A method of virtual positioning CT-guided lung biopsy

Abstract

The application relates to the technical field of medical devices and computer-aided surgical navigation, and discloses a virtual positioning CT-guided lung biopsy method, which utilizes real-time acquired body surface point cloud data to drive a biomechanics model, deduces a displacement field of lung internal grid nodes, and generates a virtual lung model which changes with respiratory motion. Through a fiber grating sensing array embedded in a puncture needle, the three-dimensional shape of the needle body is solved in real time, and an axial force strain is separated. The system monitors the first derivative characteristics of the strain, and when a mechanical event representing penetration of an anatomical boundary is identified, a spatial deviation vector of the physical coordinates of the needle tip and the virtual predicted coordinates is calculated. A local space calibration field is constructed by using a radial basis function, the virtual lung model is non-rigidly corrected based on the deviation vector, and a dynamic navigation image is generated. The application improves the positioning accuracy of lung puncture navigation by combining body surface deduction and puncture mechanical feedback correction model prediction error without increasing the radiation dose.

Description

Technical Field

[0001] This invention relates to the field of medical devices and computer-aided surgical navigation technology, specifically a method for lung biopsy under virtual positioning CT guidance. Background Technology

[0002] Percutaneous lung biopsy is a key technique for the qualitative diagnosis and minimally invasive treatment of pulmonary nodules. Due to the significant flexibility and deformability of lung tissue, the patient's unavoidable respiratory movements cause real-time displacement and deformation of lesions, blood vessels, bronchi, and other anatomical structures. This dynamic characteristic makes it difficult for the puncture path determined based on preoperative static CT images to completely coincide with the real-time anatomical environment during the procedure, increasing the risk of puncture failure or accidental damage to important blood vessels.

[0003] Currently, the main clinical practice using CT-guided scanning typically requires patients to hold their breath during scanning and puncture. However, patients often struggle to accurately replicate the same breath-holding amplitude and lung volume across multiple procedures, leading to a spatial mismatch between the actual target location and the planned image location. While CT fluoroscopy can provide real-time tomographic guidance and solve the motion tracking problem, this method requires continuous X-ray exposure, significantly increasing the radiation dose for both patients and medical staff, thus limiting its widespread use in routine surgeries.

[0004] To address radiation concerns, electromagnetic navigation technology has been introduced into the field of pulmonary intervention. This technology tracks the position of implanted sensors using electromagnetic fields. However, electromagnetic tracking is highly susceptible to interference from ferromagnetic materials such as metal surgical instruments, CT scanner gantry, and the operating table, leading to drift in positioning accuracy. Furthermore, dedicated electromagnetic guidance consumables are expensive.

[0005] In recent years, virtual augmented reality navigation technology based on surface optical monitoring combined with biomechanical models has gradually become a research hotspot. This type of technology captures the surface deformation of the patient's chest and abdomen using optical devices, and uses a continuum mechanics model to extrapolate the motion state of internal organs, thereby generating radiation-free dynamic virtual images. However, this extrapolation process from the surface to the interior is essentially an open-loop prediction. Limited by the complexity of biological tissue constitutive models and individual differences, the computational model will generate cumulative errors when extrapolating the displacement of deep lung tissue. Existing virtual navigation schemes lack a direct, measured feedback mechanism for deep anatomical boundaries after the puncture needle enters the body. They cannot perceive the real physical state when the puncture needle contacts specific tissues such as the pleura and lung parenchyma, resulting in the inability to obtain real-time intraoperative calibration of the model prediction results. Navigation accuracy often gradually decreases with increasing needle depth. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a virtual positioning CT-guided lung biopsy method, which solves the problems of target displacement caused by respiratory motion, high radiation dose from continuous CT scans, and susceptibility to interference with electromagnetic navigation in existing CT-guided lung biopsy processes.

[0007] To achieve the above objectives, the present invention provides the following technical solution: The first aspect of this invention provides a method for lung biopsy under virtual positioning CT guidance.

[0008] This method first constructs a baseline virtual model containing lung anatomy and a pre-defined puncture path using preoperative CT images. During surgical navigation, a surface optical monitoring device acquires real-time point cloud data of the patient's body surface, and a biomechanical model is used to establish a mapping relationship between surface deformation and internal anatomical displacement. This step registers the real-time surface point cloud with the baseline surface model to calculate the surface displacement field, which is then input as a boundary condition into a linear elastic continuum mechanical model. The system uses the Navier-Cauchy equation to describe the mechanical equilibrium state of lung tissue, and combines Young's modulus and Poisson's ratio parameters of the lung tissue to solve for the displacement field of the internal lung grid nodes, generating a virtual lung model that changes with respiratory motion.

[0009] This method utilizes a fiber optic grating sensor array embedded within the puncture needle to reconstruct the needle's morphology and sense its mechanical properties. The fiber optic grating sensor array comprises multiple fiber cores distributed circumferentially along the needle's cross-section. The system acquires the center wavelength drift of each fiber grating and, based on fiber optic coupled-mode theory, calculates the drift using the linear relationship between the center wavelength drift and strain and temperature. Multi-core differential calculations eliminate the influence of temperature variables and extract bending strain. Then, by integrating along the needle axis, the three-dimensional geometry of the needle and the physical coordinates of the needle tip are reconstructed. Simultaneously, common-mode calculations separate the axial resultant strain, which characterizes the axial tissue resistance experienced by the needle tip during puncture.

[0010] This method further achieves anatomical layer model calibration by monitoring the changing characteristics of axial resultant force strain. Because different anatomical tissues have different elastic moduli, the puncture needle will produce a specific mechanical response when breaking through the boundaries of each layer. The system calculates the first derivative of the axial resultant force strain with respect to time. When the absolute value of this first derivative exceeds a preset threshold, and the needle insertion depth calculated by the current fiber reconstruction is within the depth range of the anatomical feature boundary layer estimated based on the preoperative planned path, a perforation mechanical abrupt event is determined to have occurred.

[0011] Upon detecting the event, the system performs spatial deviation calculation and model correction. The physical coordinates of the needle tip, reconstructed by the fiber optic sensor array at that moment, are acquired, and the intersection point between this intersection and the predicted anatomical boundary is searched along the puncture needle axis in the virtual lung model. This intersection point is defined as the virtual predicted coordinates. The difference vector between the physical coordinates of the needle tip and the virtual predicted coordinates is calculated as the spatial deviation vector. Subsequently, a local spatial calibration field is constructed centered on the physical coordinates of the needle tip, and calibration weights are calculated using radial basis functions (RBF). These weights decrease exponentially with the Euclidean distance from the spatial point to the physical coordinates of the needle tip. The system then superimposes the product of the spatial deviation vector and the weights onto the original predicted coordinates of the virtual lung model, performing a local non-rigid correction to the anatomical model to align the virtual anatomical boundary with the physical position of the needle tip.

[0012] Finally, the reconstructed needle body shape is fused and rendered with the corrected deformation mapping model to generate a dynamic navigation image that includes a three-dimensional panoramic view and a cross-sectional view based on the needle tip coordinate system, showing the relative positional relationship between the needle tip and the lesion.

[0013] A second aspect of the present invention provides a virtual positioning CT-guided lung biopsy system.

[0014] The system includes a CT scanning device, a body surface optical monitoring device, a puncture needle with an embedded fiber Bragg grating sensor array, a fiber Bragg grating demodulator, a data processing terminal, and a display terminal.

[0015] CT scanning equipment is used to acquire preoperative imaging data of the patient. A surface optical monitoring device is configured to acquire real-time point cloud data of the patient's chest and abdomen. A fiber optic demodulator is connected to the puncture needle to acquire fiber optic reflection spectra and demodulate wavelength data.

[0016] The data processing terminal receives input data from the aforementioned devices and executes the navigation method described in the first aspect, including: deformation field deduction based on surface monitoring data, needle morphology reconstruction and strain separation based on wavelength data, membrane penetration event identification based on mechanical characteristic mutations, model non-rigid correction based on spatial deviation vectors, and generation of navigation images. The display terminal is used to present the corrected dynamic navigation interface and quantitative guidance indicators.

[0017] This invention utilizes surface monitoring to extrapolate and address organ displacement caused by respiratory movements. It also leverages the biomechanical sensing capabilities of the puncture needle to capture changes in mechanical signals as it penetrates anatomical boundaries, using this as a positional reference for closed-loop calibration of the virtual model. This method corrects target drift caused by model prediction errors without increasing intraoperative radiation dose, thus improving the real-time spatial accuracy of lung puncture navigation.

[0018] This invention provides a virtual positioning CT-guided lung biopsy method. It has the following beneficial effects: 1. This invention resolves the conflict between real-time dynamic tracking and radiation dose control during surgery by using surface monitoring data to drive a biomechanical model. Utilizing real-time acquired surface point clouds as boundary conditions, the displacement field of internal lung grid nodes is derived by solving the Navier-Cauchy equilibrium equations, generating a virtual lung model that updates in real-time with respiratory motion. This method allows surgeons to obtain images of internal anatomical structures synchronized with the patient's current respiratory state without requiring continuous CT scans, significantly reducing the patient's radiation exposure risk while effectively avoiding target point spatial drift caused by respiratory motion.

[0019] 2. This invention utilizes an embedded fiber optic grating sensor array to achieve simultaneous sensing of the morphology of the puncture needle and the mechanical response of the tissue. By demodulating and separating the wavelength drift data from the multi-core fiber, it is possible to simultaneously acquire the differential strain reflecting the bending morphology of the needle body and the axial resultant strain reflecting the tissue penetration resistance. This technical feature not only accurately reconstructs the physical coordinates of the needle tip when flexible deformation occurs during long-distance punctures, but also objectively determines whether the puncture needle has penetrated key anatomical layers such as the parietal pleura by utilizing the abrupt change characteristics of the axial resultant strain, providing objective mechanical judgment evidence for clinical operations beyond imaging.

[0020] 3. This invention establishes a closed-loop calibration mechanism for the anatomical model based on feedback from puncture biomechanical events, improving the absolute positioning accuracy of the navigation system. At the specific moment the system detects the puncture needle breaking through the anatomical boundary, it calculates the spatial deviation vector between the physical coordinates reconstructed by the optical fiber and the virtual coordinates derived from the model, and uses radial basis functions to construct a local calibration field to perform non-rigid correction on the virtual model. This mechanism uses the actual contact position during the physical puncture process as a benchmark, automatically correcting the estimation errors accumulated in the deep tissue derivation of the biomechanical model, ensuring a high degree of overlap between the virtual lesion location and the actual anatomical location in the key operational area. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the overall process of the method in an embodiment of the present invention; Figure 2 This is a schematic diagram of the hardware architecture connection of the surgical navigation system provided in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the deformation field deduction principle based on body surface monitoring in an embodiment of the present invention; Figure 4 This is a schematic diagram of the cross-sectional structure and strain separation principle of the fiber grating sensing array in an embodiment of the present invention; Figure 5 This is a line graph showing the mechanical signal response characteristics of the puncture needle penetrating the pleura in an embodiment of the present invention. Figure 6This is a schematic diagram illustrating the calibration principle of the anatomical model based on puncture biomechanical feedback in an embodiment of the present invention; Figure 7 This is a schematic diagram illustrating the display effect of the multi-view joint navigation interface provided in an embodiment of the present invention. Detailed Implementation

[0022] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Example

[0023] Please see the appendix Figure 1 - Appendix Figure 7 This invention provides a method for lung biopsy under virtual positioning CT guidance, comprising: Step S100 is used to establish the anatomical reference environment required for surgical navigation, in particular to provide a precise geometric reference for subsequent biomechanical calibration. To ensure that the navigation system can distinguish different tissue media and identify boundary crossing events during puncture, this step is specifically implemented through the following sub-steps S101 to S103.

[0024] Step S101: Acquire standardized preoperative imaging data. The system receives DICOM format data of the patient's chest from the CT imaging device via a data interface. To reduce baseline errors caused by respiratory motion, the CT data is thin-slice scan data acquired during the patient's breath-holding state at the end of inspiration, with the slice thickness preferably set between 0.625 mm and 1.25 mm. The system maps the read CT data to a unified three-dimensional Cartesian coordinate system, which is defined as the CT image coordinate system. In this coordinate system, each voxel point x has a unique gray value. The grayscale value is expressed in Hounsfield Units (HU).

[0025] Step S102: Perform multi-target anatomical structure segmentation. The data processing terminal uses an adaptive threshold segmentation algorithm to extract the geometric model of each anatomical tissue based on the physical differences in X-ray absorption rates of different human tissues.

[0026] For the skin surface, the system sets a skin threshold range (e.g., -700HU to -200HU) to extract the external contour of the human body. By triangulating and smoothing the contour data using Laplacian, a skin mesh model is generated, denoted as [model name missing]. The model It includes geometric features for subsequent registration with the optical point cloud.

[0027] For the lung parenchyma, the system extracts lung tissue using either a region growing algorithm or a graph cut-based segmentation algorithm. A point within the tracheal lumen is selected as a seed point, and a growth threshold range is set (e.g., -950 HU to -400 HU). By iteratively calculating its connected components, a three-dimensional voxel set of the lung parenchyma is obtained. For lesions within the lung parenchyma, the system segments them based on abrupt changes in local gray-level gradients to determine the centroid coordinates of the lesions. and lesion volume The specific algorithm implementation for the above image segmentation is a conventional technique for those skilled in the art and will not be elaborated upon here.

[0028] Step S103: Extract the anatomical boundary layer. This is a crucial step in establishing the biomechanical-geometric relationship. After lung parenchyma segmentation, the system needs to accurately locate the potential interface between the parietal and visceral pleura. Due to the resolution limitations of CT images, the pleural layer typically appears as a high-contrast edge between the lung field and the chest wall soft tissue.

[0029] The system processes the set of lung parenchyma voxels obtained in step S102. Morphological expansion and edge detection were performed to extract the outermost envelope surface of the lung parenchyma. The system defines this envelope surface as the anatomical feature boundary layer. Mathematically, this boundary layer It can be represented as a set of points that satisfy a specific gradient condition: ; in, This represents the gray-level gradient vector at voxel point x. The preset gradient magnitude threshold, This indicates the boundary of the lung parenchyma assembly. This anatomical boundary layer... In subsequent steps, this surface will serve as the geometric reference plane for determining whether the puncture needle has entered the pleural cavity. Specifically, when the physical coordinates of the needle tip coincide with this surface, a sudden change in the mechanical signal is expected. Through the above processing, the system completes the conversion from the original tomographic image to a parametric anatomical model that can be used for navigation calculations.

[0030] Step S200 aims to address the mismatch between static CT images and real-time intraoperative anatomical states caused by the patient's respiratory movements. This step utilizes external optical observation data to drive the deformation of the internal anatomical model and calculates the real-time position of the lung tissue by solving an inverse biomechanical problem, specifically through the following sub-steps S201 to S203.

[0031] Step S201: Perform spatial registration and displacement extraction of real-time body surface data. The body surface monitoring device acquires real-time point cloud data of the patient's chest at a preset sampling frequency (e.g., 30Hz). The system uses the Iterative Closest Point (ICP) algorithm to process the real-time point cloud data. Compared with the skin mesh model generated in step S100 Rigid registration is performed to eliminate overall rigid displacement caused by patient positional changes. After rigid alignment is completed, the system performs registration for each node on the mesh model. In real-time point cloud Search for its nearest neighbor and calculate the displacement vector between them. This set of displacement vectors constitutes the surface deformation boundary conditions required to solve the volumetric deformation at the current moment. The specific iterative convergence process of the ICP registration algorithm is well-known in this field and will not be elaborated upon here.

[0032] Step S202: Construct a biomechanical-based surface-to-internal deformation mapping relationship. To infer the displacement of lung tissue from surface displacement, the system treats the chest as a linear elastic continuous medium and establishes governing equations. The system sets boundary conditions based on human anatomy: the displacement of skin nodes on the body surface... The system uses a non-zero Dirichlet boundary condition, while setting the spinal and posterior back skeletal regions as zero-displacement boundary conditions. The Navier-Cauchy equation is used to describe the elastic deformation behavior of the lung tissue, and its governing equations are as follows: ; in, This represents the displacement vector of any point x within the body. Let λ be the Hamiltonian operator, and f be the body force (which can be set to 0 to ignore the influence of gravity). λ and μ are Lamé constants, which are determined by Young's modulus E and Poisson's ratio ν of lung tissue, with the following specific relationship: ; The system employs the finite element method (FEM) to numerically solve the aforementioned partial differential equations. By discretizing the lung parenchyma into a tetrahedral mesh and constructing a global stiffness matrix K, the differential equations are transformed into a linear system of equations KU=F for solution, thereby obtaining the predicted displacement field of all voxel points within the lung parenchyma. This process establishes a surface-to-interior deformation mapping relationship.

[0033] Step S203: Generate a virtual lung model and predicted boundaries. The system will use the predicted displacement field calculated in step S202. This is applied to the static anatomical model constructed in step S100. For each voxel point in the static model... Its predicted position at the current time t Updated to: ; Based on the updated coordinates, the system renders and generates a virtual lung model reflecting the current respiratory state in real time. Specifically, the system applies the anatomical feature boundary layer defined in step S103. Perform the same displacement transformation to generate the predicted pleural boundary at the current moment. This predicts the pleural boundary. This will serve as a benchmark for mechanical calibration in subsequent steps. Through these steps, the system estimates the dynamic location of intrapulmonary anatomy based solely on surface monitoring data without requiring additional X-ray scans.

[0034] Step S300 is used to acquire the three-dimensional spatial orientation of the puncture needle in the body in real time without relying on X-ray fluoroscopy. This step utilizes an embedded fiber optic sensing array as a shape sensing element and reconstructs the needle morphology by solving the distribution characteristics of the grating wavelength, specifically through the following sub-steps S301 to S303.

[0035] Step S301: Perform wavelength demodulation and multidimensional strain separation. The fiber optic demodulator transmits broadband light at a high frequency (e.g., 100Hz to 1kHz) to the fiber optic sensing array and receives the reflected signal. The system reads the center wavelength shift at the k-th grating sensing point in real time. (Where i=1,2,3 represent the three fiber cores distributed along the circumference). Based on the fiber grating sensing principle, wavelength drift is a coupling function of strain and temperature. To achieve accurate shape reconstruction and provide data for subsequent mechanical calibration, the system needs to separate the bending strain from the axial average strain. Utilizing the geometric symmetry of the three fiber cores on the same cross-section, the system constructs the following set of decoupling equations: ; Since the three fiber cores are close together and exposed to the same heat environment, the system eliminates the temperature term using a differential algorithm. The influence of this directly solves for the pure mechanical strain of each fiber core. Specifically, the system calculates the average strain of the three fiber cores, defined as the axial resultant strain. : ; The axial resultant strain Instead of being used for shape calculations, it will serve as a key input parameter in step S400 to identify membrane piercing mechanical events, reflecting the axial compressive or tensile resistance experienced by the needle body.

[0036] Step S302: Calculate the local geometric parameters along the needle axis. Based on the Euler-Bernoulli Beam Theory, the system uses the differential strain of each fiber core to derive the local curvature of the needle body at arc length s. and bending direction angle Let r be the radius of the fiber core from the neutral axis of the needle body. The specific calculation formula is as follows: ; ; By performing the above calculations on all grating sensing points distributed along the needle body and employing a spline interpolation algorithm, the system obtains a continuous curvature function along the entire length of the needle body. and bending angle function .

[0037] Step S303: Reconstruct the three-dimensional spatial curve of the needle body. The system establishes a local coordinate system with the handle of the puncture needle as the origin, and defines a moving frame (Frenet-Serret Frame) on the needle body curve, containing the tangent vector. Normal vector and binormal vector The spatial morphological evolution of the needle body follows the following set of differential equations: ; in The deflection is determined by the rate of change of the bending direction angle along the axial direction, i.e. The system uses the fourth-order Runge-Kutta method to numerically integrate the above differential equation, integrating from the needle tail to the needle tip, thereby obtaining the three-dimensional coordinates of any point on the needle body. : ; The endpoint of integration is the physical coordinate of the needle tip. Through the above steps, the system converts the optical signal into a visualized geometric model, thereby reconstructing the three-dimensional morphology of the puncture needle. This process is radiation-free and resistant to electromagnetic interference, making it suitable for the complex electromagnetic environment of a CT scan room.

[0038] Steps S400 and S500 constitute the core closed-loop calibration mechanism of this invention. Addressing the issue of decreased prediction accuracy in deep tissues using a purely surface-driven model, this section utilizes in-situ mechanical information generated by the interaction between the puncture needle and human tissue to dynamically correct the virtual anatomical model, specifically through the following sub-steps S401 to S403.

[0039] Step S401: Real-time identification of mechanical events penetrating the anatomical feature boundary layer. The system reads the axial resultant force and strain calculated in step S301 at high frequency. And calculate its first derivative with respect to time. Because the parietal pleura has a high elastic modulus (stiffness) while the inflated lung parenchyma has a low elastic modulus, the axial resistance experienced by the needle tip decreases abruptly the moment it penetrates the pleura and enters the lung parenchyma. This results in a negative abrupt change in the axial strain sensed by the fiber optic grating. The system uses a sliding window algorithm to monitor this abrupt change and defines a perforation event trigger function. : ; in, A preset strain rate threshold is used to distinguish tissue layer penetration signals from normal operating noise; The insertion depth calculated for the current fiber reconfiguration; This is the pleural depth range estimated based on the preoperative CT-planned path. By introducing depth range constraints, the system can effectively filter out spurious signals generated during skin or muscle punctures, ensuring that the calibration mechanism is activated only when the needle tip reaches the vicinity of the anatomical boundary layer (pleura).

[0040] Step S402: Calculate the calibration deviation vector between the physical and virtual spaces. When the system detects... At that time, lock the current moment as the event moment. The system immediately acquires the physical coordinates of the needle tip reconstructed by the FBG sensor array at this moment, denoted as... Simultaneously, in the current virtual lung model generated in step S203, the system searches for the boundary between the puncture needle and the predicted pleura along the axial ray. The intersection point is denoted as the virtual predicted coordinate. The virtual predicted coordinates represent the position where the system believes the needle tip should contact the pleura in an uncalibrated, purely surface-driven model. The system calculates the spatial deviation vector between the two points. : ; The deviation vector The cumulative geometric error of the surface-in-body deformation mapping model in deep anatomical regions was quantitatively characterized.

[0041] Step S403: Construct a local spatial calibration field and correct the model. To eliminate the aforementioned errors and maintain the topological continuity of the anatomical structure, the system should not simply translate the entire model, but rather use a non-rigid deformation field to locally correct the model. The system constructs a spatial calibration field based on the Gaussian radial basis function (RBF). For any voxel point x in the virtual lung model, its corrected coordinates... The calculation is as follows: ; in, The original predicted coordinates calculated based on the body surface in step S203. σ represents the Euclidean distance from the voxel point to the puncture point, and σ is the radius of influence parameter (attenuation coefficient). This calibration algorithm ensures that anatomical structures near the puncture point are strongly pulled towards their true physical location. In regions far from the puncture path, the model deformation gradually decays to the original predicted state, thus avoiding global image distortion. This is achieved by adjusting the coordinates... The system re-renders and corrects the deformation mapping model, enabling the navigation image to automatically achieve precise "alignment" when the puncture needle contacts key anatomical layers.

[0042] Step S600 is used to visually present the spatial relationship between the calibrated anatomical structure and the puncture needle to the operator, providing visual guidance and feedback. This step integrates the needle morphology reconstructed in step S303 with the virtual anatomical model corrected in step S503, and performs data fusion and graphical rendering, specifically through the following sub-steps S601 to S603.

[0043] Step S601: Perform spatial registration and scene fusion of multimodal data. The system establishes a unified rendering scene in the graphics processing unit (GPU). The corrected virtual lung model output in step S503 is then used. Volume rendering was employed, using an opacity transfer function to set bones to high-opacity white, lung parenchyma to semi-transparent, and lesions to bright red to enhance anatomical depth. Simultaneously, the set of points along the central axis of the puncture needle calculated in step S303 was used... The needle is rendered as a cylindrical primitive within the same scene. To visually display the real-time deformation of the needle, the system renders the puncture needle model in a color with high contrast to the background tissue (e.g., green), and dynamically updates the bending shape of the cylinder based on the curvature data calculated from the optical fiber.

[0044] Step S602: Construct a navigation view interface with multi-perspective linkage. To meet the requirements of clinical operation for spatial positioning accuracy, the system generates a composite navigation interface containing four sub-windows. The main window displays the above-mentioned three-dimensional panoramic view, supporting rotation and observation from any angle. The remaining three sub-windows respectively display real-time multi-planar reconstruction (MPR) images based on the needle tip coordinates . The system resamples the voxel data in real time according to the current tangential vector and normal vector of the needle tip, and generates a cross-sectional view (Probe's Eye View) perpendicular to the needle axis and a longitudinal section view parallel to the needle axis. This dynamic slicing display method centered on the needle tip enables the operator to clearly identify the organizational structures in front of the needle tip, such as blood vessels or bronchial branches.

[0045] Step S603: Calculate guiding quantization indicators and risk warnings. In addition to visual images, the system also calculates quantization parameters of the puncture path in real time to assist in decision-making. The system calculates the Euclidean distance between the physical coordinates of the needle tip and the centroid coordinates of the lesion : ; The system numerically displays this remaining distance in real time on the navigation interface . In addition, the system uses the ray casting algorithm to detect the minimum distance between the extension line of the needle tip and important anatomical structures (such as large blood vessels, heart edge). When the predicted path passes through the risk area or is less than the preset sampling threshold (such as 5 mm), the system triggers a visual warning, changing the border color of the navigation interface or displaying a highlighted aperture at the needle tip position. Through the above steps, the system converts the calibration results of complex internal mechanical deformations into intuitive image-guided information, generating and displaying dynamic navigation images.

Claims

1. A method for lung biopsy under virtual positioning CT guidance, characterized in that, Includes the following steps: Construct a baseline virtual model that includes lung anatomy and a pre-defined puncture path; Real-time acquisition of point cloud data of the patient's body surface; mapping of body surface deformation to internal anatomical structure through a biomechanical model to generate a virtual lung model that changes with respiratory motion. By using a fiber optic grating sensor array embedded in the puncture needle, the needle body shape of the puncture needle can be calculated in real time and the axial resultant force strain can be separated. Monitor the variation characteristics of the axial resultant force strain, and when a mechanical event characterizing the puncture needle penetrating the anatomical boundary layer is identified, calculate the spatial deviation vector between the physical coordinates of the needle tip and the virtual predicted coordinates; A local spatial calibration field is constructed based on the spatial deviation vector, and the virtual lung model is non-rigidly corrected to obtain a corrected deformation mapping model. The reconstructed needle shape is fused and rendered with the modified deformation mapping model to generate and display a dynamic navigation image.

2. The method for virtual positioning CT-guided lung biopsy according to claim 1, characterized in that, The process of mapping body surface deformation to internal anatomical structures using a biomechanical model to generate a virtual lung model that changes with respiratory movements specifically includes: The real-time collected point cloud data of the body surface is registered with the preset reference body surface model to calculate the displacement field of the body surface. The biomechanical model is solved using the Navier-Cauchy equation or the finite element method, and the displacement field of the body surface is used as the boundary condition to deduce the internal displacement field of the internal grid nodes of the lung. The internal displacement field is superimposed onto the baseline virtual model to generate the virtual lung model.

3. The method for virtual positioning CT-guided lung biopsy according to claim 1, characterized in that, The fiber grating sensing array comprises three fiber cores distributed at 120-degree angles along the circumference of the needle body cross-section; the real-time calculation of the needle body morphology and separation of the axial resultant force strain specifically includes: The center wavelength drift of each fiber core is collected, and temperature term decoupling is performed to obtain pure mechanical strain. The bending strain is calculated using differential mode to reconstruct the three-dimensional curve of the needle body and the physical coordinates of the needle tip. The axial resultant force strain is calculated using a common-mode method to characterize the axial resistance experienced by the needle tip.

4. The method for virtual positioning CT-guided lung biopsy according to claim 1, characterized in that, The monitoring of the axial resultant force strain variation characteristics, when identifying a mechanical event characterizing the penetration of the puncture needle through the anatomical boundary layer, specifically includes: Calculate the first derivative of the axial resultant force strain with respect to time; Determine whether the absolute value of the first derivative is greater than a preset strain rate threshold; At the same time, it is determined whether the needle insertion depth calculated by the current fiber reconstruction is within the anatomical feature boundary layer depth range estimated based on the preoperative planning path; When the above conditions are met simultaneously, a membrane permeation mechanical mutation event is determined to have occurred, and the current moment is locked as the event moment.

5. The method for virtual positioning CT-guided lung biopsy according to claim 4, characterized in that, The calculation of the spatial deviation vector between the physical coordinates of the needle tip and the virtual predicted coordinates specifically includes: The physical coordinates of the needle tip reconstructed by the fiber grating sensor array at the time of the event are obtained; In the virtual lung model, the intersection point between the axial ray of the puncture needle and the predicted anatomical feature boundary is found, and this intersection point is used as the virtual predicted coordinate. The vector obtained by subtracting the virtual predicted coordinates from the physical coordinates of the needle tip is calculated and used as the spatial deviation vector.

6. The method for virtual positioning CT-guided lung biopsy according to claim 5, characterized in that, The step of constructing a local spatial calibration field based on the spatial deviation vector and performing non-rigid correction on the virtual lung model specifically includes: A local spatial calibration field based on Gaussian radial basis functions is constructed with the physical coordinates of the needle tip as the center. For any voxel point in the virtual lung model, calculate its Euclidean distance to the physical coordinates of the needle tip; The attenuation weight of the calibration displacement is calculated based on the Euclidean distance, and the calibration displacement decreases exponentially with the increase of distance. The product of the spatial deviation vector and the attenuation weight is superimposed on the original predicted coordinates of the voxel point to achieve local non-rigid stretching correction of the anatomical structure.

7. The method for virtual positioning CT-guided lung biopsy according to claim 1, characterized in that, The generation and display of the dynamic navigation image specifically includes: The modified deformation mapping model is rendered using volume rendering technology, and the puncture needle model is drawn using cylindrical primitives. Construct a composite navigation interface that includes a 3D panoramic view and a multi-plane reconstructed view; The multi-plane reconstructed view includes a cross-sectional view perpendicular to the needle axis and a longitudinal section view parallel to the needle axis, generated based on the current tangential and normal vectors of the needle tip.

8. The method for virtual positioning CT-guided lung biopsy according to claim 1, characterized in that, The method further includes: The Euclidean distance between the physical coordinates of the needle tip and the centroid coordinates of the lesion is calculated in real time and the remaining distance is displayed on the navigation interface; The minimum distance between the extended line of the needle tip and important anatomical structures is detected using a ray projection algorithm. When the minimum distance is less than a preset safety threshold, a visual warning is triggered.

9. A method for virtual positioning CT-guided lung biopsy according to claim 4, characterized in that, The anatomical feature boundary layer is the parietal pleura; the mechanical event that characterized the puncture needle penetrating the anatomical feature boundary layer was identified when the needle tip broke through the parietal pleura and entered the lung parenchyma, and the axial resistance decreased abruptly due to the difference in the elastic modulus of the tissues, which in turn caused a negative change in the axial resultant strain.

10. A system applied to the virtual positioning CT-guided lung biopsy method of claim 1, characterized in that, include: CT scanning equipment is used to acquire preoperative images of patients to build a baseline virtual model; A body surface optical monitoring device is used to acquire point cloud data of the patient's body surface in real time; The puncture needle has an embedded fiber optic grating sensor array. A fiber optic grating demodulator, connected to the puncture needle, is used to collect optical signals and demodulate them into wavelength data; A data processing terminal configured to perform the steps of the method as described in any one of claims 1 to 9; The display terminal is used to display the generated dynamic navigation image.

Images