A radiation-free ultrasound three-dimensional reconstruction ar spinal puncture real-time perspective guiding system

CN122643038APending Publication Date: 2026-08-28FUJIAN ZHANGZHOU HOSPITAL
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202611068345.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-17
Publication Date
2026-08-28

AI Technical Summary

Technical Problem

[0003]常规修正方式采用增加标志贴片密度或提高相机分辨率以抑制偏差,但此类手段未包含层间滑移的生物力学约束,贴片重叠反而增加布设复杂度且加剧变形干扰,并未解决静态建模与动态现实的错配,此外,现有空间数据配准与跟踪控制算法在应对动态解剖形变时同样存在局限,例如,公开号为CN116847799A的中国发明专利申请公开了一种用于导航脊柱手术的增强现实脊柱棒规划和弯曲的计算机实现方法,通过医疗导航装置和增强现实装置获取并跟踪脊柱螺钉位置与脊柱棒形状,实现虚拟模型与真实构件重叠显示,方案底层控制逻辑建立在解剖结构静态或各椎体间保持刚性相对静止的理想假设之上,在穿刺手术实时随动工况下,人体呼吸与微震及体位变动引发浅表皮肤与皮下骨性组织间层间剪切运动,导致静态螺钉位置采样特征无法映射皮下腰椎椎体实时多维位姿畸变,核心预设前提与实际复杂动力学边界条件的根本错配,使配准矩阵在动态多目标跟踪时发生坐标污染与解算时延,无法达到毫米级精确对齐要求

Benefits of technology

1、在无辐射超声三维重建AR脊柱穿刺实时透视引导中,脂肪抑制超声三维脊柱重建模块与呼吸心跳体动三级滤波与皮肤形变补偿模块协同,利用六维姿态光学跟踪模块采集瞬态形变应变张量,代入层间剪切滑移传递矩阵以分离皮肤浅表层间滑移剪切量,提取传递至深层腰椎筋膜层的有效驱动力矩向量并加载至多刚体运动链拓扑网络,借助相邻椎体间椎间盘的物理抗扭转刚度与黄韧带张力拉伸极限确定独立腰椎椎体空间六维位姿矩阵,同步更新虚拟模型中对应的独立骨骼网格单元,避免深浅解剖层运动异构导致的解剖结构位姿漂移。

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122643038A_ABST
    Figure CN122643038A_ABST
Patent Text Reader

Abstract

The application relates to the technical field of medical ultrasonic reconstruction and augmented reality surgical navigation, and discloses a radiation-free ultrasonic three-dimensional reconstruction AR spinal puncture real-time perspective guiding system, which comprises an ultrasonic reconstruction module, an optical tracking module, a deformation compensation module, a needle body tracking module and a needle path planning module. The system collects a transient strain tensor of a spatial triangular plane of a fixed anatomical site and inputs the transient strain tensor into a shear slip transmission matrix to separate an epidermal slip amount and a deep layer torque vector, so as to correct a pose matrix of a spinal column model. In combination with the output of a safe needle path, the application corrects the bone pose according to the nonlinear slip constraint between deep fascia layers, eliminates the dynamic drift of the model caused by the motion isomerism of deep and shallow anatomical layers, realizes virtual-real alignment, cooperatively suppresses fat low echo clutter, and ensures puncture safety.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to a radiation-free ultrasound three-dimensional reconstruction AR spinal puncture real-time fluoroscopic guidance system, belonging to the field of medical ultrasound reconstruction and augmented reality surgical navigation technology. Background Technology

[0002] Current medical navigation systems integrate medical imaging and spatial positioning technologies to assist in surgical procedures. Among them, using ultrasound to acquire tissue data and combining it with augmented reality perspective technology to display the spinal structure is a common method for minimally invasive punctures. This type of solution uses an optical camera to capture back landmarks to calculate spatial pose and then overlays a three-dimensional mesh onto the human surface. However, due to the nonlinear interlayer shear slip motion between the superficial skin and the deep fascia and peripheral structures of the deep lumbar vertebrae during puncture in the lateral decubitus position with knees bent and back arched, the elastic displacement of the superficial landmarks and the rigid body space matrix of the multi-rigid-body kinematic chain topology of the subcutaneous vertebrae produce non-affine variations. Conventional methods treat these as synchronous and homogeneous changes, resulting in mesh distortion or topological misalignment in the model. This becomes the implicit system cost of maintaining the radiation-free advantage.

[0003] Conventional correction methods employ increasing marker patch density or improving camera resolution to suppress deviations. However, these methods do not incorporate the biomechanical constraints of interlayer slippage. Patch overlap actually increases deployment complexity and exacerbates deformation interference, failing to address the mismatch between static modeling and dynamic reality. Furthermore, existing spatial data registration and tracking control algorithms also have limitations in handling dynamic anatomical deformations. For example, Chinese invention patent application CN116847799A discloses a computer-implemented method for augmented reality spinal rod planning and bending for navigation spinal surgery, which acquires and tracks spinal data through a medical navigation device and an augmented reality device. The position of the screw and the shape of the spinal rod are used to achieve the superimposed display of the virtual model and the real component. The underlying control logic of the solution is based on the ideal assumption that the anatomical structure is static or that each vertebra is relatively rigid and static. Under the real-time follow-up condition of the puncture operation, human breathing, micro-vibration and body position changes cause interlayer shearing motion between the superficial skin and subcutaneous bony tissue. This causes the static screw position sampling features to be unable to map the real-time multi-dimensional pose distortion of the subcutaneous lumbar vertebrae. The fundamental mismatch between the core preset premise and the actual complex dynamic boundary conditions causes the registration matrix to be contaminated with coordinates and delayed in solution during dynamic multi-target tracking, making it impossible to achieve the millimeter-level accurate alignment requirement.

[0004] Therefore, the technical problem to be solved by this invention is how to overcome the distortion of ultrasound modeling under the conditions of obesity and degenerative deformity, and how to determine the six-dimensional pose matrix of the multi-vertebral joint chain by separating the interlayer slip shear amount, while unifying the access of infrared and visual dual-redundant puncture needle positioning features to drive the virtual and real rendering of the fluoroscopic glasses and the intelligent correction and obstacle avoidance of the needle path. Summary of the Invention

[0005] To address the problems in the background art, the technical solution of the present invention is as follows: A radiation-free ultrasound three-dimensional reconstruction AR spinal puncture real-time fluoroscopic guidance system, the system comprising: The fat-suppressed ultrasound 3D spine reconstruction module is used to segment the subcutaneous fat layer and bone region, and output anatomical boundary features and a 3D model of the spine. The six-dimensional attitude optical tracking module is used to acquire the transient deformation strain tensor of the spatial triangular plane formed by the 12th thoracic anatomical site, the 3rd lumbar anatomical site, and the 1st sacral anatomical site. The respiratory, heartbeat, and body motion three-level filtering and skin deformation compensation module is connected to the six-dimensional attitude optical tracking module. It is used to input the transient deformation strain tensor into the interlayer shear slip transfer matrix containing a preset attenuation factor, separate the epidermal slip shear amount and the deep driving torque vector, and correct the spatial pose matrix of the three-dimensional model of the spine according to the deep fascia interlayer nonlinear slip constraint. An infrared vision dual-redundant puncture needle tracking module is used to capture the spatial coordinates of the axis of a common medical puncture needle. The needle path planning and high-risk obstacle avoidance early warning module is connected to the respiratory, heartbeat and body motion three-level filtering and skin deformation compensation module and the infrared vision dual-redundant puncture needle tracking module, respectively. It is used to calculate and output the safe puncture needle path based on the corrected spatial pose matrix and axis spatial coordinates.

[0006] Preferably, the fat-suppressed ultrasound three-dimensional spinal reconstruction module includes an automatic fat and bone segmentation unit and a spinal degeneration and deformity topology fitting unit; the automatic fat and bone segmentation unit is used to eliminate acoustic noise interference on the spinous process, lamina, and intervertebral space, output anatomical boundary features, and input them into the three-dimensional spinal model; the spinal degeneration and deformity topology fitting unit is used to obtain the discrete morphological structures of bone hyperplasia, intervertebral space narrowing, and scoliosis to construct an individualized skeletal anatomical topology.

[0007] Preferably, the six-dimensional attitude optical tracking module continuously acquires the relative three-dimensional spatial displacement between the thoracic 12 anatomical sites, the lumbar 3 anatomical sites, and the sacral 1 anatomical sites, and calculates the transient deformation strain tensor of the spatial triangular plane.

[0008] Preferably, when the respiratory, cardiac, and bodily motion three-level filtering and skin deformation compensation module calculates the deep driving torque vector using the interlayer shear slip transfer matrix, it performs amplitude reduction filtering on the epidermal slip shear amount through a preset attenuation factor to separate the deep driving torque vector.

[0009] Preferably, the infrared vision dual-redundant puncture needle tracking module includes an infrared optical marker capture unit and a needle body contour recognition unit; the infrared optical marker capture unit is used to capture the six-dimensional spatial coordinates of the infrared marker fixed at the handle of the ordinary medical puncture needle to calculate the spatial pose; the needle body contour recognition unit is used to extract the edge contour of the metal needle body of the ordinary medical puncture needle to obtain the axial spatial coordinates when the infrared marker is obstructed by the line of sight.

[0010] Preferably, the system also includes a spatial three-dimensional reference coordinate system registration module; the spatial three-dimensional reference coordinate system registration module is connected to the six-dimensional attitude optical tracking module and the infrared vision dual-redundant puncture needle tracking module respectively, and is used to uniformly integrate the anatomical boundary features and axis spatial coordinates into the spatial three-dimensional reference coordinate system constructed by the thoracic 12 anatomical site, the lumbar 3 anatomical site and the sacral 1 anatomical site for spatial alignment, and output spatial global alignment data.

[0011] Preferably, the system also includes a virtual skeleton mesh unit spatial synchronization module; the virtual skeleton mesh unit spatial synchronization module is connected to the respiratory, heartbeat, and body motion three-level filtering and skin deformation compensation module and the spatial three-dimensional reference coordinate system registration module, respectively, to receive spatial global alignment data, and to perform spatial synchronous updates on the corresponding independent skeleton mesh units in the virtual model according to the spatial pose matrix, and output synchronized independent skeleton meshes.

[0012] Preferably, the system also includes a perspective virtual-real overlay rendering module; the perspective virtual-real overlay rendering module is connected to the virtual skeleton mesh unit spatial synchronization module and the needle path planning and high-risk obstacle avoidance early warning module, respectively, and is used to display the synchronous independent skeleton mesh and safe puncture needle path on the patient's back skin surface through augmented reality spatial virtual-real overlay rendering.

[0013] Preferably, the system also includes a safe puncture monitoring and constraint module; the safe puncture monitoring and constraint module is connected to the needle path planning and high-risk obstacle avoidance early warning module, and is used to obtain the safe puncture needle path, and when the angle of deviation of the axis spatial coordinate of the ordinary medical puncture needle from the safe puncture needle path is greater than the preset angle threshold, it outputs a dual warning signal of sound and light, and at the same time limits the refresh frequency of the three-dimensional space virtual and real superposition rendering display in the system to block the guidance process.

[0014] Compared with the prior art, the beneficial effects of the present invention are: 1. In radiation-free ultrasound 3D reconstruction AR spinal puncture real-time fluoroscopic guidance, the fat-suppressed ultrasound 3D spinal reconstruction module works in conjunction with the respiratory, cardiac, and body motion three-level filtering and skin deformation compensation module. The six-dimensional posture optical tracking module collects transient deformation strain tensors, which are substituted into the interlayer shear slip transfer matrix to separate the superficial interlayer slip shear of the skin. The effective driving torque vector transmitted to the deep lumbar fascia layer is extracted and loaded into the multi-rigid-body kinematic chain topology network. The physical torsional stiffness of the intervertebral disc between adjacent vertebrae and the tensile limit of the ligamentum flavum are used to determine the six-dimensional pose matrix of the independent lumbar vertebral body. The corresponding independent skeletal mesh units in the virtual model are updated synchronously to avoid anatomical structure pose drift caused by the heterogeneity of deep and superficial anatomical layers.

[0015] 2. The fat-suppressed ultrasound 3D spinal reconstruction module automatically segments the subcutaneous fat layer and bone region, blocking the interference of low-echo clutter on the bone contours of the spinous process, lamina, and intervertebral space. It enhances the strong echo component to extract clear anatomical boundaries. Furthermore, combined with the adaptive topology fitting algorithm for spinal degeneration and deformity, it abandons the standard geometric template topology structure and constructs an individualized anatomical topology based on the discrete morphology of patient-specific bone hyperplasia, intervertebral space narrowing, and mild scoliosis. This enables the system to obtain a highly deterministic 3D model of the lumbar spine under the dual conditions of subcutaneous fat clutter and pathological deformity, providing stable skeletal anatomical entity data for subsequent six-dimensional registration and active obstacle avoidance of high-risk structures.

[0016] 3. The infrared vision dual-redundant puncture needle tracking module works in conjunction with the optimal needle path planning and high-risk obstacle avoidance early warning module to capture the spatial coordinates of the miniature infrared markers under normal conditions. When the line of sight is obstructed, it seamlessly switches to machine vision subpixel contour recognition to extract the needle edge of the ordinary medical puncture needle. The multi-source positioning features are uniformly integrated into a three-dimensional reference coordinate system constructed by three fixed anatomical sites. The relative distance between the target puncture needle axis and the geometric topology of the subcutaneous ligamentum flavum and nerve roots is calculated in real time. When the distance is lower than the safety threshold, the obstacle avoidance path is automatically updated and virtual and real superposition rendering is driven. Under continuous micro-movement of the human body, the global convergence of multi-target coordinates is forcibly maintained to ensure puncture safety. Attached Figure Description

[0017] Figure 1 This invention: Structural diagram of an ultrasound three-dimensional spinal reconstruction and puncture guidance system; Figure 2 This is a diagram of the ultrasonic three-dimensional spinal reconstruction and puncture guidance system of the present invention.

[0018] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] The technical solutions in the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0020] A radiation-free ultrasound three-dimensional reconstruction AR spinal puncture real-time fluoroscopic guidance system, the system comprising: The fat-suppressed ultrasound 3D spine reconstruction module is used to segment the subcutaneous fat layer and bone region, and output anatomical boundary features and a 3D model of the spine. The six-dimensional attitude optical tracking module is used to acquire the transient deformation strain tensor of the spatial triangular plane formed by the 12th thoracic anatomical site, the 3rd lumbar anatomical site, and the 1st sacral anatomical site. The respiratory, heartbeat, and body motion three-level filtering and skin deformation compensation module is connected to the six-dimensional attitude optical tracking module. It is used to input the transient deformation strain tensor into the interlayer shear slip transfer matrix containing a preset attenuation factor, separate the epidermal slip shear amount and the deep driving torque vector, and correct the spatial pose matrix of the three-dimensional model of the spine according to the deep fascia interlayer nonlinear slip constraint. An infrared vision dual-redundant puncture needle tracking module is used to capture the spatial coordinates of the axis of a common medical puncture needle. The needle path planning and high-risk obstacle avoidance early warning module is connected to the respiratory, heartbeat and body motion three-level filtering and skin deformation compensation module and the infrared vision dual-redundant puncture needle tracking module, respectively. It is used to calculate and output the safe puncture needle path based on the corrected spatial pose matrix and axis spatial coordinates.

[0021] Preferably, the fat-suppressed ultrasound three-dimensional spinal reconstruction module includes an automatic fat and bone segmentation unit and a spinal degeneration and deformity topology fitting unit; the automatic fat and bone segmentation unit is used to eliminate acoustic noise interference on the spinous process, lamina, and intervertebral space, output anatomical boundary features, and input them into the three-dimensional spinal model; the spinal degeneration and deformity topology fitting unit is used to obtain the discrete morphological structures of bone hyperplasia, intervertebral space narrowing, and scoliosis to construct an individualized skeletal anatomical topology.

[0022] Preferably, the six-dimensional attitude optical tracking module continuously acquires the relative three-dimensional spatial displacement between the thoracic 12 anatomical sites, the lumbar 3 anatomical sites, and the sacral 1 anatomical sites, and calculates the transient deformation strain tensor of the spatial triangular plane.

[0023] Preferably, when the respiratory, cardiac, and bodily motion three-level filtering and skin deformation compensation module calculates the deep driving torque vector using the interlayer shear slip transfer matrix, it performs amplitude reduction filtering on the epidermal slip shear amount through a preset attenuation factor to separate the deep driving torque vector.

[0024] Preferably, the infrared vision dual-redundant puncture needle tracking module includes an infrared optical marker capture unit and a needle body contour recognition unit; the infrared optical marker capture unit is used to capture the six-dimensional spatial coordinates of the infrared marker fixed at the handle of the ordinary medical puncture needle to calculate the spatial pose; the needle body contour recognition unit is used to extract the edge contour of the metal needle body of the ordinary medical puncture needle to obtain the axial spatial coordinates when the infrared marker is obstructed by the line of sight.

[0025] Preferably, the system also includes a spatial three-dimensional reference coordinate system registration module; the spatial three-dimensional reference coordinate system registration module is connected to the six-dimensional attitude optical tracking module and the infrared vision dual-redundant puncture needle tracking module respectively, and is used to uniformly integrate the anatomical boundary features and axis spatial coordinates into the spatial three-dimensional reference coordinate system constructed by the thoracic 12 anatomical site, the lumbar 3 anatomical site and the sacral 1 anatomical site for spatial alignment, and output spatial global alignment data.

[0026] Preferably, the system also includes a virtual skeleton mesh unit spatial synchronization module; the virtual skeleton mesh unit spatial synchronization module is connected to the respiratory, heartbeat, and body motion three-level filtering and skin deformation compensation module and the spatial three-dimensional reference coordinate system registration module, respectively, to receive spatial global alignment data, and to perform spatial synchronous updates on the corresponding independent skeleton mesh units in the virtual model according to the spatial pose matrix, and output synchronized independent skeleton meshes.

[0027] Preferably, the system also includes a perspective virtual-real overlay rendering module; the perspective virtual-real overlay rendering module is connected to the virtual skeleton mesh unit spatial synchronization module and the needle path planning and high-risk obstacle avoidance early warning module, respectively, and is used to display the synchronous independent skeleton mesh and safe puncture needle path on the patient's back skin surface through augmented reality spatial virtual-real overlay rendering.

[0028] Preferably, the system also includes a safe puncture monitoring and constraint module; the safe puncture monitoring and constraint module is connected to the needle path planning and high-risk obstacle avoidance early warning module, and is used to obtain the safe puncture needle path, and when the angle of deviation of the axis spatial coordinate of the ordinary medical puncture needle from the safe puncture needle path is greater than the preset angle threshold, it outputs a dual warning signal of sound and light, and at the same time limits the refresh frequency of the three-dimensional space virtual and real superposition rendering display in the system to block the guidance process.

[0029] Example 1: In a surgical environment involving lateral decubitus puncture with knee flexion and arched back in a patient with high body weight and accompanying lumbar spondylosis and intervertebral space narrowing, the patient experiences low-frequency respiratory fluctuations, micro-vibrations of the heartbeat, and sudden body movements during the puncture. This causes discontinuous and non-affine interlayer shear slippage between the superficial skin, deep subcutaneous fascia, and peripheral structures of the lumbar vertebrae. This results in elastic displacement of the markers attached to the fixed anatomical sites on the body surface, causing non-affine variations between the superficial spatial coordinates and the inherent rigid body spatial matrix of the deep lumbar vertebrae. Because the conventional medical navigation system treats the deep and superficial anatomical structures in the body as homogeneous and synchronously changing continuous elastic bodies, it cannot output the interlayer shear hysteresis compensation. This leads to mesh distortion and topological misalignment of the virtual model in the augmented reality glasses, causing the puncture path to deviate from the target position and exceed the millimeter-level error limit required for safe puncture.

[0030] To address the spatial drift caused by the aforementioned anatomical and kinematic heterogeneity, the fat-suppressed ultrasound 3D spinal reconstruction module, after receiving ultrasound signals acquired by a free-arm scan of the lumbar spine region from the first lumbar vertebra to the first sacral vertebra, runs an automatic segmentation program for the subcutaneous fat layer and bone region. This program identifies and divides the subcutaneous fat layer and bone region, eliminating low-echoic clutter interference from the thick subcutaneous fat layer under high-weight conditions. It extracts the strong echo components of the spinous process, lamina, and intervertebral space to establish clear anatomical boundary features. Based on this, it calls a topology fitting program based on degenerative deformity correction, using the discrete boundary morphology of bone hyperplasia, intervertebral space narrowing, and mild scoliosis to construct an individualized skeletal anatomical topology network. This generates a patient-specific lumbar spine 3D model with a complete geometric topology, providing skeletal entity data for subsequent six-dimensional posture registration. Specifically, during operation, the topology fitting program based on degenerative deformity correction reconstructs the spatial point cloud from the discrete strong echo anatomical boundary point set of the spinous process, lamina, and intervertebral space output by the automatic segmentation unit. The program calculates the local normal vector and curvature center of each vertebral discrete plaque surface. Then, it retrieves a preset anatomical topology template mesh and calculates the minimum feature distance between each key vertebral landmark in the template mesh and the current discrete bone boundary point cloud. For areas with edge distortion caused by osteophyte formation, the mesh fitting algorithm adjusts the radial basis function weights of the surrounding mesh vertices based on the local high curvature change characteristics. Through local adaptive deformation, the standard mesh is fitted to the osteophyte boundary. For boundary adhesion caused by intervertebral space narrowing and axial deflection caused by scoliosis, the program extracts the local rigid body central axis and anatomical connection direction vector of each vertebral segment through principal component analysis of the point cloud. It also applies non-penetrating contact geometric constraints that conform to the physiological and mechanical limits of the human body between adjacent vertebrae. Thus, while maintaining the topological connectivity of the independent vertebral mesh, the program sequentially corrects the spatial offset of adjacent vertebral vertebral vertebrae. Finally, it outputs a closed, specific lumbar spine three-dimensional mesh surface model that truly reflects the patient's osteophyte formation, narrowing, and scoliosis deformity.

[0031] During the dynamic registration phase, the six-dimensional attitude optical tracking module continuously acquires the relative three-dimensional spatial displacement between the three fixed anatomical sites by using sterile positioning patches attached to the T12, L3, and S1 anatomical sites. It captures the transient deformation strain tensor of the spatial triangular plane formed by these three fixed anatomical sites at a sampling frequency of 200Hz. This transient deformation strain tensor is then transmitted in situ to the respiratory, cardiac, and body motion three-stage filtering and skin deformation compensation module. This module substitutes the transient deformation strain tensor into an interlayer shear slip transfer matrix with a preset attenuation factor. By calculating the spatial gradient divergence of superficial skin strain between the superficial and deep fascia layers, it reduces the epidermal slip shear amount and... It separates and eliminates the epidermal slip shear from the overall body surface displacement vector, thereby independently separating the epidermal slip shear and calculating the deep driving torque vector transmitted to the deep lumbar vertebral fascia layer. During this data transmission, external mechanical obstruction or changes in the attachment state of the body surface reference points often lead to local optical signal interruptions. The system completes spatial coordinate compensation through the integrated T12 / L3 / S1 dual-mode reference positioning module. Based on the principles of rigid body spatial motion geometry and the anatomical curvature constraint model between the inherent segments of the human spine, the input end receives the original coordinate data of the three-dimensional reference points of the T12, L3, and S1 spaces collected by an infrared optical camera with a working wavelength of 850nm and a sampling frequency of 200Hz. The data is then de-identified in the local graphics processor using a preset SHA256 hash image. The system performs anonymization and identification transformation on all coordinate parameters, removing natural person identity features. Initially, the infrared optical camera has a measurement accuracy of 0.05mm, and the computing platform's floating-point operation capability is no less than 10 trillion operations per second. When the output signal strength of the optical camera at any reference point falls below the preset 15% energy threshold due to line-of-sight obstruction, the T12 / L3 / S1 dual-mode reference positioning module triggers an intelligent completion program. This switches the infrared marking precision mode to anatomical topology extrapolation mode, reads the real-time three-dimensional spatial displacement vectors and local three-dimensional rotation matrices of the other two reference points that have not lost signals, compares the spatial Euclidean distance between the reference points with the standard rigid span in the preoperatively stored specific lumbar spine three-dimensional model, and determines the curvature continuity of the spine in the horizontal and sagittal planes. The system calculates the relative displacement and torsional slope of the missing site relative to the other reference points, fits and completes the transient three-dimensional coordinates of the missing site, and outputs the restored complete spatial triangular plane geometric topology data. This ensures that the current data stream continues to converge within the local graphics processor, and the final output registration variance quantitative value is less than 0.18 mm, which is used as a real-time quantitative risk parameter to characterize the guidance deviation. Subsequently, the deep driving torque vector is directly loaded into the multi-rigid-body kinematic chain topology network as the boundary excitation condition. The network is constructed based on preoperative ultrasound scans, and the fiber anchoring junction of the lumbar spinous process and the posterior leaf of the thoracolumbar fascia is determined as the first-order displacement transmission damping boundary. The physical torsional stiffness of the intervertebral disc between adjacent vertebrae and the tensile limit of the ligamentum flavum are used as hard boundary constraints.

[0032] By employing a spatial constraint optimization rule, a joint iterative optimization calculation is performed on the spatial pose matrix of each independent lumbar vertebra from the first to the fifth lumbar vertebrae in three-dimensional space until the overall potential energy equation of the entire kinematic chain reaches convergence. This outputs the latest spatial pose matrix of each independent lumbar vertebra. In this cross-scale kinematic state solution process, the physical transformation from the superficial strain field to the pose of the deep independent rigid body relies on a multi-layered continuous biomechanical transmission chain composed of subcutaneous connective tissue, deep back fascia, and thoracolumbar fascia. Specifically, the transient strain tensor of the body surface spatial triangular plane captured by the six-dimensional attitude optical tracking module is transformed into a spatial displacement gradient field of the superficial skin layer within the framework of continuum mechanics. The interlayer shear slip transfer matrix acts as an energy damping bridge with viscoelastic dissipation characteristics, and its preset attenuation factor simulates the shear of the superficial subcutaneous fat layer under shear compression. Modulus attenuation filters out the reactive slip deformation caused by the elastic stretching of the skin itself, extracting the effective driving torque that is truly transmitted to the deep fascia and caused by macroscopic torsion or bending of the trunk. The driving torque is applied as a discrete boundary load to the multi-rigid-body kinematic chain topology network composed of the first to fifth lumbar vertebrae, as well as adjacent intervertebral discs and ligamentum flavum. Since the vertebrae themselves are high-rigidity structures, and the intervertebral discs and ligamentum flavum exhibit isotropic nonlinear spring constraints under millimeter-level micro-motion conditions, the system discretizes the continuous energy transfer path into virtual work balance equations at each joint of the kinematic chain, so that the macroscopic multidimensional deformation scalar of the body surface can be uniquely mapped to the multi-degree-of-freedom relative rotation angles and relative displacements between deep discrete rigid bodies. This solves the energy dispersion caused by heterogeneous motion in the deep and shallow anatomical layers, ensuring that the joint iterative optimization is completely closed-loop and self-consistent in terms of physical and mechanical causality.

[0033] When the separated epidermal slip shear exceeds the safety control boundary threshold of 4.5mm, the background coordinate registration and real-time rendering software module increases the slip calibration compensation weight of the corresponding traction area and applies the output spatial pose matrix of each lumbar vertebra to the virtual skeleton mesh unit spatial synchronization module in situ. Based on the spatial pose matrix, the corresponding independent skeleton mesh units in the virtual model are spatially synchronized and updated to output a synchronized independent skeleton mesh, correcting the spatial mismatch caused by body position slip. At the same time, the infrared vision dual-redundant puncture needle tracking module captures and fixes the needle under normal conditions through the infrared optical marker capture unit. The six-dimensional spatial coordinates of the infrared marker at the handle of a standard medical puncture needle are used to calculate the spatial pose in real time. When the infrared marker is obstructed by the hands or surgical drapes during the operation, the infrared vision dual-redundant puncture needle tracking module switches to the needle body contour recognition unit. The machine vision sub-pixel contour recognition program extracts the edge contour of the metal needle body of the standard medical puncture needle, thereby continuously obtaining the axial spatial coordinates of the standard medical puncture needle. The spatial three-dimensional reference coordinate system registration module then integrates the above-mentioned anatomical boundary features and axial spatial coordinates into the spatial three-dimensional reference coordinate system constructed by three fixed anatomical sites to complete global alignment.

[0034] In practice, the machine vision subpixel contour recognition program retrieves a two-dimensional infrared or visible light video stream of the surgical area captured by a high-resolution camera. It uses grayscale striping and spatial direction filtering to coarsely locate the metal edge of a common medical puncture needle, defining a region of interest (ROI) containing the needle body. Due to glare and low-echo artifacts in the surgical background, the program establishes a one-dimensional grayscale gradient profile within the ROI along a normal direction perpendicular to the needle body axis. It then uses a Gaussian first derivative operator to convolve the pixel grayscale values ​​on the profile, finding the extreme point of the most dramatic grayscale gradient change as the pixel-level center of the needle body boundary. The algorithm utilizes subpixel edge detection based on cubic spline interpolation. The detection operator performs quadratic curve fitting on the gray-level gradient of three consecutive pixels in the neighborhood around the extreme point, thereby solving for the precise zero-point location of the gradient extreme value at the sub-pixel level. This obtains a set of sub-pixel edge points of the bilaterally symmetrical metal needle body. Finally, the program performs spatial geometric constraint fitting on the high-precision point set using a random sampling consensus algorithm, eliminating outliers caused by the operator's fingers or tissue debris. The central symmetry axis between the two parallel lines is calculated, which is the axial spatial coordinate of a common medical puncture needle. The needle path planning and high-risk obstacle avoidance warning module, based on global alignment data and the corrected spatial pose matrix and axial spatial coordinates, divides and identifies the intervertebral space, ligamentum flavum, and epidural space. The system calculates and outputs a safe puncture needle path for the spinal canal, vertebral canal, and nerve roots. When the axial spatial coordinates of a standard medical puncture needle deviate from the safe puncture needle path by an angle greater than a preset threshold, the safe puncture monitoring and constraint module outputs a dual audible and visual warning signal and lowers the refresh rate of the perspective virtual-real overlay rendering module to block the guided procedure. If the deviation is within a preset threshold range, the perspective virtual-real overlay rendering module will simultaneously display the independent skeletal mesh, standard guide needle path, needle tip highlighting point, needle depth, and angle values ​​on the patient's back skin surface using augmented reality perspective virtual-real overlay rendering. This allows the physician to obtain the rigid morphology of each vertebra and the subcutaneous original shape even under dynamic changes in the patient's position during surgery. Millimeter-level alignment guidance based on bony anatomical structures achieves spatial causal synchronization that follows subtle body movements while excluding radiation beams. By transforming the epidermal strain tensor with nonlinear interlaminar shear characteristics into boundary physical constraints driving a multi-rigid-body kinematic chain topology network, the system architecture is built within a single data stream processing chain. This corrects the technical problem of mismatch between superficial slip deformation and spatial positioning of subcutaneous rigid bones, ensuring that the continuity of augmented reality graphics rendering is consistent with the physical motion laws that are decoupled from human physiological kinematics. Thus, by utilizing the displacement parameters of superficial anatomical sites, the variations in superficial spatial geometry are converted into an estimation of the six-dimensional pose state of independent subcutaneous vertebrae.

[0035] In extreme cases where the guidance process is forcibly stalled or blocked, to resolve the irreversible time blind spot contradiction between the lag in visual information refresh and the registration state transition in the background system, this invention establishes a time-sharing asynchronous decoupled anti-conflict mutation fallback protection mechanism at the software control layer. Specifically, when the rendering refresh rate is forcibly reduced to 1 Hz or lower to block visual cues, the system's real-time spatial image stream calculation thread and the core spatial registration calculation thread are physically isolated into independent parallel buses. Although the image on the front-end visual glasses freezes to warn the physician to stop needle insertion, the background registration state transition thread is not interrupted. Instead, it switches to a static suspension protection mode based on the previous safe 3D reference coordinate system. When the background algorithm automatically restores the registration state and recaptures the complete lossless reference point signal and calculates the new transient deformation strain tensor, the system will not directly refresh the glasses screen. Instead, the monitoring module will silently verify in the background whether the latest axis deviation angle of the puncture needle has converged back to within the preset angle threshold within three consecutive sampling cycles. Only after confirming that the angle is safe and the system's reconstructed pose matrix has achieved complete static self-consistency will the module send a high-priority wake-up pulse command to the rendering engine, instantly restoring the normal high-frequency refresh rendering flow of the augmented reality glasses. This will prevent the clinical puncture danger caused by blind operation or screen lag during visual screen lag, and ensure the closed loop of the control causal logic of the virtual-real alignment system.

[0036] Example 2: When the system is used in a puncture surgery on a patient with high body weight and accompanying lumbar spondylosis and intervertebral space narrowing, in order to quantitatively verify the gain of interlayer shear slip compensation on the registration accuracy of the rigid body spatial matrix of deep bone under dynamic conditions, the experimental design adopts an experimental environment composed of a rigid-flexible coupling spinal training platform with physical parameter feedback. The rigid-flexible coupling spinal training platform has built-in multi-axis mechanical and displacement sensors, and the technical specifications of the multi-axis mechanical and displacement sensors are limited to a spatial measurement range of 0mm to 500mm, a spatial resolution of 0.01mm, a measurement accuracy of 0.05mm, and an output sampling rate of 500Hz, which are used as the raw data source. In the experimental environment, a six-dimensional attitude optical tracking module... The spatial attitude sampling frequency setting of the block includes transient bandwidth analysis of respiratory fluctuations, heartbeat micro-vibrations, and sudden muscle spasm motion interference signals. The engineering trade-off point for its parameter selection lies in the balance between the real-time performance of high-frequency dynamic capture and the data processing load and processor heat dissipation of the real-time coordinate registration software. Its textual decision rule is established that the spatial attitude sampling frequency is greater than 20 times the highest transient bandwidth of sudden muscle spasm motion interference, thereby avoiding aliasing of sampling signals and keeping the algorithm convergence delay below 5ms. Accordingly, when the highest transient bandwidth of the simulated sudden muscle spasm motion interference reaches 10Hz, the spatial attitude sampling frequency is determined to be a fixed engineering value of 200Hz according to the aforementioned decision rule.

[0037] To verify the synergistic effect between the fat suppression ultrasound three-dimensional spinal reconstruction module and the respiratory, cardiac, and body motion three-level filtering and skin deformation compensation module, a multi-dimensional control system consisting of an experimental group, a first control group, and a second control group was designed. The experimental group employed a complete technical solution including automatic subcutaneous fat layer segmentation and interlayer shear slip transfer matrix deformation compensation. The first control group removed the interlayer shear slip transfer matrix to isolate interference from epidermal elastic displacement. Furthermore, to simulate the non-ideal electromagnetic environment of the operating room, Gaussian white noise interference with a signal-to-noise ratio of 20dB and 50Hz power frequency harmonic disturbances were actively injected into the raw input data streams of all channels. Based on this, the simulated subcutaneous fat layer of the rigid-flexible coupling spinal training platform was adjusted... Three difficulty gradients were constructed for different fat layer thicknesses, ranging from 15mm, 35mm, to 60mm. In the 15mm fat layer interference condition, the average target registration error measured in the first control group was 3.42mm, while the experimental group's target registration error decreased to 0.52mm. When the fat thickness increased to 35mm, the target registration error in the first control group rose to 5.87mm due to the superposition of ultrasound noise and skin slippage caused by the thick fat layer. The experimental group, however, stabilized the target registration error at 0.74mm by using an interlayer shear slip transfer matrix to separate epidermal slippage and combining it with a degenerative deformity topology fitting program. When facing a 60mm extremely thick fat layer accompanied by body movement interference, the depth of the first control group... The mismatch caused by heterogeneous motion in the anatomical layers increased the target registration error to 9.13 mm. However, the experimental group relied on the rigid physical constraints formed by the physical torsional stiffness of the intervertebral discs between adjacent vertebrae and the tensile limit of the ligamentum flavum in the multi-rigid-body kinematic chain topology network, forcibly locking the spatial pose matrix from the first to the fifth lumbar vertebrae, thus limiting the final target registration error to 1.12 mm. This quantitative data confirms that the complete technical solution produces nonlinear error suppression performance exceeding the sum of individual features. The system has a pre-calibration control module used to determine the step response parameters of the aforementioned slip calibration compensation weights and verify the boundary thresholds. Based on the nonlinear viscoelastic constitutive model in continuum mechanics and the shear strain transmission law of multilayer connective tissue, after system startup... The pre-calibration process is guided by a rigid-flexible coupling spinal training platform containing multi-axis mechanical and displacement sensors. The platform's spatial measurement range is 0mm to 500mm, with a measurement accuracy of 0.05mm and an output sampling rate of 500Hz. Discrete step thrusts of 10N to 50N are applied to the simulated back skin surface of the training platform via an external force application device. The multi-axis mechanical and displacement sensors simultaneously acquire the relative displacement parameters of the surface and deep bone models. The pre-calibration control module reads 10 consecutive time-series sampling points and constructs a sliding analysis interval. It calculates the discrete statistical variance of the spatial displacement vector on the body surface within the sliding analysis interval to obtain the equivalent spatial fluctuation amplitude. Test data shows that when the discrete statistical variance is less than or equal to 0...At 12mm, since the values ​​are within the system electrical noise bias variance range of the infrared optical tracking camera, the slip calibration compensation weight is set to a fixed base value of 0.10 to suppress high-frequency noise interference and maintain the coordinate stability of the virtual skeleton mesh unit greater than 99.1%. When the discrete statistical variance is within the asymptotic working range of 0.12mm to 4.5mm, the slip calibration compensation weight increases monotonically and linearly with the variance, keeping the response delay of dynamic pose iteration calculation within 5ms. When the discrete statistical variance is greater than 4.5mm, it indicates that the simulated subcutaneous tissue is experiencing violent body movement beyond the background of micro-motion. At this point, the slip calibration compensation weight is forced to saturate to the maximum limit value of 1.00, with full weight response to true slip, eliminating coordinate contamination of the deep skeleton registration matrix by large-scale shear displacement.

[0038] To assess the rationality of the numerical range of the safety control boundary threshold in virtual-real overlay rendering, the experimental scheme, while maintaining a fat thickness of 35mm and a pose sampling frequency of 200Hz, performed gradient optimization tests within the range of 1.0mm to 10.0mm for the safety control boundary threshold. Test data showed that when the safety control boundary threshold was lowered to 1.0mm (less than the lower limit of 2.0mm), the coordinate registration and real-time rendering software modules misinterpreted baseline breathing micro-motions and high-frequency noise from sensors as interlayer slippage, frequently outputting slippage calibration compensation weights. This resulted in frequent high-frequency correction jitter in the virtual skeleton mesh units, causing the coordinate registration stability to drop to 82.4%. When the safety control boundary threshold was set within the preset working window of 2.0mm to 8.0mm... Specifically, when the median value of 4.5mm is selected, the slip deformation compensation network distinguishes between basic background disturbances and shallow and deep non-affine slip, outputting a target registration error of 0.74mm and a target alignment success rate of 99.1%. However, when the safety control boundary threshold is set to 10.0mm, which is greater than the upper limit of 8.0mm, the system experiences a perception lag when facing transient large-scale epidermal slip caused by sudden body movement. This causes a large shear displacement to directly bypass the deformation compensation module and contaminate the deep bone pose, increasing the transient target tracking error to 4.23mm. This performance inflection point confirms that the range of 2.0mm to 8.0mm is a control boundary that balances noise tolerance and slip sensitivity, providing direct experimental evidence for the numerical limitation range in the weights.

[0039] When processing technical aspects involving positioning signal acquisition and medical needle video streams, the system incorporates a data anonymization and de-identification mechanism to meet engineering requirements for privacy protection. Before the spatial three-dimensional reference coordinate system registration module receives the optical relative displacement of the 12th thoracic anatomical site, the 3rd lumbar anatomical site, and the 1st sacral anatomical site, or before the needle path planning and high-risk obstacle avoidance warning module receives sub-pixel edge data from the needle contour recognition unit, the system performs anonymization and de-identification transformation of all coordinate parameters through a preset hash image de-identification interface, blocking the backtracking to the original natural human characteristics. Furthermore, all motion chain topology calculations and virtual-real overlay rendering scheduling are completed within the graphics processor of the local workstation, prohibiting any original physical signals or sensitive geometric features from being uploaded to the external cloud network. The final output of the data processing chain is not a clinical medical diagnosis conclusion, but a set of neutral indicators used for technical reference, specifically including the quantitative value of spatial registration variance and real-time quantitative risk parameters used to characterize guidance deviations. Combined with angle thresholds and audio-visual cues, closed-loop constraints are completed to ensure that the data processing flow fully complies with the restrictions on non-diagnostic technical data processing.

[0040] Example 3: This example combines Figures 1 to 2 This document describes a radiation-free ultrasound three-dimensional reconstruction AR spinal puncture real-time fluoroscopic guidance system, such as... Figure 1 As shown, the fat-suppressing ultrasound 3D spinal reconstruction module outputs anatomical features and a 3D model, and transmits the output data to the respiratory, cardiac, and body motion three-level filtering compensation module. Simultaneously, the six-dimensional posture optical tracking module acquires the transient deformation strain tensor and transmits the acquired data to the respiratory, cardiac, and body motion three-level filtering compensation module. The respiratory, cardiac, and body motion three-level filtering compensation module corrects the model spatial pose matrix and transmits the corrected matrix data to the needle path planning and high-risk obstacle avoidance warning module. In addition, the infrared vision dual-redundant puncture needle tracking module captures the puncture needle axis coordinates and transmits the captured data to the needle path planning and high-risk obstacle avoidance warning module. Finally, the needle path planning and high-risk obstacle avoidance warning module outputs a safe puncture needle path.

[0041] like Figure 2 As shown, the system acquires the transient deformation strain tensor and separates the epidermal slip shear quantity from the input transfer matrix. After loading the epidermal slip shear quantity into the kinematic chain network, the spatial pose matrix is ​​corrected. Then, it is applied to the virtual model in situ to output a synchronous independent skeletal mesh. The virtual and real superimposed rendering is performed according to the puncture needle path. If the detected deviation angle is greater than the threshold, a dual audible and visual warning signal is output. The dual audible and visual warning signal blocks the guidance process by limiting the refresh rate on the one hand, and automatically restores the registration state to return to the acquisition of the transient deformation strain tensor on the other hand.

[0042] Example 4: When the system is faced with a patient with high body weight and narrowed intervertebral space, nonlinear anisotropic dissipation occurs due to the connective tissue between the thick fat layer of the epidermis and the deep bone structure. The patient's respiratory movements and local back muscle tremors produce non-affine random shear deformation on the back skin surface. This causes execution deviation and time delay between the shallow spatial coordinates tracked by the six-dimensional posture optical tracking module and the pose of the subcutaneous rigid lumbar vertebral body. At this time, the determination of the slip calibration compensation weight needs to establish a predictable quantitative deduction rule to balance the noise fault tolerance control of high frequency measurement and the real-time capture performance of large displacement slip, and avoid the position synchronization accuracy from being mismatched due to noise interference caused by a single fixed weighting factor.

[0043] To resolve the tracking delay conflict between shallow random stretching deformation and coordinate smoothing filtering, a three-stage filtering module for respiratory, cardiac, and body motion, along with a skin deformation compensation module, receives transient deformation strain tensors continuously transmitted by a six-dimensional attitude optical tracking module at a sampling frequency of 200Hz. Within a preset time window, 10 consecutive time-series sampling points are selected to construct a sliding analysis interval. The discrete statistical variance of the body surface spatial displacement vector within this interval is calculated. This calculated discrete statistical variance is compared to a preset first noise control threshold of 0.12mm and a safety control boundary threshold of 4.5mm. When the discrete statistical variance is less than or equal to 0.12mm, the slip calibration compensation weight is set to a fixed base value of 0.10, allowing the coordinate registration software to perform filtering for smoothing. To eliminate logical conflicts caused by direct comparisons between different dimensions and units, the system introduces a dimensional alignment physical mapping interface based on root mean square transformation. Specifically, the physical dimension of the calculated discrete statistical variance, which is originally represented by the square of its length, is automatically extracted by the program before being numerically compared with the first noise control threshold and the safety control boundary threshold. This reverses the second-order statistical variance scalar into an equivalent spatial fluctuation amplitude scalar with a first-order linear length attribute and a unit of standard millimeters. Through this dimensionality reduction and dimensional normalization mapping, the statistical features originally belonging to the square dimension are losslessly transformed into the same logical dimension equivalent to the preset length threshold, thereby eliminating the noise. The logical discontinuity caused by direct cross-dimensional comparison ensures the absolute accuracy and closed-loop self-consistency of subsequent conditional decision branches at the software execution level. When the discrete statistical variance is greater than 0.12mm and less than or equal to 4.5mm, the system runs a linear incremental deduction program, controlling the slip calibration compensation weight to monotonically and linearly increase with the increase of discrete statistical variance, so that its value smoothly transitions from the basic value of 0.10 to the preset maximum weighting coefficient of 0.85, thereby improving the response rate to the amount of epidermal slip while ensuring the stability of filtering. When the discrete statistical variance is greater than 4.5mm, the system executes the boundary coverage rule, sets the slip calibration compensation weight to the maximum saturation limit value of 1.00, skips the filtering attenuation stage, and directs the calculated effective deep driving torque to The parameters are loaded into a multi-rigid-body kinematic chain topology network, which drives the network to perform joint iterative optimization calculations on the spatial pose matrices of the first to fifth lumbar vertebrae in three-dimensional space. This ensures that the parameter determination process is clear and procedural. By converting the interval variation of discrete statistical variance into a dynamic step response of slip calibration compensation weights, the system architecture is built within a single data flow processing chain. This corrects the technical problem of mismatch between superficial slip deformation and spatial positioning of subcutaneous rigid bones, ensuring that the continuity of augmented reality graphics rendering is consistent with the physical motion law that is decoupled from human physiological kinematics. Thus, by utilizing the displacement parameters of superficial anatomical sites, the variation of superficial spatial geometric features is converted into an estimate of the six-dimensional pose state of independent subcutaneous vertebrae.

[0044] The numerical windows for the above parameters and thresholds were determined based on large-sample static and dynamic calibration experiments of the rigid-flexible coupling spinal training platform under various extreme body positions. Specifically, the preset first noise control threshold of 0.12 mm represents the maximum high-frequency instrument system electrical noise variance of the six-dimensional attitude optical tracking module under normal static conditions. If the variance is lower than this critical lower limit, it indicates that the displacement is caused only by sensor noise. At this time, setting the slip calibration compensation weight to a fixed base value of 0.10 can maximize the smoothing effect of the first-order low-pass filter and prevent the virtual skeleton mesh unit from generating high-frequency, high-intensity correction jitter. The preset safety control boundary threshold of 4.5 mm is based on multiple high-weight... The maximum engineering mean of non-affine slippage caused by superficial subcutaneous tissue deformation in the superficial skin under arched back puncture conditions is determined. If the variance is higher than this critical upper limit, it indicates that the patient has experienced sudden and severe body movement or muscle spasm, and the body surface slippage has completely broken away from the micro-motion background. At this time, the weights are forced to saturate to the maximum limit value of 1.00, skipping the filter lag stage, and responding to the real slippage with full weights. This prevents large-amplitude abrupt shear displacement from bypassing the system and contaminating the bone registration matrix. In the asymptotic working range of 0.12 mm to 4.5 mm, the setting of the maximum weighting coefficient of 0.85 balances the filter convergence speed and the anti-step overshoot performance, so that the system delay in linear incremental extrapolation is always less than 5 milliseconds.

[0045] Example 5: When the system faces patients with different anatomical heterogeneities and needs to establish the initial zero position of the three-dimensional spatial reference coordinate system, a pre-calibration program is deployed on-site to eliminate the transmission deviation caused by individual anatomical differences. The six-dimensional posture optical tracking module acquires the initial pose of the anatomical sites of T12, L3, and S1 under static conditions and inputs them into the baseline calibration software. The multi-rigid-body kinematic chain topology network is controlled to align the zero position, and a discrete step thrust is applied to the back skin surface through an external force application device. The displacement is recorded using an infrared vision dual-redundant puncture needle tracking module, thereby calculating the non-affine shear attenuation factor and the first-order displacement transmission damping boundary of the corresponding patient. Finally, the basic values ​​are written into the initial parameter register of the interlaminar shear slip transfer matrix, and the baseline data is output.

[0046] During the guidance process, the respiratory, cardiac, and body motion three-level filtering and skin deformation compensation modules correct interlayer non-affine slippage based on the baseline values ​​in the initial parameter register. When the needle path planning and high-risk obstacle avoidance warning modules calculate the puncture path, the system performs hash anonymization and desensitization on the collected optical positioning parameters to protect privacy. At the same time, the system controls the fluoroscopic virtual-real overlay rendering module to simultaneously overlay and display the independent skeletal mesh and standard guide needle path on the back skin surface, enabling physicians to obtain millimeter-level guidance under the dual correction of static baseline and dynamic compensation. Under the condition of excluding radiation beam radiation, the system outputs real-time quantitative risk parameters that characterize guidance deviation, thereby improving the stability of the system.

[0047] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0048] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A radiation-free ultrasound three-dimensional reconstruction AR spinal puncture real-time fluoroscopic guidance system, characterized in that the system... include: The fat-suppressed ultrasound 3D spine reconstruction module is used to segment the subcutaneous fat layer and bone region, and output anatomical boundary features and a 3D model of the spine. The six-dimensional attitude optical tracking module is used to acquire the transient deformation strain tensor of the spatial triangular plane formed by the 12th thoracic anatomical site, the 3rd lumbar anatomical site, and the 1st sacral anatomical site. The respiratory, heartbeat, and body motion three-level filtering and skin deformation compensation module is connected to the six-dimensional attitude optical tracking module. It is used to input the transient deformation strain tensor into the interlayer shear slip transfer matrix containing a preset attenuation factor, separate the epidermal slip shear amount and the deep driving torque vector, and correct the spatial pose matrix of the three-dimensional model of the spine according to the deep fascia interlayer nonlinear slip constraint. An infrared vision dual-redundant puncture needle tracking module is used to capture the spatial coordinates of the axis of a common medical puncture needle. The needle path planning and high-risk obstacle avoidance early warning module is connected to the respiratory, heartbeat and body motion three-level filtering and skin deformation compensation module and the infrared vision dual-redundant puncture needle tracking module, respectively. It is used to calculate and output the safe puncture needle path based on the corrected spatial pose matrix and axis spatial coordinates.

2. The radiation-free ultrasound three-dimensional reconstruction AR spinal puncture real-time fluoroscopic guidance system according to claim 1, characterized in that, The fat-suppressed ultrasound 3D spinal reconstruction module includes an automatic fat and bone segmentation unit and a spinal degeneration and deformity topology fitting unit. The automatic fat and bone segmentation unit is used to eliminate acoustic noise interference on the spinous process, lamina, and intervertebral space, output anatomical boundary features, and input them into the 3D spinal model. The spinal degeneration and deformity topology fitting unit is used to obtain the discrete morphological structures of bone hyperplasia, intervertebral space narrowing, and scoliosis to construct an individualized skeletal anatomical topology.

3. The radiation-free ultrasound three-dimensional reconstruction AR spinal puncture real-time fluoroscopic guidance system according to claim 1, characterized in that, The six-dimensional attitude optical tracking module continuously acquires the relative three-dimensional spatial displacement between the thoracic 12 anatomical sites, the lumbar 3 anatomical sites, and the sacral 1 anatomical sites, and calculates the transient deformation strain tensor of the spatial triangular plane.

4. The radiation-free ultrasound three-dimensional reconstruction AR spinal puncture real-time fluoroscopic guidance system according to claim 1, characterized in that, When the respiratory, heartbeat, and body motion three-level filtering and skin deformation compensation module uses the interlayer shear slip transfer matrix to solve the deep driving torque vector, it uses a preset attenuation factor to reduce the amplitude of the epidermal slip shear and separate the deep driving torque vector.

5. The radiation-free ultrasound three-dimensional reconstruction AR spinal puncture real-time fluoroscopic guidance system according to claim 1, characterized in that, The infrared vision dual-redundant puncture needle tracking module includes an infrared optical marker capture unit and a needle body contour recognition unit. The infrared optical marker capture unit is used to capture the six-dimensional spatial coordinates of the infrared marker fixed at the handle of a common medical puncture needle to calculate the spatial pose. The needle body contour recognition unit is used to extract the edge contour of the metal needle body of a common medical puncture needle to obtain the axial spatial coordinates when the infrared marker is obstructed by the line of sight.

6. The radiation-free ultrasound three-dimensional reconstruction AR spinal puncture real-time fluoroscopic guidance system according to claim 1, characterized in that, The system also includes a spatial three-dimensional reference coordinate system registration module; the spatial three-dimensional reference coordinate system registration module is connected to the six-dimensional attitude optical tracking module and the infrared vision dual-redundant puncture needle tracking module respectively, and is used to uniformly integrate the anatomical boundary features and axis spatial coordinates into the spatial three-dimensional reference coordinate system constructed by the thoracic 12 anatomical site, the lumbar 3 anatomical site and the sacral 1 anatomical site for spatial alignment, and output spatial global alignment data.

7. The radiation-free ultrasound three-dimensional reconstruction AR spinal puncture real-time fluoroscopic guidance system according to claim 1, characterized in that, The system also includes a virtual skeleton mesh unit spatial synchronization module; the virtual skeleton mesh unit spatial synchronization module is connected to the respiratory, heartbeat and body motion three-level filtering and skin deformation compensation module and the spatial three-dimensional reference coordinate system registration module, respectively, to receive spatial global alignment data, and to perform spatial synchronization update of the corresponding independent skeleton mesh units in the virtual model according to the spatial pose matrix, and output synchronized independent skeleton meshes.

8. The radiation-free ultrasound three-dimensional reconstruction AR spinal puncture real-time fluoroscopic guidance system according to claim 1, characterized in that, The system also includes a perspective virtual-real overlay rendering module; the perspective virtual-real overlay rendering module is connected to the virtual skeleton mesh unit spatial synchronization module and the needle path planning and high-risk obstacle avoidance early warning module, respectively, to perform augmented reality spatial virtual-real overlay rendering display on the patient's back skin surface by synchronizing the independent skeleton mesh and the safe puncture needle path.

9. The radiation-free ultrasound three-dimensional reconstruction AR spinal puncture real-time fluoroscopic guidance system according to claim 1, characterized in that, The system also includes a safe puncture monitoring and constraint module; the safe puncture monitoring and constraint module is connected to the needle path planning and high-risk obstacle avoidance early warning module, which is used to obtain the safe puncture needle path, and output a dual warning signal of sound and light when the axial spatial coordinate of the ordinary medical puncture needle deviates from the safe puncture needle path by an angle greater than a preset angle threshold. At the same time, it limits the refresh frequency of the three-dimensional space virtual and real superposition rendering display in the system to block the guidance process.

Citation Information

Patent Citations

  • Computer-implemented method of augmented reality spinal rod planning and bending for navigating spinal surgery

    CN116847799A