Cardiovascular disease treatment puncture system

Abstract

This invention relates to the field of minimally invasive cardiovascular interventional therapy, specifically to a cardiovascular disease treatment puncture system. It aims to address the problems of traditional punctures, such as unfused vascular wall mechanical properties, respiratory motion deformation, and real-time force feedback, leading to target location deviations, easy vascular penetration, and complications. This invention integrates intravascular ultrasound, X-ray fluoroscopy, and respiratory motion signals through a multimodal image fusion unit to generate a vascular wall elastic modulus distribution map and dynamic motion trajectory. A puncture path planning unit screens low-elasticity modulus safe target points, combines deformation prediction to correct the path, and avoids bifurcation and calcified areas. An adaptive puncture execution unit relies on a needle tip pressure sensor to adjust the needle insertion angle and depth in a closed loop; if the force threshold is exceeded, the procedure pauses and replans the local path. A complication risk monitoring unit identifies vascular penetration characteristics through Doppler ultrasound, triggers biomaterial occlusion, and issues an alarm, thereby improving the accuracy and safety of minimally invasive cardiovascular punctures.

Description

Technical Field

[0001] This invention relates to the field of minimally invasive interventional treatment technology for cardiovascular diseases, and more specifically, to a puncture system for treating cardiovascular diseases. Background Technology

[0002] Cardiovascular minimally invasive interventional therapy is an important technology, specifically applied to the minimally invasive puncture procedure in the treatment of cardiovascular diseases. Its core principle is to improve the accuracy and safety of the puncture procedure through dynamic sensing and adaptive control, meeting the core clinical needs of cardiovascular minimally invasive interventional therapy. During cardiovascular puncture, the target blood vessel undergoes continuous dynamic displacement and deformation with the body's breathing movements. Simultaneously, the tissue biomechanical properties of different regions of the blood vessel wall exhibit significant individual and regional differences. Because clinical puncture procedures struggle to deeply integrate the biomechanical properties of vascular tissue, the dynamic movement patterns of the blood vessel, and the real-time contact force feedback during the puncture process to achieve dynamic optimization and precise execution of the puncture path, continuous deviations occur between the puncture target location and the actual spatial position of the blood vessel and the biomechanical safety zone. This increases the risk of accidental vascular wall penetration, tissue damage, and related serious complications, making it difficult to meet the clinical requirements of precision, safety, and controllability in cardiovascular minimally invasive interventional therapy. To address this technical problem, we have developed a cardiovascular disease treatment puncture system. Summary of the Invention

[0003] The purpose of this invention is to provide a puncture system for treating cardiovascular diseases, in order to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, a cardiovascular disease treatment puncture system is provided, comprising: The multimodal image fusion unit is used to integrate intravascular ultrasound imaging, X-ray fluoroscopy images and respiratory motion signals in real time, and generates a blood vessel wall elastic modulus distribution map and a blood vessel dynamic motion field through intelligent algorithms. The blood vessel wall elastic modulus distribution map is used to quantify the deformation characteristics of blood vessel tissue, and the blood vessel dynamic motion field includes a blood vessel wall surface grid model and its complete motion information at each moment during the respiratory cycle. The puncture path planning unit is used to select low elastic modulus regions as puncture targets in the target blood vessel area based on the elastic modulus distribution map of the blood vessel wall, predict the deformation trend of the blood vessel during puncture based on the dynamic motion field of the blood vessel, calculate the blood vessel displacement caused by the deformation trend through real-time simulation, and make reverse corrections to the planned navigation path based on the blood vessel displacement. At the same time, it combines the anatomical structure model of the blood vessel to avoid blood vessel bifurcation and calcified areas, and generate a real-time navigation path leading to the puncture target. The adaptive puncture execution unit is used to perform puncture operations according to the real-time navigation path, and includes a flexible puncture needle, a drive mechanism for driving the flexible puncture needle, and a pressure sensor disposed at the tip of the flexible puncture needle. The pressure sensor is used to monitor the contact force between the flexible puncture needle and the blood vessel wall in real time. The drive mechanism adjusts the insertion angle and depth of the flexible puncture needle according to the deviation signal of the real-time navigation path and the feedback of the contact force. When the contact force exceeds the safety threshold, the needle insertion is paused and the local path is replanned according to the changing trend of the contact force. The complication risk monitoring unit is used to monitor the blood flow spectrum in the puncture target area using Doppler ultrasound, and to determine in real time whether the blood vessel has been accidentally penetrated based on the abnormal characteristics of the blood flow spectrum. When the characteristic signal of blood vessel penetration is detected, the coaxial microcatheter occlusion mechanism is triggered to perform local occlusion and an alarm signal is issued simultaneously.

[0005] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention integrates intravascular ultrasound, X-ray fluoroscopy, and respiratory motion signals through a multimodal image fusion unit to generate a distribution map of the elastic modulus of the vascular wall and a dynamic motion trajectory. This allows for precise screening of safe puncture targets with low elastic modulus, reducing the risk of vascular injury from the source. The puncture path planning unit performs displacement reversal correction based on vascular deformation prediction and finite element simulation. Combined with a vascular anatomy model, it automatically avoids high-risk areas such as bifurcation and calcification, generating a real-time navigation path that dynamically adapts to respiratory motion. This effectively eliminates positioning deviations caused by dynamic vascular displacement. The adaptive puncture execution unit relies on a needle tip pressure sensor to achieve real-time feedback of contact force, adjusting the needle insertion angle and depth in a closed loop. When the contact force exceeds a threshold, it automatically pauses and replans the local path, significantly improving the accuracy and stability of puncture manipulation. The complication risk monitoring unit monitors the blood flow spectrum in real time using Doppler ultrasound, quickly identifying vascular penetration characteristics and simultaneously triggering biomaterial occlusion and alarms to promptly curb serious complications such as bleeding. Overall, this invention achieves precise navigation, adaptive execution, and proactive safety protection throughout the entire cardiovascular puncture process, improving the operational precision, success rate, and clinical safety of minimally invasive interventional treatment. Attached Figure Description

[0006] Figure 1 This is an overall block diagram of the present invention.

[0007] The meanings of the labels in the diagram are as follows: 1. Multimodal image fusion unit; 2. Puncture path planning unit; 3. Adaptive puncture execution unit; 4. Complication risk monitoring unit. Detailed Implementation

[0008] The technical solutions of 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.

[0009] This invention provides a puncture system for treating cardiovascular diseases. Please refer to [link / reference]. Figure 1 As shown, it includes: The multimodal image fusion unit 1 is used to integrate intravascular ultrasound imaging, X-ray fluoroscopy images and respiratory motion signals in real time, and generate a blood vessel wall elastic modulus distribution map and a blood vessel dynamic motion field through intelligent algorithms. The blood vessel wall elastic modulus distribution map is used to quantify the deformation characteristics of blood vessel tissue, and the blood vessel dynamic motion field includes a blood vessel wall surface grid model and its complete motion information at each moment during the respiratory cycle. The puncture path planning unit 2 is used to screen out low elastic modulus areas as puncture targets in the target blood vessel area based on the elastic modulus distribution map of the blood vessel wall, predict the deformation trend of the blood vessel during puncture based on the dynamic motion field of the blood vessel, calculate the blood vessel displacement caused by the deformation trend through real-time simulation, and make reverse corrections to the planned navigation path based on the blood vessel displacement. At the same time, it combines the anatomical structure model of the blood vessel to avoid blood vessel bifurcation and calcification areas, and generate a real-time navigation path leading to the puncture target. The adaptive puncture execution unit 3 is used to perform puncture operations according to the real-time navigation path, including a flexible puncture needle, a drive mechanism for driving the flexible puncture needle, and a pressure sensor set at the tip of the flexible puncture needle. The pressure sensor is used to monitor the contact force between the flexible puncture needle and the blood vessel wall in real time. The drive mechanism adjusts the insertion angle and depth of the flexible puncture needle according to the deviation signal of the real-time navigation path and the feedback of the contact force. When the contact force exceeds the safety threshold, the needle insertion is paused and the local path is replanned according to the changing trend of the contact force. Complication risk monitoring unit 4 is used to monitor the blood flow spectrum of the puncture target area by Doppler ultrasound, and to determine in real time whether the blood vessel has been accidentally penetrated based on the abnormal characteristics of the blood flow spectrum. When the blood vessel penetration characteristic signal is detected, the coaxial microcatheter occlusion mechanism is triggered to perform local occlusion and an alarm signal is issued simultaneously.

[0010] When generating the elastic modulus distribution map of the blood vessel wall and the dynamic motion field of the blood vessel, the intelligent algorithm uses a deep learning model to simultaneously process the reflected wave signal of intravascular ultrasound imaging, the spatial coordinate data of blood vessels from X-ray fluoroscopy images, and the temporal waveform of respiratory motion signals. The deep learning model extracts the tissue stiffness features in the reflected wave signal through a convolutional neural network to generate an elastic modulus distribution map that quantifies the deformation capacity of the blood vessel wall, and aligns the temporal waveform with the spatial coordinate data of the blood vessel through a recurrent neural network to generate the dynamic motion field of the blood vessel.

[0011] When selecting regions with low elastic modulus as puncture targets, the elastic modulus distribution map of the blood vessel wall is input into the region segmentation algorithm. The region segmentation algorithm calculates the elastic modulus value of each pixel unit in the elastic modulus distribution map of the blood vessel wall, marks the regions that are continuously adjacent and whose elastic modulus values ​​are lower than the average value of the elastic modulus distribution map of the blood vessel wall, and outputs the regions with low elastic modulus as puncture targets.

[0012] When generating a real-time navigation path to the puncture target, the system employs a dynamic path planning module. Through a multi-constraint optimization algorithm, it uses vascular deformation displacement, bifurcation and calcification risk areas, and the distribution of the vascular wall elastic modulus as parallel constraints to generate a real-time navigation path that both compensates for dynamic vascular deformation and completely avoids high-risk areas. Specifically, this multi-constraint optimization algorithm uses the initial navigation path coordinate sequence as the optimization variable, with the objective function being the shortest total path length and the minimum curvature change, while simultaneously applying three types of core constraints: The first type is deformation compensation constraint. Based on the radial and tangential displacement dataset of blood vessels, the offset of each coordinate point on the planned path is required to be equal to the equivalent displacement vector superimposed in the opposite direction, ensuring that the path adapts to the dynamic deformation of blood vessels in real time. The second type is obstacle avoidance constraint. Based on the bifurcation point coordinates and calcification zone boundary data pre-stored in the vascular anatomy model, the planned path is required to maintain a safe distance (≥0.5mm) from risk areas throughout the entire process. The third type is mechanical safety constraint. Based on the elastic modulus distribution map of the blood vessel wall, the planned path is required to prioritize passing through low elastic modulus safe areas. The algorithm uses Sequential Quadratic Programming (SQP) to iteratively solve this multi-constraint optimization problem. Under the premise of satisfying all constraints, a globally optimal real-time navigation path coordinate sequence is generated, ensuring that the puncture path can not only dynamically compensate for the deformation displacement caused by vascular respiratory motion, but also completely avoid high-risk areas such as vascular bifurcation and calcification, providing a safe and accurate navigation basis for the adaptive puncture execution unit.

[0013] When making reverse corrections to the planned navigation path based on the vessel displacement, the radial and tangential displacement data of the vessel are input into the path optimization algorithm. The path optimization algorithm analyzes the direction and magnitude of the displacement vectors in the radial and tangential displacement data of the vessel, and superimposes the equivalent displacement vectors in reverse into the initial navigation path coordinates to generate a corrected path coordinate sequence to compensate for deformation.

[0014] When generating a real-time navigation path to the puncture target, the corrected path coordinate sequence is input into the three-dimensional obstacle avoidance algorithm. The three-dimensional obstacle avoidance algorithm calls the pre-stored bifurcation point coordinates and calcification zone boundary data in the vascular anatomy model, detects the overlapping segments of the corrected path and the risk area, dynamically replans the path direction, and outputs a real-time navigation path that bypasses the obstacle.

[0015] When adjusting the insertion angle and depth of the flexible puncture needle, the deviation value between the needle tip position coordinates and the coordinate sequence of the real-time navigation path is compared in real time. Simultaneously, the contact force data monitored by the pressure sensor is received. The drive mechanism generates a micro-angle deflection torque proportional to the path deviation value through the piezoelectric ceramic array, and executes the depth feed amount associated with the contact force data through the stepper motor.

[0016] When the contact force data continuously exceeds the safety threshold, the temporal gradient change direction of the contact force is extracted. The drive mechanism calculates the needle tip obstacle avoidance vector based on the temporal gradient change direction, generates a local path segment to avoid the current pressure point, and drives the stepper motor to perform displacement along the local path segment before resuming needle insertion.

[0017] The frequency domain signal of the blood flow spectrum around the needle tip is acquired by a Doppler ultrasound probe. The turbulent energy peak and characteristic frequency components in the frequency domain signal are extracted by fast Fourier transform. When the growth rate of the energy peak exceeds the preset threshold within a preset time period and the characteristic frequency matches the blood vessel penetration mode, the biomaterial occlusion device is triggered to release the occlusion colloid.

[0018] Further explanation is needed. After the multimodal image fusion unit completes the real-time integration of intravascular ultrasound imaging, X-ray fluoroscopy images, and respiratory motion signals, the system generates a vascular wall elastic modulus distribution map and a vascular dynamic motion field through a multimodal fusion deep learning model. This deep learning model is a deep learning model that integrates convolutional neural networks (CNN) and recurrent neural networks (LSTM), specifically designed for processing heterogeneous images and temporal signals. It can simultaneously achieve spatial feature extraction and temporal feature alignment. This model performs synchronous parallel processing on three types of input data: the reflected wave signal from intravascular ultrasound imaging is the time-domain echo signal of ultrasound from the vascular wall tissue, used to reflect differences in tissue stiffness; the vascular spatial coordinate data from X-ray fluoroscopy images is the sequence of point coordinates of the blood vessel in three-dimensional space, used to characterize the anatomical spatial position of the blood vessel; and the time-domain waveform of the respiratory motion signal is the time-domain curve of vascular displacement changing with the respiratory cycle, used to characterize the dynamic disturbance of the blood vessel caused by respiration. During processing, the intravascular ultrasound reflected wave signal is first converted into a two-dimensional time-frequency feature map through a short-time Fourier transform. The transformation formula is... ,in The reflected wave is a time-domain signal. It is a Gaussian window function. Position of the time window The angular frequency is then used as the time-frequency feature map. This map is then input into the convolutional neural network branch of the model. Through multiple convolutional and pooling operations, tissue stiffness features are extracted. The convolution operation formula is... ,in This is a time-frequency feature map. It is a 3×3 convolution kernel. As a bias term, after the convolutional layer extracts hardness-related features such as ultrasonic reflection intensity and attenuation coefficient, these features are mapped to elastic modulus values ​​through a fully connected layer. The quantization relationship between elastic modulus and ultrasonic features is as follows: ,in This refers to the elastic modulus of the blood vessel wall. These are characteristic values ​​of ultrasonic reflection. , , For in vitro experiments, calibration coefficients are used, for example, when the characteristic value of ultrasonic reflection is... When, the corresponding elastic modulus ,when hour, The model reconstructs the elastic modulus of each pixel, ultimately generating a distribution map of the elastic modulus of the blood vessel wall that quantifies its deformation capacity. Simultaneously, the model's fusion architecture of a 3D convolutional neural network (3D-CNN) and a recurrent neural network (LSTM) processes the 3D surface mesh of the blood vessel wall reconstructed from intravascular ultrasound imaging, the spatial coordinate data of the blood vessel from X-ray fluoroscopy images, and the temporal waveform of respiratory motion signals. Specifically, firstly, based on the reflected wave signal from intravascular ultrasound imaging, a 3D surface mesh model of the blood vessel wall is generated using a 3D reconstruction algorithm. This mesh model contains complete geometric information of the adventitia and intima of the blood vessel wall, with a mesh node density of 0.1mm × 0.1mm, which can finely characterize the local morphological features of the blood vessel wall. Then, the node coordinate data of the 3D surface mesh model are processed... As the initial vascular morphology benchmark, the time-domain waveform of the respiratory motion signal is received synchronously. Spatial coordinates of blood vessels in X-ray fluoroscopy images The local deformation features of the blood vessel wall surface mesh are extracted using 3D-CNN, and then the temporal memory unit of LSTM is used to complete the time-by-time alignment of the temporal waveform, spatial coordinates and mesh deformation. The temporal update formula is as follows: ,in Hide the current state. This is the hidden state from the previous moment. Given the current coordinates of the mesh nodes on the blood vessel wall surface, and by aligning the respiratory phase, the spatial position of the blood vessel, and the mesh deformation, the nodal displacement vector field of the blood vessel wall surface mesh model at each time step is output. Ultimately, a dynamic motion field of blood vessels is generated, which includes a mesh model of the blood vessel wall surface and complete motion information at each moment during the respiratory cycle. This enables the synchronous output of the spatial distribution of tissue stiffness and the dynamic motion law of local areas of the blood vessel wall, providing complete local motion information for accurate deformation prediction of the subsequent puncture target area.

[0019] After generating a vascular wall elastic modulus distribution map that accurately quantifies the deformation characteristics of vascular wall tissue using a multimodal fusion deep learning model, the puncture path planning unit immediately initiates an intelligent screening process for puncture targets. This process relies on a region segmentation algorithm to locate and extract safe regions with low elastic modulus. Region segmentation is a computer vision processing algorithm that divides continuous regions with uniform attributes from a two-dimensional physical distribution image based on the numerical features of image pixels. This system employs an improved threshold connected component segmentation algorithm adapted to vascular images, balancing real-time performance and positioning accuracy, fully meeting the clinical real-time requirements of cardiovascular puncture. In the specific execution process, the algorithm first traverses every pixel unit in the vascular wall elastic modulus distribution map, calculates the average elastic modulus of the entire image, and uses this average value as the judgment threshold. Pixels with an elastic modulus below the average value are marked as candidate low elastic modulus pixels. The continuity of candidate pixels is verified by the eight-connected neighborhood analysis rule in image processing. Interconnected candidate pixels are classified into continuous regions, while isolated pixels are removed to ensure that the selected regions have sufficient space. Finally, these continuous regions with an elastic modulus below the average value are determined as the optimal puncture target points. This ensures that the selected regions have enough space to meet the puncture operation. The blood vessel wall tissue in these regions is softer and more deformable, which can effectively reduce the risk of blood vessel rupture and tissue damage during puncture, laying a solid foundation for the subsequent safe and accurate puncture path planning.

[0020] After calculating the average value of the elastic modulus distribution map of the entire blood vessel wall and marking all candidate low elastic modulus pixels with values ​​lower than the average, the algorithm will conduct a strict continuity check on all candidate pixels using the eight-connected neighborhood analysis rule. This is to filter out the continuous region that truly meets the puncture requirements. The eight-connected neighborhood analysis rule is a standardized rule for determining the spatial continuity of pixels in digital image processing. Specifically, it refers to taking any marked candidate low elastic modulus pixel as the central reference pixel, and defining all adjacent pixels in the four orthogonal adjacent directions directly above, below, to the left, and to the right of this pixel, as well as the four diagonal adjacent directions in the four diagonal adjacent directions in the four diagonal adjacent directions in the eight directions above, below, to the left, and to the right of this pixel as connected neighbors of this reference pixel. The rule requires that only when the central reference pixel has at least one other pixel marked as a candidate low elastic modulus in the above eight adjacent directions is the central reference pixel considered a valid pixel belonging to the continuous region. The algorithm will apply this rule to all candidate pixels. The algorithm performs a row-by-row, column-by-column traversal scan, grouping candidate pixels that satisfy the eight-connected relation into a single continuous region and assigning it a unique region number, thus gradually constructing multiple complete sets of continuous low elastic modulus regions. Isolated low elastic modulus pixels, on the other hand, are those candidate pixels where no other pixels of the same type exist in any of their eight connected neighborhoods, and all eight adjacent pixels are ordinary pixels with elastic modulus values ​​higher than the average of the distribution map. These isolated pixels, existing alone without any adjacent pixels of the same type, correspond to only a very small point-like area in the actual blood vessel wall tissue. They lack the spatial dimensions required for puncture operations and cannot guarantee the stability and safety of the puncture. Therefore, the algorithm automatically removes these isolated pixels from the candidate target area, retaining only low elastic modulus regions that pass the eight-connected neighborhood analysis rule, are continuous, and meet the area requirements, ensuring that the final output puncture target area meets the spatial and safety requirements for clinical puncture.

[0021] After accurately identifying safe puncture targets with low elastic modulus, the puncture path planning unit performs real-time prediction of vascular deformation trends during puncture based on the vascular dynamic motion field generated by the multimodal image fusion unit. This adapts to the dynamic displacement of the cardiovascular system caused by respiratory movements, ensuring the dynamic adaptability of the puncture path. In practice, the vascular dynamic motion field is first input into a lightweight real-time deformation prediction model. This model first extracts complete motion information of the local area of ​​the blood vessel wall where the puncture target is located from the vascular dynamic motion field, specifically including: The set of corresponding mesh nodes of the target region in the mesh model of the blood vessel wall surface, and the displacement vectors of these nodes at each time point during the respiratory cycle. The model performs time-series analysis on the motion information of the grid nodes in the target area, extracting the vascular displacement rate and directional change characteristics of the target area. The directional change characteristics refer to the instantaneous orientation of the displacement vector of the puncture target area on the blood vessel wall under respiratory motion, the angular deflection value in three-dimensional space, and the curvature change of the motion trajectory. Specifically, it covers the switching of the target point's motion direction, the magnitude of the deflection angle, and the degree of trajectory curvature in the radial, tangential, and axial dimensions of the blood vessel. These are core temporal features for accurately depicting the dynamic laws of blood vessel swaying and contracting with respiration. Simultaneously, the model also retrieves the elastic modulus distribution of the blood vessel wall. The figure shows the stiffness parameter of the blood vessel wall material corresponding to the puncture target area. This parameter is a physical quantitative index characterizing the ability of the blood vessel wall tissue to resist external deformation. It is directly converted from the average elastic modulus value of the target area. The stiffness parameter is positively correlated with the elastic modulus. The larger the elastic modulus, the higher the stiffness of the blood vessel wall and the stronger its resistance to deformation. Conversely, the smaller the elastic modulus, the softer the blood vessel wall and the more easily it deforms. It is the core basic parameter for calculating vascular biomechanical deformation. After completing feature extraction and parameter retrieval, the model combines the temporal phase of the complete respiratory cycle and divides the respiratory cycle into three key stages: inspiration, expiration, and rest. The vascular displacement rate is first calculated using the temporal difference method. ,in Let t be the difference in displacement of the vascular target point between time t and the previous time. The image sampling time interval is used as the basis for determining the deflection angle of the directional change characteristics through vector dot product operations. , Let be the displacement velocity vector at time t, and then the stiffness parameters of the blood vessel wall material are... ,according to Conversion, The average elastic modulus of the target region. The stiffness conversion coefficient specific to vascular tissue is substituted into the biomechanical deformation response formula to calculate the deformation displacement of the blood vessel under the combined action of the puncture force and respiratory motion during the respiratory cycle. ( To estimate the contact force at the puncture needle tip, This is the effective stress-bearing length of the blood vessel wall. The area of ​​the needle tip subjected to force. The deformation amplitude, direction, and rate of the vascular target at each time point are obtained by iteratively calculating the deformation amount of the vascular autonomous displacement induced by respiratory movement, by substituting each time point into the data. Finally, these time-by-time deformation calculation results are integrated into a continuous time series, and the final output is deformation trend time series prediction data covering the entire respiratory cycle and including deformation amplitude, direction, and rate. This provides accurate dynamic deformation basis for subsequent real-time simulation calculation of vascular displacement and reverse correction of puncture path.

[0022] After obtaining the deformation trend time-series prediction data for the complete respiratory cycle, the puncture path planning unit relies on the finite element simulation engine (a real-time mechanical simulation calculation core developed based on continuum mechanics, numerical discretization methods, and the mechanical properties of biological soft tissues, capable of discretizing continuous vascular wall soft tissue into quantifiable and calculable mesh elements, accurately solving the stress distribution and displacement changes of the vascular wall under external force, meeting the real-time and high-precision requirements of cardiovascular puncture) to perform simulation calculations of vascular displacement caused by deformation trends. First, the deformation trend time-series prediction data is input into the finite element simulation engine, and the engine then initiates the process of constructing a meshed biomechanical model of the vascular wall. Based on the three-dimensional anatomical structure of the target vascular obtained by multimodal image fusion, the puncture target is first... The blood vessel wall region where the point is located is abstracted as a thin layer of isotropic viscoelastic soft tissue that conforms to human physiological characteristics. The geometric modeling domain is strictly defined according to the actual anatomical thickness of the blood vessel wall (0.2-0.5mm). An adaptive hybrid mesh discretization method is adopted. First, triangular face meshes with an accuracy of 0.1mm×0.1mm are generated on the inner and outer surfaces of the blood vessel wall to ensure the complete restoration of the curved contour of the blood vessel wall. Then, the face meshes are stretched in the direction of blood vessel wall thickness to generate hexahedral volume meshes that correspond one-to-one with the face meshes. The node spacing of the volume mesh is adaptively matched according to the blood vessel diameter. At the same time, a unique biomechanical property is precisely assigned to each mesh unit. The elastic modulus E and the Poisson's ratio of the fixed soft tissue are extracted from the elastic modulus distribution map of the blood vessel wall at the corresponding position of the mesh. Blood vessel wall tissue density Then, the two ends of the blood vessel are set as fixed constraints, and the puncture target area is set as a free force surface, which perfectly matches the actual mechanical constraint state of the blood vessel in the body. At this point, the construction of the meshed biomechanical model, including geometric discretization, material parameter assignment, and boundary condition definition, is completed. After the model is built, the simulation engine starts loading the deformation trend time series prediction data and simulating the stress distribution under puncture pressure, with a time step of 10ms. The time-series prediction data will include respiratory motion deformation excitation and the estimated contact force of the standard needle tip in clinical minimally invasive puncture. The data is applied to the core mesh nodes of the puncture target in the gridded biomechanical model at each time step, based on the core mechanical equilibrium equations of the finite element method. Conduct mechanical calculations, among which The overall stiffness matrix of the model is derived from the stiffness matrices of all elements. Assembled, element stiffness matrix , This is a strain-displacement matrix used to correlate mesh element strain with nodal displacement. The elasticity matrix is ​​calculated from the elastic modulus and Poisson's ratio, and the specific formula is as follows: , This is the external load vector at the current moment, which integrates respiratory deformation load and puncture pressure load. Let be the displacement vector of the mesh nodes to be solved; then, the Newton-Raphson iterative method is used to solve the displacement vector iteratively, first initializing the displacement vector. Calculate the residual vector When the residual vector magnitude is The iteration is considered to have converged when a preset convergence threshold is reached. If the threshold is not met, the displacement vector is continuously updated and the mechanical calculation is repeated until all mesh node displacements converge. During the iteration process, the stress calculation formula is used simultaneously. The stress value of each grid cell is calculated to generate a stress distribution cloud map of the blood vessel wall under the combined effects of puncture pressure and respiratory motion. After the iteration fully converges, the simulation engine uses the blood vessel centerline as the axial reference and decomposes the three-dimensional displacement vector of each grid node into radial displacement (the displacement component perpendicular to the blood vessel centerline, representing the displacement of expansion or contraction of the blood vessel wall) and tangential displacement (the displacement component along the tangential direction of the blood vessel wall, representing the displacement of swinging or slipping of the blood vessel wall). The specific decomposition formula is as follows: Radial displacement... Tangential displacement ,in The radial normal vector of the blood vessel. The unit tangential vector of the blood vessel is used. Finally, the radial and tangential displacement values ​​of the core grid nodes of the puncture target area are recorded sequentially according to the time step. The displacement data of all moments in the complete respiratory cycle are sorted and integrated by timestamp. Finally, a standardized dataset of radial and tangential displacement of the blood vessel containing time information, radial displacement of the blood vessel, and tangential displacement of the blood vessel is output, which provides a precise displacement compensation basis for the reverse correction of the subsequent puncture navigation path.

[0023] After iteratively solving the mechanical deformation of the blood vessel wall using a finite element simulation engine and outputting a dataset of radial and tangential displacements of the blood vessel covering the entire respiratory cycle, the puncture path planning unit will use this displacement dataset to perform precise dynamic reverse correction of the initial navigation path. This completely offsets the positional offset caused by vascular respiratory motion and puncture deformation, ensuring that the puncture needle tip is always accurately aligned with the target point. Based on the principles of spatial coordinate transformation, vector superposition, and temporal compensation, it can dynamically match and compensate the static navigation path according to the real-time deformation and displacement of the blood vessel, ensuring that the puncture path and the dynamic position of the blood vessel are perfectly matched. In specific implementation, the radial and tangential displacement data of the blood vessel are first processed... The algorithm optimizes the input path. It first performs frame-by-frame detailed analysis of the displacement data corresponding to each time stamp in the dataset. The direction of the displacement vector refers to the spatial orientation of the deformation displacement of the target area of ​​the blood vessel wall during the respiratory cycle. It is divided into a radial direction perpendicular to the blood vessel axis and a tangential direction along the vessel wall tangent, with positive and negative directions indicated. A positive direction represents radial displacement away from the blood vessel center and tangential displacement along the vessel wall, while a negative direction represents radial displacement towards the blood vessel center and tangential displacement along the vessel wall. The magnitude of the displacement vector refers to the actual physical distance of the deformation displacement of the target area, which is the magnitude of the displacement vector and a core value characterizing the magnitude of the deformation. During the analysis process, the algorithm first extracts arbitrary time points from the dataset. Corresponding radial displacement With tangential displacement The formula for calculating the vector magnitude. The magnitude of the displacement vector at that moment is accurately calculated, and then the direction angle is calculated using the formula for planar polar coordinates. The algorithm solves for the orientation angle of the displacement vector within the vessel wall plane, fully determining the spatial orientation and deformation amplitude of the displacement vector. After analyzing the direction and amplitude of the displacement vector at all time points, the algorithm retrieves the three-dimensional coordinate sequence of the initial navigation path planned based on the static vascular anatomy. The initial navigation path coordinates are a sequence of static puncture path coordinate points that do not consider dynamic vascular deformation. Each coordinate point corresponds to a unique timestamp and puncture depth. Subsequently, the algorithm performs a crucial correction step of inverse superposition of equivalent displacement vectors. The equivalent displacement vector is a compensation displacement vector that is exactly equal in magnitude and opposite in direction to the vascular deformation displacement. Its function is to accurately compensate for the path deviation caused by vascular deformation. Its radial equivalent displacement component is... The tangential equivalent displacement components are The algorithm converts the equivalent displacement vector components at each time t into coordinate offsets in a three-dimensional spatial coordinate system based on the orthogonal correspondence of the axial, radial, and tangential directions of the vascular three-dimensional anatomical coordinate system. Then, by correcting the coordinate calculation formula The algorithm precisely superimposes the coordinate offset onto the coordinate points corresponding to the initial navigation path at each time point. The algorithm performs the above reverse superposition calculation on all coordinate points of the initial navigation path one by one according to the timestamp order. At the same time, it removes path inflection points and coordinate change points caused by deformation compensation. The corrected coordinate points are smoothed by cubic spline interpolation to ensure the continuity and smoothness of the corrected path and the feasibility of puncture. Finally, a corrected path coordinate sequence covering the entire respiratory cycle and capable of real-time compensation for vascular dynamic deformation is generated, providing an accurate corrected coordinate basis for the subsequent generation of real-time navigation paths to avoid risk areas.

[0024] After generating the corrected path coordinate sequence for vascular dynamic deformation compensation, the puncture path planning unit inputs this corrected path coordinate sequence into the three-dimensional obstacle avoidance algorithm, formally entering the final real-time navigation path generation stage. This ensures that the puncture path adapts to vascular dynamic deformation while safely avoiding high-risk areas in the cardiovascular anatomy. The three-dimensional obstacle avoidance algorithm is a real-time path optimization algorithm based on three-dimensional spatial collision detection, path smoothing replanning, and vascular anatomical constraints. It is specifically designed for complex cardiovascular anatomy scenarios, has millisecond-level response speed, and can quickly identify path risks and make dynamic adjustments, fully meeting the clinical needs of real-time intraoperative navigation. After the algorithm starts, it first calls the pre-loaded reconstructed vascular anatomy model within the system. This model is based on... The target vessel's three-dimensional digital model, reconstructed from preoperative multimodal image fusion, contains pre-stored precisely calibrated bifurcation point coordinates and calcified area boundary data. The bifurcation point coordinates are a set of three-dimensional spatial coordinates where the main vessel trunk and branches intersect; this area has a thin vessel wall structure and complex orientation, making it a high-risk area for puncture. The calcified area boundary data is a set of three-dimensional contour coordinates of calcified plaques in the vessel wall; calcified areas have hard, highly brittle tissue, making puncture extremely likely to cause vessel rupture, thus constraining it as a core avoidance area. The algorithm converts these two types of risk area data into a three-dimensional spatially restricted bounding mesh, constructing a spatial safety constraint boundary for the puncture path. Subsequently, the algorithm decomposes the corrected path coordinate sequence into continuous three-dimensional straight line segments arranged according to the puncture needle insertion sequence. Two adjacent coordinate points constitute an independent path vector. The axis-aligned bounding box (AABB) fast collision detection method is used to perform segment-by-segment spatial overlap verification between each path vector and the forbidden bounding grid. By comparing the extreme values ​​of the x, y, and z coordinates of the path segment with the corresponding extreme values ​​of the forbidden grid, it is accurately determined whether the path intersects with the risk area. If any dimension of the path segment falls within the coordinate range of the forbidden grid, the path segment is determined to be an overlapping segment, indicating a puncture risk, requiring path replanning. For detected overlapping segments, the algorithm uses the start and end coordinates of the overlapping segment as fixed constraint anchor points, combined with the distribution range of the low elastic modulus safe area in the blood vessel wall elastic modulus distribution map, and employs a fast random tree (RRT) optimization method. The algorithm generates multiple collision-free candidate path segments within a safe space. It comprehensively calculates three core indicators: path length, curvature change, and distance to the puncture target. The optimal path segment, with the shortest path, smallest curvature, and complete avoidance of risk areas, is selected and replaced with the original overlapping segments. Simultaneously, B-spline curve interpolation is used to smoothly connect the replanned path segments with the original corrected path, eliminating path inflection points and abrupt angle changes to ensure the smooth movement of the flexible puncture needle. After all overlapping segments have been replanned and the entire path has been smoothed, the algorithm sorts the final three-dimensional coordinate sequence according to the puncture timestamp and needle insertion depth, generating a real-time navigation path that can compensate for dynamic vascular deformation and completely avoid vascular bifurcation and calcification risk areas.This provides the final dynamic navigation basis for the precise needle insertion operation of the adaptive puncture execution unit.

[0025] After generating a real-time navigation path that adapts to dynamic vascular deformation and completely avoids high-risk areas such as vascular bifurcation and calcification, the adaptive puncture execution unit will strictly follow this path to perform precise puncture operations. Through a dual-control collaborative mode of position closed loop and force feedback closed loop, it will adjust the insertion angle and depth of the flexible puncture needle in real time to ensure that the puncture follows the navigation path throughout the entire process and maximizes the safety of vascular wall puncture. The specific implementation process is as follows: First, the system uses intraoperative X-ray fluoroscopy images and a positioning marker built into the needle tip to acquire the current three-dimensional spatial coordinates of the flexible puncture needle tip in real time at a sampling frequency of 1kHz. Simultaneously, it retrieves the target coordinates corresponding to the current needle insertion depth in the real-time navigation path and compares the deviation between the actual needle tip position and the target coordinates on the navigation path at each time step. This deviation is divided into two categories: positional deviation and angular deviation. The positional deviation is the Euclidean distance between the actual three-dimensional coordinates of the needle tip and the target coordinates on the navigation path, calculated using the following formula: , The actual coordinates of the needle tip. The target coordinates correspond to the navigation path. The angle deviation is the spatial angle between the actual needle insertion direction vector and the tangent direction vector of the corresponding segment of the navigation path. The calculation formula is: , This is the vector representing the actual needle insertion direction. As the target direction vector for the navigation path, while performing real-time deviation calculation, the system simultaneously receives contact force data collected by a miniature pressure sensor located at the tip of the flexible puncture needle. This pressure sensor is a three-dimensional force-sensitive sensor that can simultaneously collect the normal pressure and lateral shear force components between the needle tip and the blood vessel wall tissue. The sampling frequency is consistent with the needle tip positioning frequency. The system performs adaptive filtering and noise reduction processing on the collected raw force signals to eliminate periodic force signal interference caused by respiratory movements and vascular pulsation, thus obtaining a stable real-time contact force. It also has a preset safe contact force threshold calibrated by isolated blood vessel experiments. This provides a core basis for force feedback control of deep feed. For the calculated path angle deviation value, the drive mechanism generates a micro-angle deflection torque proportional to the deviation value through a piezoelectric ceramic-based micro-displacement mechanism in conjunction with a flexible hinge. This drive mechanism consists of a micro-stepper motor, a precision lead screw transmission assembly, and a flexible hinge structure. The micro-stepper motor serves as the power source; its rotational motion is converted into linear displacement via the precision lead screw, and then amplified and converted into a precise deflection torque for the flexible puncture needle seat via the flexible hinge. Specifically, the system converts the spatial angle deviation value... Decomposed into deflection components in two orthogonal directions: horizontal and vertical. , The required number of drive pulses for the stepper motor is calculated based on the pre-calibrated motor pulse-angle conversion coefficient. The drive motor outputs the corresponding angular displacement, which is amplified by the transmission of the precision lead screw and flexible hinge. This amplifies the displacement and drives the needle holder to generate a deflection torque that is opposite in direction and in strict proportion to the angular deviation. This achieves precise correction of the needle tip insertion angle until the angular deviation value converges to the preset range of ±0.1°. This driving method has been validated in an in vitro simulated vascular environment. Its output driving force is sufficient to overcome the friction between the puncture needle and the vascular wall tissue, as well as the resistance generated by the bending stiffness of the needle body. This meets the mechanical requirements for dynamic adjustment of the needle insertion angle during clinical puncture. Subsequently, the system applies a positive driving voltage to the piezoelectric ceramic on the opposite side of the deviation direction to make it elongate, and applies a reverse driving voltage to the piezoelectric ceramic on the same side of the deviation direction to make it contract. Through the coordinated extension and contraction of four sets of actuators, a deflection torque opposite to the direction of the angle deviation and in strict proportion to its magnitude is generated, driving the flexible puncture needle tip to precisely correct the needle insertion angle until the angle deviation value converges to the preset ±0.1° range, achieving micro-nano-level precise closed-loop adjustment of the needle insertion angle. While the needle insertion angle is being adjusted in a closed loop, the drive mechanism executes a depth feed operation strongly correlated with real-time contact force data through a high-precision stepper motor. This stepper motor, combined with a ball screw drive structure, can accurately convert the motor step angle into a single-step feed depth of 0.01mm, fully meeting the accuracy requirements of cardiovascular puncture. The system pre-constructs a linear mapping rule between contact force and feed speed. When the real-time contact force... Far below the safety threshold ,Right now When the needle tip is in a soft tissue gap with no resistance, the stepper motor feeds at a preset standard speed. Perform uniform feed, with each feed increment precisely matched to the target depth sequence of the navigation path; when real-time contact force... In When the needle tip has contacted the elastic tissue of the blood vessel wall, the system linearly reduces the feed rate according to the increase in contact force. The feed rate calculation formula is as follows: Simultaneously, the single-step feed rate is reduced to achieve slow feed and highly sensitive force feedback control, preventing the needle tip from penetrating the blood vessel wall; when the real-time contact force... Exceed When the needle insertion angle is adjusted and the contact force returns to a safe range, the system immediately pauses the stepper motor's feeding action. The feeding operation is resumed only after the needle insertion angle is adjusted and the contact force returns to a safe range. Throughout the feeding process, the cumulative feeding depth of the stepper motor is compared with the target depth of the navigation path in real time to ensure that the final needle insertion depth does not deviate from the target depth of the navigation path by more than ±0.05mm. The entire adjustment process is controlled in 10ms cycles. Within each cycle, the needle tip position acquisition, deviation value calculation, contact force acquisition, piezoelectric ceramic deflection drive, and stepper motor feeding control are completed synchronously. This achieves real-time, dynamic, and coordinated precise adjustment of the needle insertion angle and depth, ensuring that the flexible puncture needle travels safely and stably along the real-time navigation path throughout the entire process.

[0026] During the insertion of the flexible puncture needle according to the real-time navigation path and the real-time adjustment of the insertion angle and depth through position and force feedback dual closed-loop control, if the system monitors for three or more consecutive 10ms control cycles and confirms that the contact force data collected by the built-in pressure sensor at the needle tip continuously exceeds the preset safety threshold, this means that the real-time contact force value is higher than the maximum safe contact force upper limit for avoiding mechanical damage to the blood vessel wall, as determined by in vitro blood vessel experiments and clinical data, and the force value shows no downward trend. The system has already eliminated the interference of instantaneous force fluctuations caused by vascular pulsation and respiratory movements through an adaptive filtering algorithm. The system will immediately pause the regular feed action of the stepper motor, terminate the current insertion process, and simultaneously initiate needle tip obstacle avoidance and local path replanning operations. First, the system will extract the temporal gradient change direction of the contact force. The temporal gradient change direction refers to the direction of the increase in contact force with the spatial position of the needle tip in a continuous time series. This is used to accurately locate the pressure point with high resistance and high risk of damage at the needle tip and the optimal avoidance direction. The specific extraction process is as follows: Retrieve contact force time-series data collected synchronously within five consecutive monitoring cycles. And the three-dimensional spatial coordinates of the needle tip at each time point. to The three-dimensional spatial gradient field of the contact force around the current position of the needle tip is obtained by fitting the least squares method. The vector direction of this gradient field is the direction in which the contact force increases fastest with the needle tip displacement, which is also the direction of the high-resistance calcification point and the high-stiffness region of the blood vessel wall. The opposite direction of the gradient vector is the low-resistance safe direction where the contact force gradually decreases. After accurately extracting the direction of the time-domain gradient change, the control core of the drive mechanism will calculate the needle tip obstacle avoidance vector based on this gradient change direction. The needle tip obstacle avoidance vector here refers to the three-dimensional spatial displacement vector that can guide the needle tip to safely avoid the current high-resistance pressure point while not deviating from the safe range of the main navigation path. It includes a clear displacement direction and a controllable displacement amplitude. The specific calculation process is as follows: Using the opposite direction of the contact force time-domain gradient as the main obstacle avoidance direction, and combining the distribution of the low elastic modulus safe zone in the current region of the blood vessel wall elastic modulus distribution map, the maximum amplitude of the obstacle avoidance displacement is limited to 0.3mm to prevent excessive displacement of the needle tip from deviating from the puncture target point of the original navigation path. At the same time, the safety of the obstacle avoidance direction is verified by calling the blood vessel anatomy model to ensure that the direction does not approach prohibited risk areas such as blood vessel bifurcation points and calcified plaques. Finally, a three-dimensional obstacle avoidance vector that meets the spatial safety constraints is generated. Then, the system takes the current three-dimensional coordinates of the needle tip as the starting point of the path, the end point of the obstacle avoidance vector as the intermediate transition node of the path, and the next target coordinate corresponding to the current needle insertion depth in the original real-time navigation path as the end point. A smooth and continuous local path segment with curvature that completely meets the maximum bending limit of the flexible puncture needle is generated by the cubic spline interpolation algorithm. At the same time, the local path segment is quickly collided to ensure that the entire path is within the low elastic modulus safe zone of the blood vessel wall without any overlap with risk areas. After the compliance generation of the local path segment is completed, the drive mechanism first precisely adjusts the flexible puncture needle through a micro stepper motor and a lead screw mechanism. The needle insertion angle is precisely adjusted so that the needle tip is aligned with the starting segment of the local path. A high-precision stepper motor is then driven to perform uniform micro-displacement along the local path segment with a minimum feed step of 0.01 mm / step. During the displacement, the change in needle tip contact force is monitored in real time at an ultra-high sampling frequency of 2 kHz. When the contact force stabilizes and falls below a safe threshold, it is confirmed that the needle tip has successfully avoided the current high-resistance pressure point. The system immediately stops the micro-displacement operation and then readjusts the needle tip insertion angle back to the original real-time navigation path using a piezoelectric ceramic array. Following the tangent direction and completing angle calibration, the system resumes the normal dual-closed-loop needle insertion control process. If the contact force remains above the safety threshold after completing the full displacement along a local path segment, the system will repeat the gradient extraction, obstacle avoidance vector calculation, and local path replanning process, performing obstacle avoidance operations a maximum of three times. If the contact force cannot be reduced to the safe range after three attempts, the system will immediately lock the puncture drive mechanism and push a risk warning to the operator through the human-machine interface. The operator will then manually judge the subsequent operation based on the intraoperative images, ensuring the safety and controllability of the puncture operation throughout the process.

[0027] During the entire puncture and needle insertion process executed by the adaptive puncture execution unit according to the real-time navigation path, the complication risk monitoring unit simultaneously initiates uninterrupted real-time monitoring of vascular injury risk. A high-frequency miniature Doppler ultrasound probe, coaxially mounted with the puncture needle, continuously acquires blood flow echo signals from the vascular region surrounding the needle tip. This high-frequency miniature Doppler ultrasound probe is a 20MHz or higher high-frequency ultrasound probe specifically designed for cardiovascular minimally invasive puncture scenarios. It can accurately capture blood flow motion signals within a 5mm radius around the needle tip, possessing millimeter-level spatial resolution and microsecond-level temporal resolution, fully meeting the accuracy and response speed requirements for real-time intraoperative monitoring. The raw blood flow echo signal acquired by the probe is a time-domain electrical signal that varies with time. The system first performs adaptive bandpass filtering and baseline drift removal preprocessing operations on this time-domain signal to remove low-frequency interference from vascular pulsation and respiratory motion, as well as high-frequency noise generated by the circuit system, obtaining a pure blood flow time-domain signal. Subsequently, a fast Fourier transform (FFT) is used to convert the preprocessed time-domain signal into a frequency-domain signal of the blood flow spectrum. The FFT is a method for efficiently converting discrete time-domain signals into frequency-domain signals. The mathematical algorithm can decompose complex time-domain blood flow signals into a superposition of different frequency components, clearly presenting the frequency distribution characteristics of blood flow motion. The converted frequency domain signal, with frequency as the abscissa and signal energy as the ordinate, fully characterizes the distribution of blood flow at different velocities around the needle tip. After completing the frequency domain conversion, the system accurately extracts the turbulent energy peak and characteristic frequency components from the frequency domain signal of the blood flow spectrum. The turbulent energy peak refers to the maximum energy value of the corresponding turbulent blood flow in the frequency domain signal. Blood flow in normal blood vessels is a regular laminar flow state, and the frequency domain signal energy is concentrated in... Within the low-frequency narrow band, when the blood vessel wall is accidentally penetrated, blood overflows at high speed from the rupture, forming turbulent flow. Its frequency domain signal exhibits a wideband peak with a sharp increase in energy, which is the turbulent energy peak. The characteristic frequency component refers to the specific frequency range corresponding to the turbulent blood flow caused by blood vessel penetration. This range is a characteristic frequency range distinguished from normal laminar flow, vasospasm, and other abnormal blood flow states, defined through extensive in vitro blood vessel penetration experiments and clinical case data. It is typically a wideband range of 1.5kHz-8kHz. The specific extraction process is as follows: First, a sliding window with a fixed step size is used to traverse the frequency domain signal. With a frequency step size of 100Hz, the signal energy integral in each frequency interval is calculated segment by segment. The frequency interval and peak value corresponding to the maximum energy value are marked, which are the turbulence energy peak values. At the same time, 1 is extracted by using a preset characteristic frequency bandpass filter.The system extracts frequency domain signal components within the 5kHz-8kHz range, calculating key parameters such as energy percentage, peak frequency, and spectral width within this range to obtain complete characteristic frequency components. After feature extraction, the system performs real-time vascular penetration compliance determination on blood flow characteristics. First, it continuously monitors the changing trend of turbulent energy peaks and calculates their growth rate within a preset time period. This preset time period is a continuous monitoring window set to eliminate misjudgments caused by instantaneous blood flow disturbances, typically set to three consecutive 20ms monitoring cycles, totaling 60ms. The preset threshold is a critical growth rate calibrated using large clinical sample data to distinguish between normal blood flow fluctuations and vascular penetration, typically set to 500 energy units / millisecond. Within a 60ms preset time period, when the rate of increase of the turbulent energy peak exceeds a preset threshold, the system simultaneously verifies whether the extracted feature frequency components match the vascular penetration pattern. The vascular penetration pattern is a pre-stored frequency domain feature template for vascular penetration, validated through numerous in vitro and in vivo experiments. It includes multiple criteria such as feature frequency range matching degree, spectral bandwidth ratio, energy growth slope, and symmetry of positive and negative frequency components. Only when the matching degree between the feature frequency components and the template exceeds 90% is it considered a match for the vascular penetration pattern. This completely eliminates misjudgments caused by vascular branch blood flow or minor needle vibrations. The system simultaneously verifies whether the rate of increase of the turbulent energy peak exceeds the preset threshold. When the system detects that the characteristic frequency components match the vessel penetration pattern, it immediately determines that a serious complication of accidental vessel wall penetration has occurred. It simultaneously triggers two emergency actions: first, it sends a high-priority alarm signal to the operator via an audible and visual alarm module; second, it immediately activates the coaxial microcatheter occlusion mechanism to seal the vessel rupture. This mechanism includes a microcatheter coaxially fitted onto the outer wall of a flexible puncture needle. The tip of the microcatheter can be pushed along the puncture needle to the vessel rupture site. It is pre-filled with biodegradable medical bioadhesive. When the system detects vessel penetration, it drives the microcatheter forward along the puncture needle, ensuring its tip precisely reaches the vessel wall rupture location. The sealing adhesive is then released via an electrically controlled microvalve, achieving [the desired effect]. The system provides targeted and precise closure of the rupture. The clotting colloid, validated in an in vitro simulated vascular environment, has a coagulation time of 3-5 seconds. The resulting colloidal thrombus possesses sufficient mechanical strength to withstand intravascular blood flow impact (withstanding flow velocities ≥50cm / s) without displacement or fragmentation, ensuring stable and reliable closure. After release, the microcatheter is withdrawn, and changes in the blood flow spectrum are continuously monitored. If the turbulent energy peak gradually returns to the range corresponding to normal laminar flow, the closure is considered effective. If the abnormal state persists, the system will trigger a secondary closure operation and simultaneously send an emergency manual intervention prompt. This system enables real-time monitoring, rapid identification, and precise targeted closure of vascular penetration complications throughout the entire process, maximizing the clinical safety of the puncture procedure.

[0028] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A puncture system for treating cardiovascular diseases, characterized in that, include: The multimodal image fusion unit (1) is used to integrate intravascular ultrasound imaging, X-ray fluoroscopy images and respiratory motion signals in real time, and generate a blood vessel wall elastic modulus distribution map and a blood vessel dynamic motion field through intelligent algorithms. The blood vessel wall elastic modulus distribution map is used to quantify the deformation characteristics of blood vessel tissue. The blood vessel dynamic motion field includes a blood vessel wall surface grid model and its complete motion information at each moment during the respiratory cycle, including grid node displacement vectors, local area deformation rate and direction change characteristics. When the intelligent algorithm generates the elastic modulus distribution map of the blood vessel wall and the dynamic motion field of the blood vessel, it uses a deep learning model to simultaneously process the reflected wave signal of intravascular ultrasound imaging, the spatial coordinate data of blood vessels in X-ray fluoroscopy images, and the temporal waveform of respiratory motion signals. The deep learning model extracts the tissue stiffness features in the reflected wave signal through a convolutional neural network to generate an elastic modulus distribution map that quantifies the deformation capacity of the blood vessel wall. Through a fusion architecture of three-dimensional convolutional neural network and recurrent neural network, it spatiotemporally aligns the three-dimensional surface mesh of the blood vessel wall reconstructed by intravascular ultrasound imaging with the spatial coordinate data of blood vessels in X-ray fluoroscopy images and the temporal waveform of respiratory motion signals, and outputs the surface mesh model of the blood vessel wall and the displacement vector field of the mesh nodes at each time point in the respiratory cycle as the dynamic motion field of the blood vessel. The puncture path planning unit (2) is used to select low elastic modulus areas as puncture targets in the target blood vessel area based on the elastic modulus distribution map of the blood vessel wall, predict the deformation trend of the blood vessel during puncture based on the dynamic motion field of the blood vessel, calculate the blood vessel displacement caused by the deformation trend through real-time simulation, and make reverse corrections to the planned navigation path according to the blood vessel displacement. At the same time, it avoids blood vessel bifurcation and calcification areas by combining the anatomical structure model of the blood vessel, and generates a real-time navigation path leading to the puncture target. When selecting regions with low elastic modulus as puncture targets, the elastic modulus distribution map of the blood vessel wall is input into a region segmentation algorithm. The region segmentation algorithm calculates the elastic modulus value of each pixel unit in the elastic modulus distribution map of the blood vessel wall, marks regions that are consecutively adjacent and whose elastic modulus values ​​are lower than the average value of the elastic modulus distribution map of the blood vessel wall, and outputs regions with low elastic modulus as puncture targets. The adaptive puncture execution unit (3) is used to perform puncture operations according to the real-time navigation path, including a flexible puncture needle, a drive mechanism for driving the flexible puncture needle, and a pressure sensor disposed at the tip of the flexible puncture needle. The pressure sensor is used to monitor the contact force between the flexible puncture needle and the blood vessel wall in real time. The drive mechanism adjusts the insertion angle and depth of the flexible puncture needle according to the deviation signal of the real-time navigation path and the feedback of the contact force. When the contact force exceeds the safety threshold, the needle insertion is paused and the local path is replanned according to the changing trend of the contact force. When adjusting the insertion angle and depth of the flexible puncture needle, the deviation value between the needle tip position coordinates and the coordinate sequence of the real-time navigation path is compared in real time, and the contact force data monitored by the pressure sensor is received simultaneously. The drive mechanism generates a micro-angle deflection torque proportional to the path deviation value through a micro stepper motor and a lead screw mechanism, and executes the depth feed amount associated with the contact force data through the stepper motor. When the contact force data continuously exceeds the safety threshold, the time-domain gradient change direction of the contact force is extracted. The drive mechanism calculates the needle tip obstacle avoidance vector according to the time-domain gradient change direction, generates a local path segment to avoid the current pressure point, and drives the stepper motor to perform displacement along the local path segment before resuming needle insertion. Complication risk monitoring unit (4) is used to monitor the blood flow spectrum of the puncture target area by Doppler ultrasound, and to determine in real time whether the blood vessel has been accidentally penetrated based on the abnormal characteristics of the blood flow spectrum. When the blood vessel penetration characteristic signal is detected, the coaxial microcatheter occlusion mechanism is triggered to perform local occlusion and an alarm signal is issued simultaneously. The frequency domain signal of the blood flow spectrum around the needle tip is acquired by a Doppler ultrasound probe. The turbulent energy peak and characteristic frequency components in the frequency domain signal are extracted by fast Fourier transform. When the growth rate of the energy peak exceeds the preset threshold within a preset time period and the characteristic frequency matches the blood vessel penetration mode, the coaxial microcatheter occlusion mechanism is triggered to release the occlusion colloid.

2. The cardiovascular disease treatment puncture system according to claim 1, characterized in that: When predicting the deformation trend of blood vessels during puncture based on the dynamic motion field of blood vessels, the dynamic motion field of blood vessels is input into the real-time deformation prediction model. The real-time deformation prediction model extracts the displacement rate and direction change characteristics of blood vessels in the dynamic motion field of blood vessels, combines the stiffness parameters of blood vessel wall material contained in the elastic modulus distribution map of blood vessel wall, calculates the deformation response of blood vessels under external force during the respiratory cycle, and outputs time-series prediction data of deformation trend.

3. The cardiovascular disease treatment puncture system according to claim 1, characterized in that: When calculating the vascular displacement caused by the deformation trend through real-time simulation, the time-series prediction data of the deformation trend is input into the finite element simulation engine. The finite element simulation engine constructs a gridded biomechanical model of the vascular wall, loads the time-series prediction data to simulate the stress distribution under puncture pressure, iteratively solves the displacement vector of the grid nodes in the gridded biomechanical model, and outputs the radial and tangential displacement datasets of the vascular vessel.

4. The cardiovascular disease treatment puncture system according to claim 1, characterized in that: When generating the real-time navigation path to the puncture target, a dynamic path planning module is used. The multi-constraint optimization algorithm takes the blood vessel deformation displacement, bifurcation and calcification risk areas, and the distribution of the elastic modulus of the blood vessel wall as parallel constraints to generate the real-time navigation path.

Images