the surgical instrument are acquired in real time, the coordinate data are transmitted to a navigation guiding module in a wireless transmission mode, the coordinate conversion is performed, the coordinate of the surgical instrument under the sensor coordinate system is converted into the coordinate under the intraoperative coordinate system based on a fine registration transformation matrix T, and the conversion formula is that Wherein For the coordinates of the surgical instrument in the sensor coordinate system, Is the coordinate of the surgical instrument in the intraoperative coordinate system, the deviation value delta is set to be 0.3mm, when When the diameter is less than or equal to 0.3mm, determining that the surgical instrument is aligned with the surgical path when And triggering a dynamic error compensation mechanism when the length of the transmission line is greater than 0.3 mm.
  8. 8. The three-dimensional registration-based bone surgery navigation positioning method and system as claimed in claim 1, wherein the dynamic error compensation mechanism in the step 6 adopts a PID control algorithm, and is based on a deviation value The method comprises the steps of adjusting the guiding direction and speed of a surgical instrument, setting parameters of a PID control algorithm to be a proportionality coefficient Kp=5.0, an integral coefficient Ki=0.1, a differential coefficient Kd=0.5, a compensation response time less than or equal to 50ms and a compensation precision less than or equal to 0.05mm, dynamically updating the registration and positioning parameters, updating the intraoperative dynamic point cloud data once every 0.5s, re-executing registration calculation, updating a fine registration transformation matrix T and a deviation value delta of the surgical instrument, ensuring that the navigation positioning parameters can follow in real time when the intraoperative bone undergoes micro deformation or body position deviation, maintaining the positioning precision, and simultaneously, automatically starting a point cloud complement algorithm to complement missing point cloud data when a great amount of blood stains and soft tissues are blocked in an operation area, and complementing precision less than or equal to 0.1mm.
  9. 9. The orthopedic surgery navigation positioning method and system based on three-dimensional registration according to claim 2 are characterized in that a PCIe 4.0 interface is adopted by the high-speed data bus, the data transmission rate is more than or equal to 1000Mbps, multi-module parallel data transmission is supported, data transmission delay is avoided, the navigation display unit further has surgery progress display and abnormal alarm display functions besides displaying relevant navigation information in real time, when an offset value delta is more than 0.3mm or registration error is more than 0.1mm, an audible and visual alarm is sent out, alarm volume is adjustable, alarm lamplight is red and flashes to remind doctors to process timely, the instrument guide unit supports adaptation of various surgical instruments, including orthopedic dril, orthopedic hammers, screw insertion instruments and the like, the diameter range of the adaptive instruments is 1-10mm, and guide parameters can be automatically adjusted according to the sizes of different instruments.
  10. 10. The orthopedic surgery navigation positioning method and system based on three-dimensional registration according to claim 2, wherein the control module is internally provided with a fault diagnosis unit, can monitor the working state of each module in real time, give out fault alarm in time when one module breaks down, display fault position and fault type, and automatically switch to a standby working mode at the same time to ensure that surgery navigation is not interrupted, the data storage module supports a data timing backup function, the backup period can be set to be 1-5min, backup data is stored in an independent backup partition to prevent data loss, and simultaneously supports a data query function, corresponding navigation positioning data and surgery reports can be quickly queried according to keywords such as surgery time, patient information, surgery type and the like, and the query response time is less than or equal to 0.5s.

Description

Three-dimensional registration-based orthopedic surgery navigation positioning method and system Technical Field The invention belongs to the technical field of orthopedic surgery navigation, and particularly relates to an orthopedic surgery navigation positioning method and system based on three-dimensional registration, which are suitable for various orthopedic surgeries, including spinal surgery, joint replacement surgery, fracture reduction surgery and the like, and can be used for realizing the accurate positioning of surgical instruments and the accurate guiding of surgical paths. Background Accurate positioning of the bone surgery is a key for ensuring success of the surgery, traditional bone surgery mainly depends on clinical experience of doctors, and the position and the surgical path of the surgical instruments are judged through perspective images in the surgery, so that the problems of low positioning precision, large surgical wounds, more postoperative complications, strong dependence on experience of the doctors and the like exist. With the development of medical technology, the orthopedic surgery navigation system is gradually applied to clinic, and alignment of preoperative images and actual bones in surgery is realized through a three-dimensional registration technology, so that accurate navigation information is provided for doctors, and surgery precision and efficiency are effectively improved. At present, the existing navigation and positioning method and system for the orthopedic operation still have a plurality of technical defects, and the high-precision requirement of the clinical operation is difficult to meet. On one hand, the existing registration algorithm mostly adopts a single ICP algorithm or SAC-IA algorithm, has low registration precision and weak anti-interference capability, is easily influenced by factors such as tissue deformation, body position deviation, blood trace shielding and the like in operation, so that registration errors are large, accurate positioning cannot be realized, on the other hand, the existing system mostly relies on physical markers for registration, the implantation of the physical markers can increase wounds of patients, the markers are easily shifted in operation to influence the registration precision, meanwhile, the existing method lacks a real-time error monitoring and dynamic compensation mechanism, when deviation occurs in operation, operation risks are increased, in addition, the multimode image fusion effect of the existing system is poor, the advantages of different images cannot be fully utilized, so that the three-dimensional model reconstruction precision is insufficient, and the navigation positioning effect is further influenced. According to the retrieval, in the prior art, related patents such as Beijing Tianzhihuang medical science and technology Co-Ltd are focused on the directions of automatic screw setting of robots, tracking of acetabular joint replacement tools, intelligent planning of osteotomy planes and the like, the problems that a registration mode still depends on physical markers, a single registration algorithm is achieved, dynamic error compensation is avoided and the like exist, the existing point cloud registration method based on an attention mechanism is mainly optimized for public data sets, clinical scenes of orthopedic surgery are not combined, the interference resistance is insufficient, full-flow automation of registration and positioning is not achieved, the existing image registration technology is judged by adopting fixed threshold values, dynamic adjustment cannot be achieved according to actual conditions in operation, and the registration accuracy and robustness are to be improved. Therefore, the orthopedic surgery navigation positioning method and system based on three-dimensional registration, which do not need to rely on physical markers, have high registration precision and strong anti-interference capability, have a real-time error monitoring and dynamic compensation mechanism, and can avoid the defects of the prior art, become the technical problem to be solved in the technical field of current orthopedic surgery navigation. Disclosure of Invention Aiming at the technical defects of low registration precision, weak anti-interference capability, dependence on physical markers, lack of a real-time error compensation mechanism, poor multi-mode image fusion effect and the like of the traditional orthopedic operation navigation positioning method and system, the invention simultaneously avoids the design ideas of single registration algorithm, fixed threshold judgment, marker dependence and the like in the prior art, provides the orthopedic operation navigation positioning method and system based on three-dimensional registration, realizes high-precision, high-stability and full-process automatic orthopedic operation navigation positioning, reduces operation risks, improves operation efficiency and s