Auxiliary operation positioning system based on AI intelligent navigation

Abstract

The invention discloses an auxiliary operation positioning system based on AI intelligent navigation, and the system comprises an image data collection module which is used for obtaining the original image data and operation planning information of an operation part of a patient; the navigation model construction module is used for constructing a three-dimensional navigation model based on the original image data and generating a target key point and a target path through a preset AI model based on the operation planning information; the positioning and tracking module is used for acquiring the position information of the surgical instrument in real time in the surgical process and mapping the position information into the three-dimensional navigation model; and the real-time positioning correction module is used for acquiring continuous image frames acquired by the surgical instrument in real time, acquiring intraoperative motion deviation through optical flow estimation based on real-time image data, and correcting the target key point and the target path based on the intraoperative motion deviation. According to the method, the intraoperative motion deviation is obtained in real time through optical flow estimation, target positioning is corrected, and the accuracy and reliability of operation positioning can be improved through the real-time dynamic correction mechanism.

Description

Technical Field

[0001] This application relates to the field of visual positioning technology, and in particular to a surgical-assisted positioning system based on AI intelligent navigation. Background Technology

[0002] AI-based intelligent navigation surgical positioning systems utilize artificial intelligence and navigation technologies to assist in surgical positioning before surgery, reducing human error and improving surgical accuracy. This is especially beneficial for surgeons with limited experience, significantly improving surgical efficiency and safety.

[0003] However, during surgery, the patient's physical condition may undergo slight changes, such as tissue displacement. In order to ensure the accuracy of surgical positioning, existing AI-based intelligent navigation-assisted surgical positioning systems usually repeatedly collect image data of the patient's surgical site during the operation and regenerate or adjust the three-dimensional model constructed before the operation, and then reposition. As a result, due to the repeated calculations involved globally, delays can easily occur, which will affect the execution of the operation. Summary of the Invention

[0004] The purpose of this application is to provide a surgical-assisted positioning system based on AI intelligent navigation. By using optical flow estimation, the real-time positioning calculation is focused on key local motion information, avoiding the time-consuming process of real-time 3D reconstruction. This can effectively solve the latency problem caused by using AI intelligent navigation for precise surgical positioning.

[0005] In a first aspect, this application provides a surgical assistance positioning system based on AI intelligent navigation, comprising: The image data acquisition module is used to acquire raw image data of the patient's surgical site and surgical planning information; The navigation model construction module is used to build a 3D navigation model based on the original image data, and generate target key points and target paths based on surgical planning information through a preset AI model; The positioning and tracking module is used to acquire the position information of surgical instruments in real time during the operation and map the position information into the three-dimensional navigation model; The real-time positioning correction module is used to acquire continuous image frames from surgical instruments in real time, and based on the real-time image data, it obtains intraoperative motion deviation through optical flow estimation, and corrects the target key points and target path based on the intraoperative motion deviation.

[0006] The above technical solution uses optical flow estimation to obtain intraoperative motion deviations in real time and correct the target positioning. This dynamic navigation mechanism has higher accuracy and reliability compared to traditional static navigation. Furthermore, by using optical flow estimation for dynamic correction, it is not necessary to repeatedly collect image data of the patient's site during the operation and to update and adjust the three-dimensional navigation model in real time, which greatly reduces the time delay caused by dynamic positioning correction.

[0007] Optionally, the navigation model construction module includes: The image preprocessing unit is used to preprocess the raw image data to generate standardized image data; The 3D navigation construction unit is used to generate a 3D navigation model based on standardized image data and a preset 3D reconstruction method. The key point generation unit is used to generate target key points for the 3D navigation model based on surgical planning information; The path generation unit is used to generate a target path based on the target key points and 3D navigation information.

[0008] Optionally, the positioning and tracking module includes: The spatial mapping unit is used to construct a mapping model from the patient space to the three-dimensional navigation model space, denoted as the spatial mapping model; The spatial positioning unit is used to acquire the spatial pose information of surgical instruments in real time during the operation. The coordinate mapping unit is used to map the spatial pose information of surgical instruments into the three-dimensional navigation model through the spatial mapping model.

[0009] Optionally, the positioning and tracking module further includes: The mapping model verification unit is used to verify the spatial mapping model and generate the model verification results. The mapping model correction unit is used to perform compensation corrections based on the model validation results using a preset method.

[0010] Optionally, the real-time positioning correction module includes: The image frame acquisition unit is used to acquire continuous image frames acquired by surgical instruments in real time. Motion offset unit, used to obtain intraoperative motion deviation based on optical flow estimation from consecutive image frames; The positioning correction unit is used to correct the target key points and target path based on the intraoperative motion deviation when the intraoperative motion deviation reaches the preset correction threshold.

[0011] Optionally, the surgical planning information includes the surgical type, and the system further includes: The individual association module is used to construct a physiological dynamic impact model based on the surgical type and target key points. During the surgery, the module monitors the associated physiological indicators based on the physiological dynamic impact model to obtain the reference offset.

[0012] Optionally, the individual association module includes: The physiological indicator determination unit is used to determine the relevant physiological indicators based on the type of surgery, using a pre-set surgical type knowledge base. The physiological data sampling unit is used to collect the patient's related physiological index data and the location data of the target key points simultaneously, based on the related physiological indicators, to ensure that the timestamps of the data are completely aligned, so as to form the patient's individual physiological data; The physiological dynamic model generation unit is used to construct a physiological dynamic impact model based on individual patient physiological data. The physiological motion monitoring unit is used to monitor relevant physiological indicators during surgery and, based on the physiological dynamic influence model, obtain the motion offset of key points, which is denoted as the reference offset.

[0013] Optionally, the system further includes: The data storage management module is used to store the patient's original image data of the surgical site and the positioning data during the operation, and to store the data in a preset auxiliary positioning database.

[0014] Secondly, this application provides a surgical-assisted positioning method based on AI intelligent navigation, comprising the following steps: Obtain raw imaging data of the patient's surgical site and surgical planning information; Based on the original image data, a 3D navigation model is constructed, and based on the surgical planning information, target key points and target paths are generated through a preset AI model; During the surgery, the position information of the surgical instruments is acquired in real time and mapped onto the three-dimensional navigation model; The system acquires continuous image frames from surgical instruments in real time, and obtains intraoperative motion deviations through optical flow estimation based on the real-time image data. Based on the intraoperative motion deviations, the system corrects the target key points and target paths.

[0015] Thirdly, this application provides a computer-readable storage medium storing a computer program that can be loaded by a processor and executed as described above for an AI-based intelligent navigation-assisted surgical positioning method.

[0016] In summary, this dynamic navigation mechanism, which uses optical flow estimation to acquire intraoperative motion deviations in real time and correct target localization, improves the accuracy and reliability of surgical localization. Furthermore, dynamic correction using optical flow estimation eliminates the need for repeated acquisition of patient site images during surgery and allows for real-time updates and adjustments to the 3D navigation model, significantly reducing the time delay associated with dynamic localization correction. Additionally, after constructing the 3D navigation model, it is validated and corrected to account for individual differences in tissue structure, ensuring that the 3D navigation model used for assisted localization matches the current patient, further increasing the reliability of assisted surgical localization. Moreover, besides using optical flow estimation for dynamic localization correction, a model influencing the patient's surgical site structural offset and physiological indicators is established to adjust the assisted localization. This, combined with motion deviation correction achieved through optical flow estimation, further enhances the accuracy of assisted surgical localization. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the modules of a surgical assistance positioning system based on AI intelligent navigation provided in an embodiment of this application; Figure 2 These are schematic diagrams of the various units of the navigation model construction module provided in the embodiments of this application; Figure 3 This is a schematic diagram of each unit of the positioning and tracking module provided in the embodiments of this application; Figure 4 This is a schematic diagram of each unit of the real-time positioning correction module provided in the embodiments of this application; Figure 5 This is a flowchart of a surgical-assisted positioning method based on AI intelligent navigation provided in an embodiment of this application. Detailed Implementation

[0018] The following is in conjunction with the appendix Figure 1 -Appendix Figure 5 This application will be described in further detail below.

[0019] This application provides a surgical assistance positioning system based on AI intelligent navigation, see [link to relevant documentation]. Figure 1 It includes an image data acquisition module 10, a navigation model construction module 20, a positioning and tracking module 30, and a real-time positioning correction module 40.

[0020] The image data acquisition module 10 is used to acquire raw image data of the patient's surgical site and surgical planning information.

[0021] The navigation model construction module 20 is used to construct a three-dimensional navigation model based on the original image data, and to generate target key points and target paths based on surgical planning information through a preset AI model.

[0022] The positioning and tracking module 30 is used to acquire the position information of surgical instruments in real time during the operation and map the position information into the three-dimensional navigation model.

[0023] The real-time positioning correction module 40 is used to acquire continuous image frames collected by surgical instruments in real time, and based on the real-time image data, to obtain intraoperative motion deviation through optical flow estimation, and to correct the target key points and target path based on the intraoperative motion deviation.

[0024] The raw image data is unprocessed image data collected by medical imaging equipment that can truly reflect the anatomical structure of the patient's surgical site. Essentially, it digitizes the patient's internal structures such as bones, organs, blood vessels, and lesions to provide a visualized anatomical map for surgery. It is mainly acquired through medical imaging equipment such as CT and MRI.

[0025] Surgical planning information refers to the surgical operation plan generated by doctors by combining original imaging data, patient condition (such as lesion stage and physical condition), and surgical goals (such as complete / partial tumor resection and precise implantation of prosthesis). In essence, it is to digitize and standardize the complete surgical execution plan, so as to facilitate integration with AI navigation for assisted positioning.

[0026] Surgical planning information includes target area information (target location, such as the specific location, size, and number of stones) and surgical approach (surgical type and starting point of instrument entry path).

[0027] In this embodiment of the application, firstly, before the patient undergoes surgery, the image acquisition module 10 acquires the original image data of the patient's surgical site and surgical planning information.

[0028] Then, the navigation model construction module 20 constructs a three-dimensional navigation model based on the original image data, and generates target key points and target paths based on surgical planning information through a preset AI model.

[0029] Among them, the three-dimensional navigation model is a digital model that transforms two-dimensional image data into a three-dimensional spatial structure; the target key point is a key point defined based on the patient's surgical goal. For example, in stone surgery, the target point can be regarded as the center point of the stone; the target path is the path formed with the patient's external entry point as the starting point and the target key point as the ending point.

[0030] Specifically, see Figure 2 The navigation model construction module 20 includes an image preprocessing unit 21, a 3D navigation construction unit 22, a key point generation unit 23, and a path generation unit 24.

[0031] The image preprocessing unit 21 is used to preprocess the original image data to generate standardized image data.

[0032] The 3D navigation construction unit 22 is used to generate a 3D navigation model based on standardized image data and a preset 3D reconstruction method.

[0033] The key point generation unit 23 is used to generate target key points for the three-dimensional navigation model based on surgical planning information.

[0034] The path generation unit 24 is used to generate a target path for the 3D navigation model based on the target key points.

[0035] In this embodiment of the application, after obtaining the original image data of the patient's surgical site and surgical planning information, the original image data is first preprocessed by the image preprocessing unit 21, including denoising and enhancing the original image data to improve image quality, and normalizing the image pixel values ​​to map to a uniform range to generate standard image data.

[0036] Then, the 3D navigation construction unit 22 will generate a 3D navigation model from the preprocessed standard image data using a preset 3D reconstruction method.

[0037] Specifically, the standard image data is first segmented to obtain two-dimensional segmented images, which can reflect the anatomical structure information related to the surgery. Then, using a preset three-dimensional reconstruction method, such as a surface-based method, the boundary contours of the target object are extracted from the two-dimensional segmented images. Then, algorithms such as triangular mesh generation are used to connect these contours into a three-dimensional surface mesh. Finally, techniques such as texture mapping are used to add details and realism to the three-dimensional surface model, thereby constructing a detailed three-dimensional model containing spatial information of the surgical site and the relative positional relationships of various tissues, i.e., a three-dimensional navigation model.

[0038] After constructing the 3D navigation model, target key points and target paths can be generated from the 3D navigation model. Since the surgical planning information contains the target location, the target key points can be determined based on the 3D navigation model, and the target path can be generated based on the target key points combined with the surgical planning information.

[0039] For example, percutaneous nephrolithotomy is used as an example. A three-dimensional navigation model (STL mesh of kidney, stone, and blood vessel + coordinate parameters) and surgical planning information (target area: 8mm stone in the lower calyx of the left kidney; approach: percutaneous puncture; obstacle avoidance: ≥5mm from the renal artery) are used.

[0040] The target key points are determined as follows: locate the center of the stone, the skin puncture point, and the safe boundary point of the blood vessel, and obtain the position coordinates of the target key points. Then, based on the target key points, a straight puncture path that avoids the blood vessel is generated through a constraint optimization algorithm, which is the target path.

[0041] After constructing the 3D navigation model and determining the target path, the surgery can begin. During the surgery, the positioning and tracking module 30 will acquire the position information of the surgical instruments in real time and map the position information of the surgical instruments into the 3D navigation model.

[0042] Because the pre-operative planned target path can be simulated and visualized in a 3D model, by mapping the real-time position information of surgical instruments into the 3D navigation model, doctors can intuitively compare the deviation between the "current operation" and the "planned path" during the operation. For example, in urological surgery, the model can show whether the puncture needle has deviated from the preset angle, helping doctors to make timely adjustments.

[0043] Specifically, see Figure 3 The positioning and tracking module 30 includes a spatial mapping unit 31, a spatial positioning unit 32, and a coordinate mapping unit 33.

[0044] Among them, the spatial mapping unit 31 is used to construct a mapping model from the patient space to the three-dimensional navigation model space, denoted as the spatial mapping model.

[0045] The spatial positioning unit 32 is used to acquire the spatial pose information of surgical instruments in real time during the operation.

[0046] The coordinate mapping unit 33 is used to map the spatial pose information of surgical instruments into the three-dimensional navigation model through the spatial mapping model.

[0047] Surgical instruments include image acquisition instruments and operating instruments. The former is used to continuously acquire real-time image frames during surgery, while the latter is the instrument for performing surgical operations. The mapping of the real-time position information of surgical instruments into a three-dimensional navigation model mentioned here refers to the operating instruments.

[0048] First, in order to ensure that the anatomical structures in the model correspond one-to-one with their actual positions in the patient's body, the three-dimensional navigation model needs to be registered with the surgical space coordinate system. That is, a mapping model from the patient space to the three-dimensional navigation control is constructed through the spatial mapping unit 31, which is denoted as the spatial mapping model.

[0049] Specifically, before surgery, reference markers are calibrated. For example, several reference markers (such as optical markers attached to the waist) are fixed on the patient's body surface. These markers are visible in preoperative CT / MRI images and have been located in the three-dimensional navigation model, which means that the spatial coordinates of the markers in the three-dimensional navigation model can be determined.

[0050] The next step is to calculate the transformation matrix. The physical coordinates of the patient's surface markers (denoted as P1, P2, P3) are identified by the positioning device. For example, infrared light reflected from optical markers is captured by an infrared camera, and the coordinates are calculated based on triangulation. At the same time, the coordinates of these markers in the 3D navigation model (denoted as M1, M2, M3) are obtained. The transformation matrix from the patient space to the 3D navigation model is calculated using a point set registration algorithm, such as the ICP algorithm. The transformation matrix includes the rotation parameter R and the translation parameter T. That is, after rotation and translation, the coordinate point P in the patient space becomes the coordinate point M in the 3D navigation model, which can be expressed as: M = R × P + T.

[0051] For example, the physical coordinates of a certain reference mark on the patient's waist are (50,30,20) mm, and the coordinates in the three-dimensional model are (150,80,60) mm. By the correspondence of multiple mark points, R and T can be solved, and thus the spatial mapping model can be obtained.

[0052] Next, the spatial positioning unit 32 can acquire the spatial orientation information of the surgical instruments in real time during the operation.

[0053] Spatial pose information includes position information (physical spatial coordinates) and attitude information (pitch angle, yaw angle, roll angle).

[0054] Obtaining the spatial pose information of surgical instruments is mainly achieved by positioning and marking the surgical instruments, for example, by using optical positioning methods or electromagnetic positioning methods.

[0055] Taking a certain type of urological surgery as an example, the surgical instrument includes a flexible ureteroscope. Using optical positioning, three infrared reflective spheres (arranged in an isosceles triangle with sides of 20mm, pre-calibrated) are fixed to the end of the flexible ureteroscope. During the surgery, an infrared camera captures the real-time coordinates of the three reflective spheres (e.g., A: (300, 200, 100)mm, B: (310, 210, 100)mm, C: (305, 205, 105)mm). The centroid coordinates of points A, B, and C ((305,205,101.7) mm) can be calculated based on the real-time coordinates. This is the physical position of the end of the ureteroscope, which is also the physical spatial coordinate of the surgical instrument.

[0056] By comparing the angle difference between the triangle formed by the three points in real time and the triangle calibrated before surgery, the yaw angle and pitch angle of the flexible endoscope can be obtained. Then, the roll angle can be calculated through spatial vectors, which is the attitude information of the surgical instrument.

[0057] After determining the spatial pose information of the surgical instruments, the coordinate mapping unit 33 maps the spatial pose information of the surgical instruments to the three-dimensional navigation model through the spatial mapping model. That is, the spatial pose information of the surgical instruments is converted into the spatial pose information in the three-dimensional navigation model through the spatial mapping model, and then the position status of the surgery can be displayed in real time through the three-dimensional navigation model.

[0058] Because changes in patient position and tissue morphology during surgery can affect the spatial mapping model, any deviation in the spatial mapping model can cause discrepancies between the actual position of surgical instruments and the position displayed in the 3D navigation model. Therefore, it is necessary to verify the spatial mapping model during surgery.

[0059] Specifically, see Figure 3 The positioning and tracking module 30 also includes a mapping model verification unit 34 and a mapping model correction unit 35.

[0060] The mapping model verification unit 34 is used to verify the spatial mapping model and generate the model verification result.

[0061] The mapping model correction unit 35 is used to perform compensation correction based on the model verification results using a preset method.

[0062] Verification of the spatial mapping model mainly involves capturing real-time images of the patient's surgical site (such as C-arm X-ray, ultrasound, and endoscopy), setting at least three reference points, obtaining the coordinates of the reference points (recorded as physical coordinates), and performing image registration between the real-time images and the 3D navigation model. The corresponding positions of the reference points are then found in the 3D navigation model and recorded as model coordinates.

[0063] By comparing the physical coordinates and the model coordinates, the deviations (Δx, Δy, Δz) of each reference point in the three axes of x, y, and z can be calculated to obtain the deviation vector, thus yielding the positional deviations in each axis.

[0064] For example, if the physical coordinates of reference point A are (100, 50, 80) and the model coordinates are (101, 50, 80), then the deviation vector is (-1, 0, 0), indicating that there is a 1mm deviation in the x-axis direction.

[0065] If the position deviation exceeds the preset deviation threshold, for example, if the position deviation is >0.5mm, it means that the spatial mapping model has failed and the transformation matrix needs to be corrected based on the deviation. Therefore, the position deviation is fed back to the mapping model correction unit 35 as the model verification result.

[0066] The mapping model correction unit 35 will correct the spatial mapping model according to the model verification results using a preset method.

[0067] One of the preset methods, such as the least squares method, involves constructing an objective function that minimizes the deviation. By solving for R and T, which minimize the sum of squared deviations of all reference points, the transformation matrix can be updated, thus completing the correction of the spatial mapping model.

[0068] Considering that the auxiliary positioning in surgery is based on a three-dimensional navigation model, but the three-dimensional navigation model constructed before surgery is static and is based on the patient's fixed position, while during the surgery, the patient and organs will move continuously due to physiological activities or external factors, which can easily lead to deviations between the target key points and the target path.

[0069] Therefore, in this embodiment, the real-time positioning correction module 40 will also acquire continuous image frames collected by the surgical instruments in real time, and based on the real-time image data, the intraoperative motion deviation will be obtained by optical flow estimation, and the target key points and target path will be corrected based on the intraoperative motion deviation.

[0070] Specifically, see Figure 4 The real-time positioning correction module 40 includes an image frame acquisition unit 41, a motion offset unit 42, and a positioning correction unit 43.

[0071] The image frame acquisition unit 41 is used to acquire continuous image frames acquired by surgical instruments in real time.

[0072] The motion offset unit 42 is used to obtain intraoperative motion deviation based on continuous image frames through optical flow estimation.

[0073] The positioning correction unit 43 is used to correct the target key points and target path based on the intraoperative motion deviation when the intraoperative motion deviation reaches the preset correction threshold.

[0074] The surgical instruments mentioned above include image acquisition instruments, which are used to continuously acquire real-time image frames during surgery to provide a real-time, intuitive view of local details.

[0075] First, the continuous image frames acquired by the surgical instruments will be obtained in real time through the image frame acquisition unit 41.

[0076] Then, the intraoperative motion deviation is obtained by optical flow estimation based on continuous image frames through the motion offset unit 42.

[0077] Specifically, feature points are first extracted based on consecutive image frames. These feature points are stable and have obvious contrast, such as sharp corners of the stone edge or intersections of vascular patterns in the renal calyx wall. These feature points can be clearly identified in both consecutive image frames. For example, the Shi-Tomasi corner detection algorithm can be used to extract a number of feature points.

[0078] Then, the change of pixel coordinates of feature points in two consecutive image frames is calculated by optical flow estimation algorithm, which is the two-dimensional pixel motion vector (u,v) of feature points. Combined with the principle of perspective projection, the pixel coordinates of the pixel plane are converted into physical space coordinates, and the two-dimensional pixel motion vector is converted into three-dimensional motion deviation (Δa,Δb,Δc) in physical space, which is the motion deviation of the target key point in physical space during the operation.

[0079] Finally, the motion deviation is judged by a preset motion correction threshold. If the motion deviation reaches the motion correction threshold, the target key points and target path are corrected.

[0080] For example, if a target key point is M0(x0,y0,z0) and (Δa,Δb,Δc), then the corrected target key point is M1=M0+(Δa,Δb,Δc).

[0081] After obtaining the corrected target key points, the target path can be readjusted to obtain the corrected target path.

[0082] In addition, when the motion deviation obtained by optical flow estimation suddenly increases and exceeds the real-time correction capability, it is possible that the patient has undergone a significant change in body position. In this case, the spatial mapping model may fail, prompting a re-verification and correction of the spatial mapping model.

[0083] In addition, considering that patients' physiological activities (such as breathing and heartbeat) can trigger periodic tissue movements, such as the expansion / contraction of the lungs during thoracic surgery, which can cause the location of the target lesion (such as a lung nodule) to change in real time, it is possible to incorporate such periodic movement changes into the auxiliary localization, which can also help to correct the localization of optical flow estimation to some extent.

[0084] Therefore, in this embodiment of the application, the surgical assisted positioning system further includes an individual association module 50.

[0085] Among them, the individual association module 50 is used to construct a physiological dynamic impact model based on the surgical type and target key points, and to obtain the reference offset by monitoring the associated physiological indicators based on the physiological dynamic impact model during the surgical process.

[0086] Specifically, the individual association module 50 includes a physiological indicator determination unit 51, a physiological data sampling unit 52, a physiological dynamic model generation unit 53, and a physiological motion monitoring unit 54.

[0087] Among them, the physiological indicator determination unit 51 is used to determine the associated physiological indicators based on the surgical type through a preset surgical type knowledge base.

[0088] The physiological data sampling unit is used to collect the patient's related physiological indicator data and the location data of the target key points simultaneously, based on the related physiological indicators, to ensure that the timestamps of the data are completely aligned, so as to form the patient's individual physiological data.

[0089] The physiological dynamic model generation unit is used to construct a physiological dynamic impact model based on individual patient physiological data.

[0090] The physiological motion monitoring unit is used to monitor relevant physiological indicators during surgery and, based on the physiological dynamic influence model, obtain the motion offset of key points, which is denoted as the reference offset.

[0091] The pre-defined surgical type knowledge base stores the associated physiological indicators for different surgical types. For example, for cardiovascular surgery (such as atrial fibrillation ablation), the associated physiological indicators are: heart rate (cardiac beat cycle) and blood pressure (vascular filling); for orthopedics (such as hip replacement), the associated physiological indicator is: none (the skeleton is a rigid structure, and physiological indicators do not change its anatomical position).

[0092] The physiological dynamics model characterizes the mapping relationship between individualized physiological indicators of patients and movement at the surgical site.

[0093] First, relevant physiological indicators can be determined based on the current surgical type. Then, based on these indicators, the patient's relevant physiological indicator data and the location data of the target key points are collected synchronously to ensure that the timestamps of the data are completely aligned, thus forming individual physiological data for the patient.

[0094] Next, by analyzing and calculating the data, such as fitting individualized data through a linear regression model, a mapping relationship between individualized physiological indicators of patients and the movement of surgical sites can be generated, i.e., a physiological dynamic influence model.

[0095] After determining the physiological dynamic influence model, during the operation, physiological indicators can be monitored, and the motion offset of key points, i.e., the reference offset, can be obtained synchronously based on the physiological dynamic influence model. This allows for the correction of the intraoperative motion deviation obtained by optical flow estimation, thereby further ensuring the accuracy of positioning.

[0096] In this embodiment, the assisted positioning system further includes a data storage management module 60.

[0097] Specifically, the data storage management module 60 is used to store the patient's original image data of the surgical site and the positioning data during the operation, and to store the positioning data in a preset auxiliary positioning database.

[0098] The positioning data includes target key points and target paths, intraoperative movement deviations, and correction records. By storing and recording this data, it can assist in pathological analysis and provide data support for subsequent positioning system optimization. For example, for positioning deviations that occur during surgery, data analysis can be used to determine the cause of the deviation and then make targeted optimizations and adjustments.

[0099] This application also provides a surgical-assisted positioning method based on AI intelligent navigation, see [link to relevant documentation]. Figure 5 It includes the following steps: S100: Obtain raw image data of the patient's surgical site and surgical planning information.

[0100] S200 constructs a 3D navigation model based on raw image data, and generates target key points and target paths through a preset AI model based on surgical planning information.

[0101] S300: During the operation, the position information of the surgical instruments is acquired in real time and mapped to the three-dimensional navigation model.

[0102] S400 acquires continuous image frames from surgical instruments in real time, and obtains intraoperative motion deviations through optical flow estimation based on real-time image data, and corrects target key points and target paths based on intraoperative motion deviations.

[0103] First, obtain the patient's original imaging data of the surgical site and surgical planning information.

[0104] Then, based on the original image data, a 3D navigation model is constructed, and based on the surgical planning information, target key points and target paths are generated through a preset AI model.

[0105] Next, during the surgery, the position information of the surgical instruments is acquired in real time and mapped into the three-dimensional navigation model.

[0106] Finally, continuous image frames acquired by surgical instruments are obtained in real time, and intraoperative motion deviation is obtained through optical flow estimation based on real-time image data. The target key points and target path are then corrected based on the intraoperative motion deviation.

[0107] This application also provides a computer-readable storage medium storing a computer program that can be loaded by a processor and executed by any of the above-described AI-based intelligent navigation surgical assistance positioning systems.

[0108] The embodiments described in this application are preferred embodiments of this application and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the principles of this application should be included within the scope of protection of this application.

Claims

1. A surgical assistance positioning system based on AI intelligent navigation, characterized in that, include: The image data acquisition module is used to acquire raw image data of the patient's surgical site and surgical planning information; The navigation model construction module is used to build a 3D navigation model based on the original image data, and generate target key points and target paths based on surgical planning information through a preset AI model; The positioning and tracking module is used to acquire the position information of surgical instruments in real time during the operation and map the position information into the three-dimensional navigation model; The real-time positioning correction module is used to acquire continuous image frames from surgical instruments in real time, and based on the real-time image data, to obtain intraoperative motion deviation through optical flow estimation, and to correct the target key points and target path based on the intraoperative motion deviation.

2. The surgical assistance positioning system based on AI intelligent navigation according to claim 1, characterized in that, The navigation model construction module includes: The image preprocessing unit is used to preprocess the raw image data to generate standardized image data; The 3D navigation construction unit is used to generate a 3D navigation model based on standardized image data and a preset 3D reconstruction method. The key point generation unit is used to generate target key points for the 3D navigation model based on surgical planning information; The path generation unit is used to generate a target path based on the target key points and 3D navigation information.

3. The surgical assistance positioning system based on AI intelligent navigation according to claim 1, characterized in that, The positioning and tracking module includes: The spatial mapping unit is used to construct a mapping model from the patient space to the three-dimensional navigation model space, denoted as the spatial mapping model; The spatial positioning unit is used to acquire the spatial pose information of surgical instruments in real time during the operation. The coordinate mapping unit is used to map the spatial pose information of surgical instruments into the three-dimensional navigation model through the spatial mapping model.

4. The surgical assistance positioning system based on AI intelligent navigation according to claim 3, characterized in that, The positioning and tracking module further includes: The mapping model verification unit is used to verify the spatial mapping model and generate the model verification results. The mapping model correction unit is used to perform compensation corrections based on the model validation results using a preset method.

5. A surgical assistance positioning system based on AI intelligent navigation according to claim 1, characterized in that, The real-time positioning correction module includes: The image frame acquisition unit is used to acquire continuous image frames acquired by surgical instruments in real time. Motion offset unit, used to obtain intraoperative motion deviation based on optical flow estimation from consecutive image frames; The positioning correction unit is used to correct the target key points and target path based on the intraoperative motion deviation when the intraoperative motion deviation reaches the preset correction threshold.

6. The surgical assistance positioning system based on AI intelligent navigation according to claim 1, characterized in that, The surgical planning information includes the surgical type, and the system also includes: The individual association module is used to construct a physiological dynamic impact model based on the surgical type and target key points. During the surgery, the module monitors the associated physiological indicators based on the physiological dynamic impact model to obtain the reference offset.

7. A surgical assistance positioning system based on AI intelligent navigation according to claim 6, characterized in that, The individual association module includes: The physiological indicator determination unit is used to determine the relevant physiological indicators based on the type of surgery, using a pre-set surgical type knowledge base. The physiological data sampling unit is used to collect the patient's related physiological index data and the location data of the target key points simultaneously, based on the related physiological indicators, to ensure that the timestamps of the data are completely aligned, so as to form the patient's individual physiological data; The physiological dynamic model generation unit is used to construct a physiological dynamic impact model based on individual patient physiological data. The physiological motion monitoring unit is used to monitor relevant physiological indicators during surgery and, based on the physiological dynamic influence model, obtain the motion offset of key points, which is denoted as the reference offset.

8. A surgical assistance positioning system based on AI intelligent navigation according to claim 1, characterized in that, The system also includes: The data storage management module is used to store the patient's original image data of the surgical site and the positioning data during the operation, and to store the data in a preset auxiliary positioning database.

9. A method for real-time detection of surface defects in wet wipes based on Transformer, characterized in that, include: Obtain raw imaging data of the patient's surgical site and surgical planning information; Based on the original image data, a 3D navigation model is constructed, and based on the surgical planning information, target key points and target paths are generated through a preset AI model; During the surgery, the position information of the surgical instruments is acquired in real time and mapped onto the three-dimensional navigation model; The system acquires continuous image frames from surgical instruments in real time, and obtains intraoperative motion deviations through optical flow estimation based on the real-time image data. Based on the intraoperative motion deviations, the system corrects the target key points and target paths.

10. A computer-readable storage medium storing a computer program capable of being loaded by a processor and executing the surgical-assisted positioning method based on AI intelligent navigation as described in claim 9.

Images