Medical instrument assembly simulation method and device, computer device and storage medium

Abstract

The application relates to a medical instrument assembly simulation method and device, computer equipment and a storage medium. The method comprises the following steps: extracting a blood vessel cavity region from a multi-dimensional blood vessel image through an image recognition algorithm, and generating a three-dimensional surface grid model of the blood vessel cavity region by using a surface reconstruction algorithm; constructing a Voronoi diagram inside the three-dimensional surface grid model, determining a maximum inscribed sphere radius in the three-dimensional surface grid model according to the Voronoi diagram; determining a smooth trunk curve of the three-dimensional surface grid model according to the maximum inscribed sphere radius; determining a target skeleton driving line of a medical instrument to be assembled according to the smooth trunk curve and a branch blood vessel geometric center line in the three-dimensional surface grid model; and performing deformation mapping on a medical instrument reference grid model of the medical instrument to be assembled based on the target skeleton driving line, so as to obtain an assembled instrument grid model. The above scheme can realize high-precision virtual assembly simulation of a medical instrument.

Description

Technical Field

[0001] This application relates to the field of digital medical device technology, and in particular to a medical device assembly simulation method, apparatus, computer equipment, and storage medium. Background Technology

[0002] TEVAR (Thoracic Endovascular Aortic Repair) is the mainstream minimally invasive interventional procedure for the clinical treatment of major vascular diseases such as aortic aneurysm and aortic dissection. For complex lesions in the aortic arch, it is usually necessary to implant branched interventional devices such as branched stents or fenestrated stents along the guidewire to complete the blood flow reconstruction of multiple vessels.

[0003] To advance surgical planning and conduct intraoperative rehearsals, clinicians commonly utilize computer simulation technology to simulate the entire process of stent or balloon delivery, deformation, and assembly within blood vessels. Current mainstream simulation methods fall into two categories: finite element method (FEM) high-precision mechanical simulation and pure geometric path simulation. However, both traditional simulation methods have significant technical limitations. First, FEM simulation is complex, prone to computational problems such as ill-conditioned stiffness matrices and mesh distortion, resulting in extremely long computation times and failing to meet the real-time interactive and rapid iteration requirements of clinical preoperative planning. Second, pure geometric path simulation, based on modeling the vessel centerline extracted from medical images, is susceptible to topological breaks at vessel bifurcation points due to inherent geometric noise in the images. Furthermore, existing algorithms treat bifurcation nodes as static rigid structures, failing to reproduce the dynamic sliding and continuous deformation of the bifurcation region under instrument traction. Ultimately, the simulated configuration of the branch instrument assembly deviates significantly from the actual implantation state. Meanwhile, the thoracic aorta and aortic arch naturally possess complex three-dimensional tortuous structures with high curvature. Existing spatial mapping and mesh deformation algorithms tend to generate stent singularities when processing such curves. This can cause self-torsion in stent simulation models that violates physical laws, and it cannot guarantee the conservation of geodesic length before and after device deformation, leading to axial stretching and compression distortions. This significantly reduces the assembly alignment accuracy of branch stents in the vessel bifurcation plane. Therefore, existing TEVAR device simulation schemes are incompatible with the dynamic topological continuity of branch vessels and cannot simultaneously consider real-time computation and solid geometric conservation constraints. The simulated device deformation and assembly results differ significantly from the actual working conditions within the diseased vessel. The industry urgently needs a spatial assembly simulation method for thoracic aortic endovascular repair devices that adapts to dynamic branching structures and combines high fidelity with anti-torsion characteristics. Summary of the Invention

[0004] Based on this, it is necessary to provide a medical device assembly simulation method, device, computer equipment, and storage medium that takes into account the real-time performance of computation, the dynamic topological continuity of branch vessels, and the geometric conservation characteristics of the device entity in TEVAR device assembly simulation, and reduces the deviation between the simulated device deformation, assembly results, and the actual working conditions of the blood vessels.

[0005] In a first aspect, this application provides a medical device assembly simulation method, the method comprising:

[0006] A multidimensional blood vessel image is acquired, and the vascular lumen region is extracted from the multidimensional blood vessel image using an image recognition algorithm. A surface reconstruction algorithm is then used to generate a three-dimensional surface mesh model of the vascular lumen region.

[0007] A Voronoi diagram is constructed inside the three-dimensional surface mesh model, and the maximum inscribed sphere radius within the three-dimensional surface mesh model is determined based on the Voronoi diagram.

[0008] Based on the maximum inscribed sphere radius, the geometric center line of the main trunk of the three-dimensional surface mesh model and the geometric center line of the branch vessels in the three-dimensional surface mesh model are determined, and the geometric center line of the main trunk is fitted with a B-spline curve to determine a smooth main trunk curve.

[0009] Based on the smooth main curve and the geometric center lines of the branch vessels in the three-dimensional surface mesh model, the target skeleton drive line of the medical device to be assembled is determined.

[0010] Based on the target skeleton driving lines, the medical device reference mesh model of the medical device to be assembled is deformed and mapped to obtain the assembled device mesh model, thereby completing the assembly simulation of the medical device.

[0011] In one embodiment, determining the main geometric centerline of the three-dimensional surface mesh model based on the maximum inscribed sphere radius includes:

[0012] The reciprocal of the square of the maximum inscribed sphere radius is used as a constraint term in the path search cost function, and the entrance endpoint and exit endpoint are determined from the three-dimensional surface mesh model. Based on the entrance endpoint and the exit endpoint, at least one spatial path curve is determined.

[0013] Based on the constraints of the path search cost function and the spatial path curves, the energy integral of the cost function corresponding to each spatial path curve is determined.

[0014] The minimum integral value is determined from the energy integral of the cost function, and the spatial path curve corresponding to the minimum integral value is taken as the geometric center line of the main trunk.

[0015] In one embodiment, determining the target skeleton drive line of the medical device to be assembled, based on the smooth main trunk curve and the geometric center lines of the branch vessels in the three-dimensional surface mesh model, includes:

[0016] Determine whether the medical device to be assembled has a branch structure;

[0017] If the medical device to be assembled does not have a branch structure, then the smooth main curve is used as the target skeleton driving line of the medical device to be assembled.

[0018] If the medical device to be assembled has a branch structure, the geometric center line of the branch blood vessels is extracted from the three-dimensional surface mesh model;

[0019] Based on the geometric centerline of the branch vessels and the smooth trunk curve, the target bifurcation anchor point is determined;

[0020] Determine the distal branch point of the geometric center line of the branch vessel, and construct a third-order Bézier curve based on the target bifurcation anchor point and the distal branch point;

[0021] Based on the third-order Bezier curve, the geometric center line of the branch vessels, and the smooth trunk curve, the target skeleton drive line of the medical device to be assembled is determined.

[0022] In one embodiment, determining the target bifurcation anchor point based on the geometric centerline of the branch vessel and the smooth trunk curve includes:

[0023] Determine the spatial distance from each discrete sampling point on the geometric center line of the branch vessels to the smooth trunk curve, and the nearest projection position of the discrete sampling point to the smooth trunk curve;

[0024] Based on the spatial distance and the maximum inscribed sphere radius of the smooth trunk curve at the nearest projection position, determine the branch detachment segment after the branch vessel's geometric center line detaches from the lumen region of the trunk vessel;

[0025] The target sampling point is determined from the discrete sampling points corresponding to the branch break-off segment based on the distance between the discrete sampling points corresponding to the branch break-off segment and the smooth trunk curve.

[0026] Bezier curve fitting is performed on the target sampling points to determine the branch reference points that characterize the extension position of the branch breakaway segment;

[0027] Based on the branch reference points, candidate trunk line segments are determined from the smooth trunk curve;

[0028] Based on the tangent vector of the smooth trunk curve at the candidate trunk position within the candidate trunk segment, and the unit direction vector of the endpoint of the candidate trunk segment pointing to the branch reference point, the orthogonal matching cost is determined, and the target bifurcation anchor point is determined from the candidate trunk positions.

[0029] In one embodiment, the distal branching point of the branch vessel's geometric centerline is determined, and a third-order Bézier curve is constructed based on the target bifurcation anchor point and the distal branching point, including:

[0030] The target bifurcation anchor point is taken as the first endpoint, and the distal branch point is taken as the second endpoint;

[0031] Determine the maximum inscribed sphere radius corresponding to the distal branch point, and take half of the maximum inscribed sphere radius corresponding to the distal branch point as the first extension length. Based on the first extension length, extend the distal branch point along the tangent direction of the geometric center line of the branch vessel at the distal branch point, and take the extension endpoint as the first intermediate control point.

[0032] Determine the maximum inscribed sphere radius corresponding to the target bifurcation anchor point, and take half of the maximum inscribed sphere radius corresponding to the target bifurcation anchor point as the second extension length. Based on the second extension length, extend the target bifurcation anchor point along the tangent direction of the smooth trunk curve at the target bifurcation anchor point, and take the extension endpoint as the second intermediate control point.

[0033] A third-order Bézier curve is constructed based on the first endpoint, the second endpoint, the first intermediate control point, and the second intermediate control point.

[0034] In one embodiment, based on the target skeleton driving lines, a deformation mapping is performed on the medical device reference mesh model of the medical device to be assembled to obtain an assembled device mesh model, including:

[0035] A smooth frame field is constructed based on the target skeleton driving line using a rotation minimization frame algorithm;

[0036] When the medical device to be assembled has a branch structure, the first normal vector of the blood vessel bifurcation plane is determined by three-dimensional cross product operation between the tangent vector of the smooth trunk curve at the target bifurcation anchor point and the tangent vector at the position closest to the target bifurcation anchor point on the geometric center line of the branch blood vessel.

[0037] Based on the smooth frame field, determine the second normal vector of the device bifurcation plane of the medical device to be assembled;

[0038] With minimizing the angle between the first normal vector and the second normal vector as the optimization objective, a target optimization function is constructed. Based on the target optimization function, the optimal rotation angle is determined, and the pre-rotation matrix corresponding to the optimal rotation angle is determined.

[0039] The pre-rotation matrix is ​​applied to the medical device reference mesh model of the medical device to be assembled to determine the calibrated mesh model;

[0040] When the medical device to be assembled does not have a branch structure, the reference mesh model of the medical device is used as the calibrated mesh model;

[0041] The skeleton nodes of the medical device to be assembled are extracted from the medical device reference mesh model. The starting node of the skeleton nodes is mapped to the target skeleton drive line to determine the first mapping node. Based on the first mapping node and node geometric constraints, the other nodes of the skeleton nodes except the starting node are mapped to the target skeleton drive line to determine the coordinates of the nodes after mapping.

[0042] Based on the mapped coordinates of the nodes and the smoothed frame field, generate the rigid body transformation matrix corresponding to each skeleton node;

[0043] The assembly-state instrument mesh model is determined using a linear hybrid skinning algorithm based on the calibrated mesh model and the rigid body transformation matrix.

[0044] In one embodiment, the assembly-state device mesh model is determined using a linear hybrid skinning algorithm based on the calibrated mesh model and the rigid body transformation matrix, including:

[0045] Determine the initial spatial coordinates of each grid vertex in the calibrated grid model;

[0046] Based on the medical device reference mesh model, an initial straight skeleton is determined, and the skeleton line segment closest to each mesh vertex is determined on the initial straight skeleton;

[0047] By projecting each grid vertex vertically onto its corresponding skeleton line segment, the local centroid coordinate parameters of the projection point of each grid vertex on the skeleton line segment are obtained.

[0048] Based on the line segment index of each skeleton line segment and the local centroid coordinate parameters corresponding to each grid vertex, the continuous parameterized measure value of each grid vertex on the initial straight skeleton is determined;

[0049] The unnormalized weights corresponding to each grid vertex are determined using the Gaussian probability density function, based on the continuous parameterized measure value and standard deviation parameter.

[0050] The unnormalized weights are normalized to determine the skinning influence weights corresponding to each grid vertex.

[0051] The linear hybrid skinning algorithm is used to determine the mapped spatial coordinates of each mesh vertex based on the initial spatial coordinates, the rigid body transformation matrix, and the skinning influence weights corresponding to each mesh vertex.

[0052] The assembled device mesh model is determined based on the mapped spatial coordinates of each mesh vertex.

[0053] Secondly, this application also provides a medical device assembly simulation device, the device comprising:

[0054] The three-dimensional mesh model construction module is used to acquire multi-dimensional blood vessel images, extract the vascular lumen region from the multi-dimensional blood vessel images through image recognition algorithms, and generate a three-dimensional surface mesh model of the vascular lumen region using a surface reconstruction algorithm.

[0055] The inscribed sphere radius determination module is used to construct a Voronoi diagram inside the three-dimensional surface mesh model and determine the maximum inscribed sphere radius within the three-dimensional surface mesh model based on the Voronoi diagram.

[0056] The smooth trunk curve determination module is used to determine the geometric center line of the trunk and the geometric center line of the branch vessels in the three-dimensional surface mesh model based on the maximum inscribed sphere radius, and to perform B-spline curve fitting on the geometric center line of the trunk to determine the smooth trunk curve.

[0057] The target skeleton driving line determination module is used to determine the target skeleton driving line of the medical device to be assembled based on the smooth trunk curve and the geometric center line of the branch blood vessels in the three-dimensional surface mesh model.

[0058] The device mesh model determination module is used to perform deformation mapping on the medical device reference mesh model of the medical device to be assembled based on the target skeleton driving lines, so as to obtain the assembled device mesh model and thus complete the assembly simulation of the medical device.

[0059] Thirdly, this application also provides a computer device, the computer device including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to perform the following steps:

[0060] A multidimensional blood vessel image is acquired, and the vascular lumen region is extracted from the multidimensional blood vessel image using an image recognition algorithm. A surface reconstruction algorithm is then used to generate a three-dimensional surface mesh model of the vascular lumen region.

[0061] A Voronoi diagram is constructed inside the three-dimensional surface mesh model, and the maximum inscribed sphere radius within the three-dimensional surface mesh model is determined based on the Voronoi diagram.

[0062] Based on the maximum inscribed sphere radius, the geometric center line of the main trunk of the three-dimensional surface mesh model and the geometric center line of the branch vessels in the three-dimensional surface mesh model are determined, and the geometric center line of the main trunk is fitted with a B-spline curve to determine a smooth main trunk curve.

[0063] Based on the smooth main curve and the geometric center lines of the branch vessels in the three-dimensional surface mesh model, the target skeleton drive line of the medical device to be assembled is determined.

[0064] Based on the target skeleton driving lines, the medical device reference mesh model of the medical device to be assembled is deformed and mapped to obtain the assembled device mesh model, thereby completing the assembly simulation of the medical device.

[0065] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0066] A multidimensional blood vessel image is acquired, and the vascular lumen region is extracted from the multidimensional blood vessel image using an image recognition algorithm. A surface reconstruction algorithm is then used to generate a three-dimensional surface mesh model of the vascular lumen region.

[0067] A Voronoi diagram is constructed inside the three-dimensional surface mesh model, and the maximum inscribed sphere radius within the three-dimensional surface mesh model is determined based on the Voronoi diagram.

[0068] Based on the maximum inscribed sphere radius, the geometric center line of the main trunk of the three-dimensional surface mesh model and the geometric center line of the branch vessels in the three-dimensional surface mesh model are determined, and the geometric center line of the main trunk is fitted with a B-spline curve to determine a smooth main trunk curve.

[0069] Based on the smooth main curve and the geometric center lines of the branch vessels in the three-dimensional surface mesh model, the target skeleton drive line of the medical device to be assembled is determined.

[0070] Based on the target skeleton driving lines, the medical device reference mesh model of the medical device to be assembled is deformed and mapped to obtain the assembled device mesh model, thereby completing the assembly simulation of the medical device.

[0071] The aforementioned medical device assembly simulation method, apparatus, computer equipment, and storage medium acquire multidimensional vascular images and extract the vascular lumen region from the multidimensional vascular images using image recognition algorithms. A surface reconstruction algorithm is then used to generate a three-dimensional surface mesh model of the vascular lumen region. A Voronoi diagram is constructed within the three-dimensional surface mesh model, and the maximum inscribed sphere radius within the three-dimensional surface mesh model is determined based on the Voronoi diagram. Based on the maximum inscribed sphere radius, the geometric center lines of the main trunk and branch vessels in the three-dimensional surface mesh model are determined, and B-spline curve fitting is performed on the main trunk geometric center line to determine a smooth main trunk curve. Based on the smooth main trunk curve and the branch vessel geometric center lines in the three-dimensional surface mesh model, the target skeleton driving line of the medical device to be assembled is determined. Based on the target skeleton driving line, deformation mapping is performed on the medical device reference mesh model of the medical device to be assembled to obtain the assembled device mesh model, thereby completing the assembly simulation of the medical device. This solution addresses the limitations of existing TEVAR device simulation schemes, which struggle to maintain the dynamic topological continuity of branch vessels and simultaneously compromise computational real-time performance with geometrical conservation constraints. This results in significant discrepancies between simulated device deformation and assembly outcomes and the actual working conditions within diseased vessels. The proposed solution reconstructs a 3D surface mesh model of the vessel lumen based on multidimensional vascular images of the patient. It then uses a Voronoi diagram to determine the maximum inscribed sphere radius, extracts the geometric centerline of the main vessel at the lumen center based on this radius, and obtains a smooth main vessel curve through B-spline fitting. This, combined with the branch vessel centerlines, constructs a target skeleton driving line adapted to the vessel morphology. Based on this target skeleton driving line, the device mesh model undergoes deformation mapping to achieve virtual assembly simulation. This approach adapts to individualized vascular anatomy in different patients, ensuring the extracted vessel centerline remains stably centered within the vessel lumen and avoiding simulation errors caused by centerline deviation and proximity to the vessel wall. Furthermore, the smoothed drive skeleton can adapt to complex vascular morphologies including bifurcations, making the mesh deformation of the medical device to be assembled more continuous and smooth during the simulation assembly process. This reduces mesh distortion during the simulation process, thereby accurately reproducing the assembly morphology of the medical device to be assembled in the blood vessel, improving the accuracy and practicality of virtual assembly simulation before vascular intervention, and providing reliable data support for surgical plan rehearsal. Attached Figure Description

[0072] Figure 1 This is a flowchart illustrating a medical device assembly simulation method in one embodiment;

[0073] Figure 2 This is a schematic diagram of a medical device assembly simulation system in one embodiment;

[0074] Figure 3 This is a structural block diagram of a medical device assembly simulation device in one embodiment;

[0075] Figure 4This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0076] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0077] In one embodiment, such as Figure 1 As shown, a medical device assembly simulation method is provided. This embodiment illustrates the method's application to a terminal. It is understood that this method can also be applied to a server, and further to a system including both a terminal and a server, and is implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0078] S110. Acquire multidimensional blood vessel images, extract the vascular cavity region from the multidimensional blood vessel images using image recognition algorithms, and generate a three-dimensional surface mesh model of the vascular cavity region using surface reconstruction algorithms.

[0079] The multidimensional vascular images can be multi-angle vascular CT images, MRI (Magnetic Resonance Imaging) images, or vascular ultrasound images. The vascular lumen region is extracted after segmentation processing of medical images such as CT or MRI; it is the entire hollow tubular flow space inside the vessel wall, including the main vessel lumen and the lumens of each branch vessel. The image recognition algorithm can be a pre-trained deep learning model. The surface reconstruction algorithm can be the moving cube algorithm.

[0080] Specifically, multidimensional vascular images of patients requiring endovascular aortic repair are acquired. Using image recognition algorithms, based on image grayscale features and vascular anatomical topology, the vascular lumen region is extracted from the multidimensional vascular images. A moving cubes algorithm is then used to generate a three-dimensional surface mesh model of the vascular lumen region from the image.

[0081] S120. Construct a Voronoi diagram inside the three-dimensional surface mesh model, and determine the maximum inscribed sphere radius within the three-dimensional surface mesh model based on the Voronoi diagram.

[0082] The maximum inscribed sphere refers to the largest sphere that can be contained within the vessel wall at a given location without exceeding its inner diameter. Its radius directly reflects the diameter of the vessel lumen at that location. The Voronoi diagram is a spatial region partitioning geometric model. It spatially divides all voxels within a 3D surface mesh model, using each sampling point as a unique generator to divide the entire 3D space of the vessel lumen into several independent sub-regions. The distance from any point within each sub-region to its own generator is less than its distance to all other generators. After constructing a 3D Voronoi diagram within the 3D surface mesh model, the inscribed sphere corresponding to each sub-region represents the width of the vessel lumen at that location, and the radius of the sphere in each region is the radius of the inscribed sphere at that location. Traversing all regions yields the maximum inscribed sphere radius at each point in the vessel lumen.

[0083] Specifically, the 3D surface mesh model is discretized using voxelization. The sampled set of spatial points is used as generators, and a Voronoi diagram is constructed within the constraint range of the vascular cavity region based on these generators. The maximum inscribed sphere radius is then determined for each point within the 3D surface mesh model based on the Voronoi diagram.

[0084] S130. Based on the maximum inscribed sphere radius, determine the geometric center line of the main trunk of the three-dimensional surface mesh model and the geometric center line of the branch vessels in the three-dimensional surface mesh model, and perform B-spline curve fitting on the geometric center line of the main trunk to determine the smooth main trunk curve.

[0085] Among them, B-spline curve fitting is also known as B-spline curve fitting.

[0086] For example, determining the backbone geometric centerline of a 3D surface mesh model based on the radius of the maximum inscribed sphere includes:

[0087] The reciprocal of the square of the maximum inscribed sphere radius is used as a constraint term in the path search cost function. The entrance and exit endpoints are determined from the 3D surface mesh model, and at least one spatial path curve is determined based on the entrance and exit endpoints. Based on the constraint term of the path search cost function and the spatial path curve, the cost function energy integral corresponding to each spatial path curve is determined. The minimum integral value is determined from the cost function energy integral, and the spatial path curve corresponding to the minimum integral value is used as the main geometric centerline.

[0088] It should be noted that in the medical device assembly simulation process for endovascular aortic repair, the geometric centerline of the main trunk is the core geometric benchmark for constructing the reference delivery track and drive frame. Its centering accuracy directly affects the accuracy of subsequent device pose calculation and medical device assembly simulation. The closer the spatial path curve is to the vessel wall, the larger the energy integral of the cost function corresponding to the spatial path curve; the more centrally located the spatial path curve is, the smaller the energy integral of the cost function corresponding to the spatial path curve.

[0089] Specifically, the reciprocal of the square of the maximum inscribed sphere radius at each point inside the 3D surface mesh model is used as a constraint term in the path search cost function, transforming the extraction of the main geometric centerline into solving the optimal path problem that minimizes the energy integral.

[0090] The expression for the energy integral of the cost function corresponding to the spatial path curve is shown in Equation (1).

[0091] (1).

[0092] in, Let s represent the spatial path curve connecting the inlet and outlet endpoints, x be the arc length of the spatial path curve, x be the coordinates of a spatial point on the curve, and R(x) be the maximum radius of the inscribed sphere at spatial point x. The closer x is to the center of the lumen, the larger R(x) becomes.

[0093] After determining the cost function energy integral corresponding to each spatial path curve, the minimum integral value is determined from the cost function energy integral, and the spatial path curve corresponding to the minimum integral value is taken as the main geometric center line.

[0094] It should be noted that the geometric centerline of the main trunk obtained above is only formed by connecting a set of discrete three-dimensional spatial coordinate points sequentially. It is a piecewise polygonal line, meaning that the points are straight line segments, and there are sharp angles at the inflection points. If the geometric centerline of the main trunk is directly used as the skeleton driving line for medical device assembly simulation, distortion problems such as geometric discontinuity and kinking distortion will occur when simulating the deformation of the medical device to be assembled along the bend of the blood vessel. Geometric discontinuity refers to abrupt changes in the line at the inflection points of the polygonal line, resulting in abrupt breaks in the deformation of the stent model. Kinking distortion refers to the fact that when matching the large tortuous spatial morphology of the aortic arch, the sharp angles of the polygonal line will force the stent to produce kinks and bends that do not conform to the actual physical state, resulting in a large deviation between the medical device assembly simulation effect and the actual assembly state of the medical device. By performing B-spline curve fitting on the geometric centerline of the main trunk, a smooth main trunk curve can be obtained.

[0095] For example, the formula for B-spline curve fitting is shown in formula (2).

[0096] (2).

[0097] Where C(u) is the fitted smooth backbone curve, P i These are the fitting control points determined based on the original set of discrete center points. Let p be the B-spline basis functions. The curve length and the degree of curve distortion of the main geometric center line are adaptively determined, and u is a node vector parameter that controls the position of the curve segments.

[0098] The above scheme, by minimizing the energy integral of the cost function, allows the path search algorithm to actively choose to traverse the central region of the lumen with a larger inscribed sphere radius, avoiding edge positions close to the vessel wall with a smaller inscribed sphere radius. The resulting main trunk geometric centerline can fit the geometric center of the vessel lumen throughout the entire process, providing a high-precision axial reference for subsequent construction of the benchmark main trunk delivery track, generation of the drive skeleton, and spatial deformation assembly of the device.

[0099] S140. Based on the smooth main curve and the geometric center lines of the branch vessels in the three-dimensional surface mesh model, determine the target skeleton drive line of the medical device to be assembled.

[0100] Among them, the medical devices to be assembled refer to interventional implantation devices that need to be delivered along the aortic guidewire and positioned and fitted inside the blood vessel lumen. The core includes special medical devices for endovascular repair of the thoracic aorta, such as branched stents, fenestrated stents, and vascular balloons.

[0101] For example, the methods for determining the target skeleton driving line include:

[0102] S1401. Determine whether the medical device to be assembled has a branch structure.

[0103] Among them, branch structures refer to secondary stent branches or fenestrated extensions extending from the medical device to be fitted. These are device-attached structures adapted to branch vessels such as the brachiocephalic artery of the aortic arch, the left common carotid artery, or the left subclavian artery. Conventional straight stents do not have branch structures and can only be fitted to a single main vessel; while fenestrated or branched stents have small, outwardly extending branch stent segments. During implantation, the main trunk is attached to the main aortic trunk, and the branch segments need to extend into the branch vessels to complete blood flow reconstruction. These outwardly extending secondary medical device segments are called branch structures.

[0104] S1402. If the medical device to be assembled does not have a branch structure, the smooth main curve will be used as the target skeleton driving line of the medical device to be assembled.

[0105] Among them, the target skeleton driving line is a three-dimensional reference curve used to constrain the spatial movement of the medical device to be assembled and control the deformation of the device mesh.

[0106] Specifically, if the medical device to be assembled is a straight tube stent without a branch structure, the smooth main curve is directly used as the target skeleton driving line of the medical device to be assembled.

[0107] S1403. If the medical device to be assembled has a branch structure, extract the geometric center line of the branch blood vessels from the three-dimensional surface mesh model.

[0108] Specifically, if the medical device to be assembled has a branched structure, a three-dimensional Voronoi diagram is constructed based on the three-dimensional surface mesh model of the blood vessel lumen. The radius of the maximum inscribed sphere within the three-dimensional surface mesh model is determined. The reciprocal of the square of the maximum inscribed sphere radius is used as a constraint term in the path search cost function. The bifurcation port point and the distal endpoint of the branch vessel are determined from the three-dimensional surface mesh model, and at least one branch vessel path curve is determined based on the bifurcation port point and the distal endpoint of the branch. Based on the constraint term of the path search cost function and the branch vessel path curve, the energy integral of the cost function corresponding to each branch vessel path curve is determined. According to the energy integral of the cost function corresponding to each branch vessel path curve, the initial centerline of the branch is determined from the branch vessel path curve. B-spline curve fitting is performed on the initial centerline of the branch to generate a smooth geometric centerline of the branch vessel.

[0109] S1404. Based on the geometric centerline of the branch vessels and the smooth trunk curve, determine the target bifurcation anchor point.

[0110] The target bifurcation anchor point is located in the intersection area of ​​the main aortic trunk and branch vessels, and serves as the spatial reference point for achieving geometric connection between the smooth trunk curve and the geometric center line of the branch vessels.

[0111] For example, determining the target bifurcation anchor point based on the geometric centerline of the branch vessels and the smooth trunk curve includes:

[0112] The spatial distances from each discrete sampling point on the geometric centerline of the branch vessel to the smooth trunk curve, and the nearest projection position of the discrete sampling point to the smooth trunk curve, are determined. Based on the spatial distances and the maximum inscribed sphere radius of the smooth trunk curve at the nearest projection position, the branch detachment segment after the branch vessel's geometric centerline detaches from the lumen of the trunk vessel is determined. Based on the distance between the discrete sampling points corresponding to the branch detachment segment and the smooth trunk curve, a target sampling point is determined from the discrete sampling points corresponding to the branch detachment segment. A Bezier curve is fitted to the target sampling point to determine a branch reference point characterizing the extension position of the branch detachment segment. Based on the branch reference point, candidate trunk segments are determined from the smooth trunk curve. Based on the tangent vector of the smooth trunk curve at the candidate trunk position within the candidate trunk segment, and the unit direction vector of the endpoint of the candidate trunk segment pointing to the branch reference point, the orthogonal matching cost is determined, and the target bifurcation anchor point is determined from the candidate trunk positions.

[0113] Discrete sampling points can be points obtained by randomly sampling the geometric centerline of branch vessels. Candidate trunk locations refer to the sampling points on the candidate trunk segment.

[0114] Specifically, all discrete sampling points on the geometric center line of the branch vessels are traversed, and the nearest projection position of each discrete sampling point on the smooth trunk curve is calculated. The maximum inscribed sphere radius corresponding to the smooth trunk curve at the nearest projection position is obtained. This maximum inscribed sphere radius represents the equivalent radius of the lumen of the trunk vessel. The spatial distance from the discrete sampling point to the smooth trunk curve is compared with the maximum inscribed sphere radius of the smooth trunk curve at the nearest projection position. If the spatial distance corresponding to the discrete sampling point is less than or equal to the maximum inscribed sphere radius of the smooth trunk curve at the nearest projection position, the discrete sampling point is determined to be within the lumen coverage area of ​​the trunk vessel. If the spatial distance corresponding to the discrete sampling point is greater than the maximum inscribed sphere radius of the smooth trunk curve at the nearest projection position, the discrete sampling point is determined to be outside the lumen region of the trunk vessel. The state transition point from the lumen coverage area to the outside of the lumen is found on the geometric center line of the branch vessel as the boundary position of the branch vessel's geometric center line leaving the trunk region. The boundary position can be obtained by interpolation between two adjacent discrete sampling points with different states. The branch center line point set corresponding to the side of the boundary position away from the smooth trunk curve is determined as the branch detachment segment. Among them, there are two states: the discrete sampling point is within the lumen coverage area of ​​the trunk vessel and the discrete sampling point is outside the lumen region of the trunk vessel. The point where the two states switch is the state jump point along the geometric center line of the branch vessel, from the area covered by the lumen of the main vessel to the outside of the lumen of the main vessel.

[0115] From the discrete sampling points corresponding to the branch breakaway segment, target sampling points whose distance from the smooth main curve is less than a distance threshold are determined. In other words, target sampling points are discrete sampling points in the branch breakaway segment that are close to the smooth main curve. These target sampling points are used as transition fitting points, and Bézier curves are fitted to them to determine the initial Bézier curve. The internal control points of the initial Bézier curve on the side closest to the smooth main curve are used as branch reference points.

[0116] It should be noted that the initial Bézier curve is controlled by two endpoints and several internal control points. The two endpoints correspond to the start and end positions of the initial Bézier curve, while the internal control points do not fall on the initial Bézier curve but determine the bending direction and extension trend of the initial Bézier curve.

[0117] The branch reference point is projected onto the smooth trunk curve, and the projected position of the branch reference point on the smooth trunk curve is taken as the nearest position. The nearest position is used as the central reference for searching the bifurcation anchor point. Based on a preset sampling range and centered on the nearest position, the smooth trunk curve is truncated and sampled. The truncated sampled curve is used as the candidate trunk line segment, thereby limiting the search range of the target bifurcation anchor point.

[0118] The orthogonal matching cost is determined based on the tangent vector of the smooth trunk curve at the candidate trunk position within the candidate trunk segment, and the unit direction vector of the endpoint of the candidate trunk segment pointing to the branch reference point. The formula for calculating the orthogonal matching cost is shown in formula (3).

[0119] (3).

[0120] in, Let t represent the orthogonal matching cost corresponding to the i-th candidate backbone position. i This represents the tangent vector of the smoothed main curve at the i-th candidate main curve position. and These represent the unit direction vectors pointing from the endpoints of the two candidate trunk line segments to the branch reference point.

[0121] Orthogonal matching cost is used to quantitatively evaluate the spatial distribution orientation of branch reference points relative to candidate trunk segments. The smaller the orthogonal matching cost value, the closer the branch reference points are to being distributed along a direction perpendicular to the local tangential direction of the trunk, and the better they match the geometry of the actual bifurcation of the blood vessel. The candidate trunk position corresponding to the smallest orthogonal matching cost value is selected as the target bifurcation anchor point.

[0122] The above scheme identifies branch detachment segments in the lumen of the branch vessel by calculating the spatial distance and nearest projection position of discrete sampling points on the geometric center line of the branch vessel relative to the smooth trunk curve, combined with the maximum inscribed sphere radius at the projection position. Target sampling points are then selected from these detachment segments, and Bézier curves are used to fit branch reference points. This reduces interference from noise in the original center line sampling and accurately characterizes the spatial position and extension trend of the branch root. Candidate trunk segments are delineated using the branch reference point as a benchmark to narrow the search range. The orthogonal matching cost is then calculated using the tangent vector of the candidate trunk segments and the unit direction vector pointing from the segment endpoints to the branch reference point to determine the target bifurcation anchor point. This comprehensively evaluates the orthogonal bifurcation posture of the branch relative to the trunk, avoiding positioning errors caused by directly using the original discrete sampling points and improving the robustness of bifurcation anchor point positioning.

[0123] S1405. Determine the distal branching point of the geometric center line of the branching vessel, and construct a third-order Bézier curve based on the target bifurcation anchor point and the distal branching point.

[0124] The distal branch point is the target connection point on the branch centerline that is far from the main trunk. A third-order Bézier curve is a continuous, smooth three-dimensional parametric curve constrained by two endpoints and two intermediate control points.

[0125] For example, determining the distal branching point of the branch vessel's geometric centerline and constructing a third-order Bézier curve based on the target bifurcation anchor point and the distal branching point includes:

[0126] The target bifurcation anchor point is taken as the first endpoint, and the distal branch point is taken as the second endpoint. The maximum inscribed sphere radius corresponding to the distal branch point is determined, and half of the maximum inscribed sphere radius corresponding to the distal branch point is taken as the first extension length. Based on the first extension length, the distal branch point is extended along the tangent direction of the branch vessel geometric center line at the distal branch point, and the extension endpoint is taken as the first intermediate control point. The maximum inscribed sphere radius corresponding to the target bifurcation anchor point is determined, and half of the maximum inscribed sphere radius corresponding to the target bifurcation anchor point is taken as the second extension length. Based on the second extension length, the target bifurcation anchor point is extended along the tangent direction of the smooth trunk curve at the target bifurcation anchor point, and the extension endpoint is taken as the second intermediate control point. A third-order Bézier curve is constructed based on the first endpoint, the second endpoint, the first intermediate control point, and the second intermediate control point.

[0127] The above scheme uses the target bifurcation anchor point and the distal branch point as the two endpoints of the third-order Bézier curve. It generates two intermediate control points by combining the tangent directions at the two locations and the maximum inscribed sphere radius constraint. Based on the two endpoints and the two intermediate control points, the third-order Bézier curve is constructed. The third-order Bézier curve can be used as a transition segment between the smooth main curve and the geometric center line of the branch blood vessel, eliminating the sharp corners of the path at the bifurcation location, realizing the smooth connection between the main and branch paths, ensuring the continuous and smooth drive line of the target skeleton, and avoiding abrupt changes in posture and abnormal bending during the assembly simulation of medical devices with branches.

[0128] S1406. Based on the third-order Bézier curve, the geometric center line of the branch vessels, and the smooth trunk curve, the target skeleton drive line of the medical device to be assembled is determined.

[0129] Specifically, a third-order Bézier curve is used as a transition segment between the smooth main curve and the geometric center line of the branch vessels to connect the smooth main curve and the geometric center line of the branch vessels, thereby determining the target skeleton drive line of the medical device to be assembled.

[0130] The above-described scheme directly uses a smooth trunk curve as the target skeleton driving line for medical devices without branching structures, resulting in a simple and efficient process. For medical devices with branching structures, the geometric center lines of the branch vessels are extracted and the target bifurcation anchor points are solved. A third-order Bézier curve is constructed to achieve a smooth transition between the trunk curve and the branch vessel center lines, forming the target skeleton driving line for the medical device to be assembled. This eliminates sharp angles at vessel bifurcation points, ensures a continuous and smooth driving skeleton, accurately matches the assembly shape of the branch stent, effectively avoids abrupt twisting and abnormal bending of the device during simulation, and improves the realism of simulation results for the delivery and bifurcation alignment of interventional devices with branches.

[0131] S150. Based on the target skeleton driving line, the medical device reference mesh model to be assembled is deformed and mapped to obtain the assembled device mesh model, thereby completing the assembly simulation of the medical device.

[0132] For example, the methods for determining the mesh model of an assembled device include:

[0133] S1501. A smooth frame field is constructed based on the target skeleton driving line by using the rotation minimization frame algorithm.

[0134] The Rotation Minimizing Frame (RMF) algorithm is a local orthogonal frame field established along a spatial curve. During continuous travel along the curve, it minimizes unnecessary spin twisting of the frame around the curve's tangent. A smooth RMF frame field is generated based on the target skeleton's driving lines, providing a continuous and stable local attitude reference for the medical device to be assembled. This avoids the frame flipping and violent rotation problems that easily occur when directly using Frenet frames. Each skeleton node in the smooth frame field corresponds to a tangent vector, a normal vector, and a binormal vector along the skeleton's forward direction. These three vectors together determine the spatial pose of the skeleton node.

[0135] Specifically, a series of spatial discrete sampling points are sampled along the target skeleton driving line. At each sampling point, the rotation minimization frame algorithm is used to recursively solve the orthogonal local frame, constraining the rotation amount of adjacent frames around the tangent vector of the curve to be minimized, thereby generating a continuous smooth frame field along the entire target skeleton driving line.

[0136] For example, the relationship between the normal vectors of adjacent skeleton nodes on the target skeleton driving line is expressed as Equation (4).

[0137] (4).

[0138] Where, N i N represents the normal vector of the i-th skeleton node on the target skeleton driving line. i+1 Let v represent the normal vector of the (i+1)th skeleton node on the target skeleton driving line, and v be the reflection vector determined based on the tangent vectors corresponding to adjacent skeleton nodes. The reflection symmetry axis vector between the normal vectors of two adjacent skeleton nodes can be obtained by summing and normalizing the normal vectors of the two adjacent skeleton nodes.

[0139] S1502. When the medical device to be assembled has a branch structure, the first normal vector of the blood vessel bifurcation plane is determined by three-dimensional cross product operation between the tangent vector of the smooth main curve at the target bifurcation anchor point and the tangent vector at the position closest to the target bifurcation anchor point on the geometric center line of the branch blood vessel.

[0140] Among them, the three-dimensional cross product operation is to calculate the cross product of two three-dimensional vectors.

[0141] Specifically, through three-dimensional cross product operation, the tangent vector of the smooth main trunk curve at the target bifurcation anchor point and the branch tangent vector at the position closest to the target bifurcation anchor point on the geometric center line of the branch vessel are solved. The normal vector of the vessel bifurcation plane between the tangent vector of the smooth main trunk curve at the target bifurcation anchor point and the branch tangent vector at the position closest to the target bifurcation anchor point on the geometric center line of the branch vessel is determined as the first normal vector.

[0142] S1503. Based on the smooth frame field, determine the second normal vector of the device bifurcation plane of the medical device to be assembled.

[0143] The device bifurcation plane is the plane formed by the tangent vector of the main stent axis at the device bifurcation position and the tangent vector of the branch stent axis at the device bifurcation position on the medical device to be assembled with branches.

[0144] Specifically, based on the local spatial reference provided by the smooth frame field, the tangent vectors of the main stent axis and the branch stent axis at the device bifurcation position of the medical device to be assembled are obtained. A three-dimensional cross product operation is performed on the tangent vectors of the main stent axis and the branch stent axis at the device bifurcation position to determine the second normal vector of the device bifurcation plane of the medical device to be assembled.

[0145] S1504. With minimizing the angle between the first normal vector and the second normal vector as the optimization objective, construct the objective optimization function, determine the optimal rotation angle based on the objective optimization function, and determine the pre-rotation matrix corresponding to the optimal rotation angle.

[0146] The pre-rotation matrix is ​​used to perform attitude pre-calibration on the original mesh of the device, so that the bifurcation orientation of the device matches the natural bifurcation direction of the blood vessel. The objective optimization function uses the rotation angle of the device around the main axis as the independent variable, and is used to quantify the difference in orientation between the first normal vector of the blood vessel bifurcation plane and the second normal vector of the device bifurcation plane.

[0147] Specifically, a target optimization function can be constructed to quantify the difference in orientation between the first and second normal vectors. This target optimization function can be: F(θ) = arccos(z, w(θ)), where z is the first normal vector and w(θ) is the second normal vector that varies with the rotation angle θ. By iterating through all rotation angles, the target optimization function is solved, and the minimum value of the solution is determined. The rotation angle corresponding to the minimum value of the solution is the optimal rotation angle. The tangent vector of the main curve at the target bifurcation anchor point is selected as the rotation axis. A three-dimensional rotation matrix is ​​generated around the rotation axis to obtain the optimal rotation angle, according to the rotation matrix construction formula, thus obtaining the pre-rotation matrix.

[0148] S1505. Apply the pre-rotation matrix to the medical device reference mesh model of the medical device to be assembled, and determine the calibrated mesh model.

[0149] Among them, the medical device reference mesh model of the medical device to be assembled is the original three-dimensional mesh model of the medical device to be assembled in the initial gripping state.

[0150] Specifically, the pre-rotation matrix is ​​applied to the coordinates of all vertices of the medical device reference mesh model to be assembled with the medical device, and the overall attitude rotation calibration is completed to obtain the calibrated mesh model.

[0151] S1506. When the medical device to be assembled does not have a branch structure, the medical device reference mesh model shall be used as the calibrated mesh model.

[0152] S1507. Extract the skeleton nodes of the medical device to be assembled from the medical device reference mesh model, map the starting node in the skeleton nodes onto the target skeleton driving line, determine the first mapping node, and based on the first mapping node and node geometric constraints, map the other nodes in the skeleton nodes except the starting node onto the target skeleton driving line, and determine the coordinates of the nodes after mapping.

[0153] In this context, the skeleton nodes are discrete driving nodes predefined on the instrument reference mesh. The node geometric constraints are distance conservation constraints between adjacent skeleton nodes.

[0154] For example, the expression for the node geometric constraints is shown in Equation (5).

[0155] (5).

[0156] Among them, P l Let P be the spatial coordinates of the mapped node formed by mapping the l-th skeleton node to the target skeleton driving line. l-1 Let L be the spatial coordinates of the mapped node formed by mapping the (l-1)th skeleton node to the target skeleton driving line. l The distance between the l-th skeleton node and the (l-1)-th skeleton node when the medical device to be assembled is in the initial gripping state.

[0157] It should be noted that cardiovascular stents have the physical characteristic of being difficult to stretch axially, therefore the distance between the skeleton nodes is conserved. Skeleton nodes of the medical device to be assembled are extracted from the medical device reference mesh model. The starting node in the skeleton nodes is mapped onto the target skeleton drive line to determine the first mapped node. Based on the first mapped node of the starting node on the target skeleton drive line and the node geometric constraints, the node coordinates of all other nodes in the skeleton nodes except the starting node mapped onto the target skeleton drive line are calculated sequentially. The coordinates of the first mapped node and the node coordinates of all other nodes in the skeleton nodes except the starting node mapped onto the target skeleton drive line are integrated to determine the mapped node coordinates.

[0158] S1508. Based on the coordinates after node mapping and the smooth frame field, generate the rigid body transformation matrix corresponding to each skeleton node.

[0159] Specifically, based on the mapped coordinates of the nodes, the local spatial reference of each node is obtained by combining the smooth frame field on the target skeleton driving line. The translation and rotation transformations of each skeleton node from the original coordinate system of the instrument to the spatial coordinate system of the blood vessel are solved in turn, and the rigid body transformation matrix corresponding to each skeleton node is obtained by combining them.

[0160] S1509. Using a linear hybrid skinning algorithm, determine the assembly-state instrument mesh model based on the calibrated mesh model and the rigid body transformation matrix.

[0161] For example, the assembly-state device mesh model is determined using a linear hybrid skinning algorithm based on the calibrated mesh model and the rigid body transformation matrix, including:

[0162] The process involves: determining the initial spatial coordinates of each grid vertex in the calibrated mesh model; determining the initial linear skeleton based on the medical device reference mesh model, and identifying the skeleton line segment closest to each grid vertex on the initial linear skeleton; projecting each grid vertex vertically onto its corresponding skeleton line segment to obtain the local centroid coordinate parameters of the projection point of each grid vertex on the skeleton line segment; determining the continuous parameterized measure value of each grid vertex on the initial linear skeleton based on the line segment index of each skeleton line segment and the local centroid coordinate parameters of each grid vertex; determining the unnormalized weight corresponding to each grid vertex using the Gaussian probability density function, based on the continuous parameterized measure value and standard deviation parameter; normalizing the unnormalized weight to determine the skin influence weight corresponding to each grid vertex; determining the mapped spatial coordinates of each grid vertex using the linear hybrid skinning algorithm, based on the initial spatial coordinates, rigid body transformation matrix, and the skin influence weight corresponding to each grid vertex; and determining the assembled device mesh model based on the mapped spatial coordinates of each grid vertex.

[0163] The standard deviation parameter is set manually in advance.

[0164] Specifically, the initial spatial coordinates of each grid vertex in the calibrated mesh model are determined; based on the medical device reference mesh model, an initial linear skeleton is determined, and the skeleton line segment closest to each grid vertex is identified on the initial linear skeleton, along with its segment index. Each grid vertex is then vertically projected onto its corresponding skeleton line segment to obtain the local centroid coordinate parameters of the projection point of each grid vertex on the skeleton line segment. ,in Based on the line segment index of each skeleton line segment and the local centroid coordinate parameters corresponding to each mesh vertex, the continuous parametric measure value of each mesh vertex on the initial straight skeleton is determined.

[0165] For example, for the j-th mesh vertex, the nearest skeleton line segment to the mesh vertex is found on the initial skeleton of the medical device, and the mesh vertex is orthogonally projected onto the skeleton line segment.

[0166] The formula for calculating the continuous parameterized measure value corresponding to the j-th grid vertex is shown in formula (6).

[0167] (6).

[0168] Where, μ j For continuous parameterized measure values, I seg This is the segment index of the skeleton segment corresponding to the mesh vertex. The local centroid coordinates of the projection point corresponding to the j-th mesh vertex on the skeleton line segment.

[0169] By using the Gaussian probability density function, based on continuous parameterized measure values ​​and standard deviation parameters, the unnormalized weights corresponding to each grid vertex are determined.

[0170] For example, the formula for calculating the unnormalized weight of the k-th skeleton node to the mesh vertex j is shown in formula (7).

[0171] (7).

[0172] in, Let σ be the unnormalized weight of the k-th skeleton node relative to mesh vertex j, and σ be a pre-set standard deviation parameter controlling the smoothing range of the weights. For example, σ can be 1. It represents the continuous parameterized measure value corresponding to the k-th skeleton node.

[0173] Normalize the unnormalized weights so that the total skinning influence weight corresponding to a single mesh vertex is 1, and determine the skinning influence weight corresponding to each mesh vertex.

[0174] The linear hybrid skinning algorithm determines the mapped spatial coordinates of each mesh vertex based on the initial spatial coordinates, rigid body transformation matrix, and skinning influence weights corresponding to each mesh vertex; and the assembled device mesh model is determined based on the mapped spatial coordinates of each mesh vertex.

[0175] For example, the expression for the mapped spatial coordinates of the mesh vertices is shown in Equation (8).

[0176] (8).

[0177] Among them, v j v' represents the spatial coordinates of mesh vertex j before mapping. j M represents the spatial coordinates of mesh vertex j after mapping. k Let w be the rigid body transformation matrix corresponding to the k-th skeleton node. j,k The skinning influence weight of mesh vertex j is affected by skeleton node k.

[0178] The above scheme establishes continuous parametric measures based on the initial straight skeleton, solves the normalized skin weights using Gaussian functions, and completes mesh deformation mapping with the help of linear hybrid skin. This can smoothly drive the device mesh to follow the skeleton to deform, effectively suppress mesh distortion, restore the assembly morphology of the medical device to be assembled in the blood vessel, and improve the accuracy of virtual simulation assembly results.

[0179] In the aforementioned medical device assembly simulation method, multidimensional vascular images are acquired, and the vascular lumen region is extracted from the multidimensional vascular images using image recognition algorithms. A surface reconstruction algorithm is then used to generate a three-dimensional surface mesh model of the vascular lumen region. A Voronoi diagram is constructed within the three-dimensional surface mesh model, and the maximum inscribed sphere radius within the three-dimensional surface mesh model is determined based on the Voronoi diagram. Based on the maximum inscribed sphere radius, the geometric center lines of the main trunk and branch vessels in the three-dimensional surface mesh model are determined, and B-spline curve fitting is performed on the main trunk geometric center line to determine a smooth main trunk curve. Based on the smooth main trunk curve and the branch vessel geometric center lines in the three-dimensional surface mesh model, the target skeleton driving line of the medical device to be assembled is determined. Based on the target skeleton driving line, deformation mapping is performed on the medical device reference mesh model of the medical device to be assembled to obtain the assembled device mesh model, thereby completing the medical device assembly simulation. This method solves the problems of existing TEVAR device simulation schemes, which are difficult to reconcile with the dynamic topological continuity of branch vessels, cannot simultaneously consider real-time computation and solid geometric conservation constraints, and result in significant differences between the simulated device deformation and assembly results and the actual working conditions within diseased blood vessels. The above-described scheme reconstructs a three-dimensional surface mesh model of the vascular lumen based on the patient's multidimensional vascular images. It then uses a Voronoi diagram to solve for the maximum inscribed sphere radius of the vessel, extracts the geometric centerline of the main trunk located at the center of the lumen based on this radius, and obtains a smooth main trunk curve through B-spline fitting. Combined with the branch vessel centerlines, a target skeleton driving line adapted to the vascular morphology is constructed. Based on this target skeleton driving line, the instrument mesh model is deformed and mapped to achieve virtual assembly simulation. This approach can adapt to the individualized vascular anatomy of different patients, ensuring that the extracted vessel centerline is stably located at the center of the vessel lumen, avoiding simulation errors caused by the centerline shifting close to the vessel wall. Furthermore, the smoothed driving skeleton can adapt to complex vascular morphologies including bifurcations, making the mesh deformation of the medical device to be assembled more continuous and smooth during the simulation assembly process. This reduces mesh distortion during simulation, thereby accurately reproducing the assembly morphology of the medical device within the vessel, improving the accuracy and practicality of preoperative virtual assembly simulation for vascular intervention, and providing reliable data support for surgical planning.

[0180] It should be noted that the initial linear skeleton of the medical device to be assembled can be extracted from the medical device reference mesh model. Alternatively, it can be directly obtained from the device design data corresponding to the reference mesh model. The initial linear skeleton consists of multiple skeleton nodes arranged sequentially along the device axis. Let the spatial coordinates of the i-th skeleton node in the initial gripping state be Qi, and the adjacent skeleton nodes... The reference distance between Qi is Li.

[0181] The initial skeleton node Q0 of the initial linear skeleton is mapped to the preset starting position of the target skeleton driving line, resulting in the mapped coordinates P0. Subsequently, following the arrangement order of the skeleton nodes, the mapping positions of the remaining skeleton nodes are determined sequentially along the forward direction of the target skeleton driving line. For the i-th skeleton node, a target position P satisfying the distance constraint between adjacent nodes is searched on the target skeleton driving line. i , so that: The target position satisfying the distance constraint along the direction of travel of the target skeleton drive line is used as the mapped spatial coordinate of the i-th skeleton node. Thus, when the medical device is spatially mapped along the curved target skeleton drive line, the reference distance between adjacent skeleton nodes remains essentially unchanged, thereby reducing unintended stretching or compression of the medical device in the axial direction.

[0182] After determining the mapped spatial coordinates of each skeleton node, a smoothed frame field is used to determine the target local frame at each skeleton node's mapped position. The target local frame includes the tangent vector, normal vector, and binormal vector of the target skeleton driving line at the corresponding position. For the i-th skeleton node, let its initial local frame in the initial gripping state be... Its position P after mapping i The corresponding local frame of the target is F. i Then, the rotation transformation of the skeleton nodes is determined based on the attitude relationship between the initial local frame and the target local frame, and based on the initial skeleton node coordinates Q. i And the mapped skeleton node coordinates P i Determine the translation transformation to generate the rigid body transformation matrix M corresponding to the i-th skeleton node. i .

[0183] For example, the initial local frame and the target local frame can be represented as rigid body transformations, respectively. and The transformation matrix corresponding to the skeleton node is determined by formula (9).

[0184] (9).

[0185] It is understandable that the rigid body transformation matrix includes both the translational component generated by the change of the instrument skeleton nodes from the initial position to the target position, and the rotational component generated by the change of spatial attitude determined by the smooth frame field.

[0186] For example, such as Figure 2 As shown, a medical device assembly simulation system is provided, which includes an operating table 10, a scanning bed 20, and a radiation generator 30. The medical device assembly simulation system is used to implement the aforementioned medical device assembly simulation method.

[0187] The patient is placed on the scanning bed 20, which can be moved by the operating console 10, so that the X-rays emitted by the X-ray generator 30 scan specific parts of the patient, determining a multidimensional vascular image, and returning the generated multidimensional vascular image to the operating console 10. The operating console 10 includes a computer device or server, which can extract the vascular lumen region from the multidimensional vascular image using image recognition algorithms, and generate a three-dimensional surface mesh model of the vascular lumen region using surface reconstruction algorithms; construct a Voronoi diagram within the three-dimensional surface mesh model, and determine the maximum inscribed sphere radius within the three-dimensional surface mesh model based on the Voronoi diagram; determine the main geometric center line of the three-dimensional surface mesh model and the geometric center lines of the branch vessels in the three-dimensional surface mesh model based on the maximum inscribed sphere radius, and perform B-spline curve fitting on the main geometric center line to determine a smooth main curve; determine the target skeleton driving line of the medical device to be assembled based on the smooth main curve and the geometric center lines of the branch vessels in the three-dimensional surface mesh model; and perform deformation mapping on the medical device reference mesh model of the medical device to be assembled based on the target skeleton driving line to obtain the assembled device mesh model, thereby completing the assembly simulation of the medical device.

[0188] For example, based on the above embodiments, the medical device assembly simulation method further includes:

[0189] Multidimensional vascular images of patients requiring medical device assembly for endovascular aortic repair were acquired. Using image recognition algorithms, based on image grayscale features and vascular anatomical topology, the vascular lumen region was extracted from the multidimensional vascular images. A three-dimensional surface mesh model of the vascular lumen region was generated from the image using a moving cube algorithm. A Voronoi diagram was constructed within the three-dimensional surface mesh model, and the maximum inscribed sphere radius within the model was determined based on the Voronoi diagram.

[0190] The reciprocal of the square of the maximum inscribed sphere radius at each point within the 3D surface mesh model is used as a constraint term in the path search cost function. The extraction of the main geometric centerline is transformed into solving the optimal path problem that minimizes the energy integral. After determining the energy integral of the cost function corresponding to each spatial path curve, the minimum integral value is determined from the cost function energy integral, and the spatial path curve corresponding to the minimum integral value is taken as the main geometric centerline. B-spline curve fitting is then performed on the main geometric centerline to obtain a smooth main curve.

[0191] Determine if the medical device to be assembled has a branching structure. If the medical device to be assembled is a straight stent without a branching structure, the smooth main trunk curve is directly used as the target skeleton driving line of the medical device to be assembled. If the medical device to be assembled has a branching structure, based on the three-dimensional Voronoi diagram constructed from the three-dimensional surface mesh model of the blood vessel lumen, determine the maximum inscribed sphere radius within the three-dimensional surface mesh model. The reciprocal of the square of the maximum inscribed sphere radius is used as a constraint term in the path search cost function. The bifurcation port point and the distal end point of the branch vessel are determined from the three-dimensional surface mesh model, and at least one branch vessel path curve is determined based on the bifurcation port point and the distal end point of the branch. Based on the constraint term of the path search cost function and the branch vessel path curve, the energy integral of the cost function corresponding to each branch vessel path curve is determined. According to the energy integral of the cost function corresponding to each branch vessel path curve, the initial centerline of the branch is determined from the branch vessel path curve. B-spline curve fitting is performed on the initial centerline of the branch to generate a smooth geometric centerline of the branch vessel.

[0192] Iterate through all discrete sampling points on the geometric centerline of the branch vessels, and for each discrete sampling point, calculate its nearest projection position on the smooth trunk curve. Obtain the maximum inscribed sphere radius corresponding to the smooth trunk curve at the nearest projection position. This maximum inscribed sphere radius represents the equivalent radius of the trunk vessel's lumen. Compare the spatial distance from the discrete sampling point to the smooth trunk curve with the maximum inscribed sphere radius of the smooth trunk curve at the nearest projection position. If the spatial distance corresponding to the discrete sampling point is less than or equal to the maximum inscribed sphere radius of the smooth trunk curve at the nearest projection position, then the discrete sampling point is determined to be within the lumen coverage area of ​​the trunk vessel. If the spatial distance corresponding to the discrete sampling point is greater than the maximum inscribed sphere radius of the smooth trunk curve at the nearest projection position, then the discrete sampling point is determined to be outside the lumen region of the trunk vessel. Find the state transition point on the geometric centerline of the branch vessels that switches from the lumen coverage area to the outside of the lumen as the boundary position of the branch vessel's geometric centerline leaving the trunk region. The boundary position can be obtained by interpolation between two adjacent discrete sampling points with different states. Determine the branch centerline point set corresponding to the side of the boundary position away from the smooth trunk curve as the branch detachment segment. The process involves two states: discrete sampling points within the lumen coverage area of ​​the main trunk vessel and discrete sampling points detached from the lumen coverage area of ​​the main trunk vessel. The transition point between these two states occurs along the geometric centerline of the branch vessel, moving from the lumen coverage area of ​​the main trunk vessel to the outside of the lumen. Target sampling points are determined from the discrete sampling points corresponding to the branch detachment segment, with a distance less than a distance threshold from the smooth main trunk curve; these target sampling points are discrete sampling points within the branch detachment segment that are close to the smooth main trunk curve. These target sampling points are used as transition fitting points, and Bézier curves are fitted to them to determine the initial Bézier curve. The internal control point of the initial Bézier curve closest to the smooth main trunk curve is used as the branch reference point. The branch reference point is projected onto the smooth main trunk curve, and the projected position of the branch reference point on the smooth main trunk curve is used as the nearest neighbor position, which serves as the central reference for the bifurcation anchor point search. Centered on the nearest neighbor position and based on a preset sampling range, the smooth main trunk curve is truncated, and the truncated curve is used as a candidate main trunk segment, thus limiting the search range for the target bifurcation anchor point. The orthogonal matching cost is determined based on the tangent vector of the smooth trunk curve at the candidate trunk position within the candidate trunk segment, and the unit direction vector of the endpoint of the candidate trunk segment pointing to the branch reference point. The candidate trunk position corresponding to the minimum orthogonal matching cost is selected as the target fork anchor point.

[0193] The target bifurcation anchor point is taken as the first endpoint, and the distal branch point as the second endpoint. The maximum inscribed sphere radius corresponding to the distal branch point is determined, and half of this radius is taken as the first extension length. Based on this first extension length, the distal branch point is extended along the tangent of the branch vessel's geometric center line at the distal branch point, with the extension endpoint taken as the first intermediate control point. The maximum inscribed sphere radius corresponding to the target bifurcation anchor point is determined, and half of this radius is taken as the second extension length. Based on this second extension length, the target bifurcation anchor point is extended along the tangent of the smooth trunk curve at the target bifurcation anchor point, with the extension endpoint taken as the second intermediate control point. A third-order Bézier curve is constructed based on the first endpoint, second endpoint, first intermediate control point, and second intermediate control point. This third-order Bézier curve serves as the transition segment between the smooth trunk curve and the branch vessel's geometric center line, connecting them to determine the target skeleton drive line of the medical device to be assembled.

[0194] A smooth frame field is constructed based on the target skeleton driving line using a rotation minimization frame algorithm. When the medical device to be assembled has no branch structure, the medical device reference mesh model is used as the calibrated mesh model. When the medical device to be assembled has a branch structure, a three-dimensional cross product operation is used to solve for the main trunk tangent vector at the target bifurcation anchor point and the branch tangent vector at the closest position to the target bifurcation anchor point on the branch vessel geometric center line. The normal vector of the vessel bifurcation plane between the main trunk tangent vector at the target bifurcation anchor point and the branch tangent vector at the closest position to the target bifurcation anchor point on the branch vessel geometric center line is determined as the first normal vector. Based on the local spatial reference provided by the smooth frame field, the tangent vectors of the main stent axis and the branch stent axis at the device bifurcation position of the medical device to be assembled are obtained. A three-dimensional cross product operation is performed on the tangent vectors of the main stent axis and the branch stent axis at the device bifurcation position to determine the second normal vector of the device bifurcation plane of the medical device to be assembled. The optimization objective is to minimize the angle between the first and second normal vectors. An objective optimization function is constructed, and the optimal rotation angle is determined based on this function. The corresponding pre-rotation matrix is ​​then determined. This pre-rotation matrix is ​​applied to the coordinates of all vertices of the medical device reference mesh model to be assembled, completing the overall attitude rotation calibration and obtaining the calibrated mesh model.

[0195] The skeleton nodes of the medical device to be assembled are extracted from the medical device reference mesh model. The starting node of the skeleton nodes is mapped onto the target skeleton driving line to determine the first mapping node. Based on the first mapping node and the node geometric constraints, the other nodes of the skeleton nodes except the starting node are mapped onto the target skeleton driving line to determine the coordinates of the nodes after mapping. According to the coordinates of the nodes after mapping, the local spatial reference of each node is obtained by combining the smooth frame field on the target skeleton driving line. The translation and rotation transformations of each skeleton node from the original coordinate system of the device to the spatial coordinate system of the blood vessel are solved in turn, and the rigid body transformation matrix corresponding to each skeleton node is obtained by combining them.

[0196] The process involves: determining the initial spatial coordinates of each grid vertex in the calibrated mesh model; determining the initial linear skeleton based on the medical device reference mesh model, and identifying the skeleton line segment closest to each grid vertex on the initial linear skeleton; projecting each grid vertex vertically onto its corresponding skeleton line segment to obtain the local centroid coordinate parameters of the projection point of each grid vertex on the skeleton line segment; determining the continuous parameterized measure value of each grid vertex on the initial linear skeleton based on the line segment index of each skeleton line segment and the local centroid coordinate parameters of each grid vertex; determining the unnormalized weight corresponding to each grid vertex using the Gaussian probability density function, based on the continuous parameterized measure value and standard deviation parameter; normalizing the unnormalized weight to determine the skin influence weight corresponding to each grid vertex; determining the mapped spatial coordinates of each grid vertex using the linear hybrid skinning algorithm, based on the initial spatial coordinates, rigid body transformation matrix, and the skin influence weight corresponding to each grid vertex; and determining the assembled device mesh model based on the mapped spatial coordinates of each grid vertex, thereby completing the assembly simulation of the medical device.

[0197] The aforementioned medical device assembly simulation method introduces the concept of a general driving skeleton, which can adaptively adjust the mathematical manifold according to the device structure type. It can perform deformation mapping for medical devices such as balloons or straight stents, and is also compatible with complex medical devices with branching structures, effectively improving the algorithm's versatility. By solving for the target bifurcation anchor point through an energy function and constructing a dynamic transition segment using Bézier curves, it overcomes the shortcomings of statically fixed bifurcation nodes in traditional solutions, realistically reproducing the traction slippage and continuous topological deformation of the bifurcation position during the transport of the medical device to be assembled, achieving a simulation effect consistent with kinematic laws. For the large curvature geometry of the aortic arch, a rotation-minimized frame field is used to eliminate the local self-torsion problem of the device caused by abrupt changes in the normal vector. Simultaneously, a two-vector constraint global alignment algorithm is set for branch devices to solve the problems of device attitude uncertainty and insufficient positioning accuracy under complex spatial paths. Combining a linear hybrid skinning algorithm with a distance conservation mechanism between adjacent skeleton nodes, while achieving device mesh deformation rendering, it adheres to the physical characteristic of the stent's axial inextensibility, avoiding model stretching and compression distortion caused by conventional geometric interpolation methods, and ensuring the physical consistency of the medical device assembly simulation.

[0198] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0199] Based on the same inventive concept, this application also provides a medical device assembly simulation apparatus for implementing the medical device assembly simulation method described above. The solution provided by this apparatus is similar to the solution described in the above method; therefore, the specific limitations in one or more embodiments of the medical device assembly simulation apparatus provided below can be found in the limitations of the medical device assembly simulation method described above, and will not be repeated here.

[0200] In one embodiment, such as Figure 3 As shown, a medical device assembly simulation device is provided, including: a three-dimensional mesh model construction module 301, an inscribed sphere radius determination module 302, a smooth principal curve determination module 303, a target skeleton driving line determination module 304, and a device mesh model determination module 305, wherein:

[0201] The three-dimensional mesh model construction module 301 is used to acquire multi-dimensional blood vessel images, extract the blood vessel lumen region from the multi-dimensional blood vessel images through image recognition algorithms, and generate a three-dimensional surface mesh model of the blood vessel lumen region using surface reconstruction algorithms.

[0202] The inscribed sphere radius determination module 302 is used to construct a Voronoi diagram inside the three-dimensional surface mesh model and determine the maximum inscribed sphere radius inside the three-dimensional surface mesh model based on the Voronoi diagram.

[0203] The smooth trunk curve determination module 303 is used to determine the geometric center line of the trunk of the three-dimensional surface mesh model and the geometric center line of the branch vessels in the three-dimensional surface mesh model according to the maximum inscribed sphere radius, and to perform B-spline curve fitting on the geometric center line of the trunk to determine the smooth trunk curve;

[0204] The target skeleton driving line determination module 304 is used to determine the target skeleton driving line of the medical device to be assembled based on the smooth trunk curve and the geometric center line of the branch blood vessels in the three-dimensional surface mesh model.

[0205] The device mesh model determination module 305 is used to perform deformation mapping on the medical device reference mesh model of the medical device to be assembled based on the target skeleton driving line, so as to obtain the assembled device mesh model and thus complete the assembly simulation of the medical device.

[0206] For example, the smoothing main curve determination module 303 is specifically used for:

[0207] The reciprocal of the square of the maximum inscribed sphere radius is used as a constraint term in the path search cost function, and the entrance endpoint and exit endpoint are determined from the three-dimensional surface mesh model. Based on the entrance endpoint and the exit endpoint, at least one spatial path curve is determined.

[0208] Based on the constraints of the path search cost function and the spatial path curves, the energy integral of the cost function corresponding to each spatial path curve is determined.

[0209] The minimum integral value is determined from the energy integral of the cost function, and the spatial path curve corresponding to the minimum integral value is taken as the geometric center line of the main trunk.

[0210] For example, the target skeleton driving line determination module 304 is specifically used for:

[0211] Determine whether the medical device to be assembled has a branch structure;

[0212] If the medical device to be assembled does not have a branch structure, then the smooth main curve is used as the target skeleton driving line of the medical device to be assembled.

[0213] If the medical device to be assembled has a branch structure, the geometric center line of the branch blood vessels is extracted from the three-dimensional surface mesh model;

[0214] Based on the geometric centerline of the branch vessels and the smooth trunk curve, the target bifurcation anchor point is determined;

[0215] Determine the distal branch point of the geometric center line of the branch vessel, and construct a third-order Bézier curve based on the target bifurcation anchor point and the distal branch point;

[0216] Based on the third-order Bezier curve, the geometric center line of the branch vessels, and the smooth trunk curve, the target skeleton drive line of the medical device to be assembled is determined.

[0217] For example, the target skeleton driving line determination module 304 is also specifically used for:

[0218] Determine the spatial distance from each discrete sampling point on the geometric center line of the branch vessels to the smooth trunk curve, and the nearest projection position of the discrete sampling point to the smooth trunk curve;

[0219] Based on the spatial distance and the maximum inscribed sphere radius of the smooth trunk curve at the nearest projection position, determine the branch detachment segment after the branch vessel's geometric center line detaches from the lumen region of the trunk vessel;

[0220] The target sampling point is determined from the discrete sampling points corresponding to the branch break-off segment based on the distance between the discrete sampling points corresponding to the branch break-off segment and the smooth trunk curve.

[0221] Bezier curve fitting is performed on the target sampling points to determine the branch reference points that characterize the extension position of the branch breakaway segment;

[0222] Based on the branch reference points, candidate trunk line segments are determined from the smooth trunk curve;

[0223] Based on the tangent vector of the smooth trunk curve at the candidate trunk position within the candidate trunk segment, and the unit direction vector of the endpoint of the candidate trunk segment pointing to the branch reference point, the orthogonal matching cost is determined, and the target bifurcation anchor point is determined from the candidate trunk positions.

[0224] For example, the target skeleton driving line determination module 304 is also specifically used for:

[0225] The target bifurcation anchor point is taken as the first endpoint, and the distal branch point is taken as the second endpoint;

[0226] Determine the maximum inscribed sphere radius corresponding to the distal branch point, and take half of the maximum inscribed sphere radius corresponding to the distal branch point as the first extension length. Based on the first extension length, extend the distal branch point along the tangent direction of the geometric center line of the branch vessel at the distal branch point, and take the extension endpoint as the first intermediate control point.

[0227] Determine the maximum inscribed sphere radius corresponding to the target bifurcation anchor point, and take half of the maximum inscribed sphere radius corresponding to the target bifurcation anchor point as the second extension length. Based on the second extension length, extend the target bifurcation anchor point along the tangent direction of the smooth trunk curve at the target bifurcation anchor point, and take the extension endpoint as the second intermediate control point.

[0228] A third-order Bézier curve is constructed based on the first endpoint, the second endpoint, the first intermediate control point, and the second intermediate control point.

[0229] For example, the instrument mesh model determination module 305 is specifically used for:

[0230] A smooth frame field is constructed based on the target skeleton driving line using a rotation minimization frame algorithm;

[0231] When the medical device to be assembled has a branch structure, the first normal vector of the blood vessel bifurcation plane is determined by three-dimensional cross product operation between the tangent vector of the smooth trunk curve at the target bifurcation anchor point and the tangent vector at the position closest to the target bifurcation anchor point on the geometric center line of the branch blood vessel.

[0232] Based on the smooth frame field, determine the second normal vector of the device bifurcation plane of the medical device to be assembled;

[0233] With minimizing the angle between the first normal vector and the second normal vector as the optimization objective, a target optimization function is constructed. Based on the target optimization function, the optimal rotation angle is determined, and the pre-rotation matrix corresponding to the optimal rotation angle is determined.

[0234] The pre-rotation matrix is ​​applied to the medical device reference mesh model of the medical device to be assembled to determine the calibrated mesh model;

[0235] When the medical device to be assembled does not have a branch structure, the reference mesh model of the medical device is used as the calibrated mesh model;

[0236] The skeleton nodes of the medical device to be assembled are extracted from the medical device reference mesh model. The starting node of the skeleton nodes is mapped to the target skeleton drive line to determine the first mapping node. Based on the first mapping node and node geometric constraints, the other nodes of the skeleton nodes except the starting node are mapped to the target skeleton drive line to determine the coordinates of the nodes after mapping.

[0237] Based on the mapped coordinates of the nodes and the smoothed frame field, generate the rigid body transformation matrix corresponding to each skeleton node;

[0238] The assembly-state instrument mesh model is determined using a linear hybrid skinning algorithm based on the calibrated mesh model and the rigid body transformation matrix.

[0239] For example, the instrument mesh model determination module 305 is also specifically used for:

[0240] Determine the initial spatial coordinates of each grid vertex in the calibrated grid model;

[0241] Based on the medical device reference mesh model, an initial straight skeleton is determined, and the skeleton line segment closest to each mesh vertex is determined on the initial straight skeleton;

[0242] By projecting each grid vertex vertically onto its corresponding skeleton line segment, the local centroid coordinate parameters of the projection point of each grid vertex on the skeleton line segment are obtained.

[0243] Based on the line segment index of each skeleton line segment and the local centroid coordinate parameters corresponding to each grid vertex, the continuous parameterized measure value of each grid vertex on the initial straight skeleton is determined;

[0244] The unnormalized weights corresponding to each grid vertex are determined using the Gaussian probability density function, based on the continuous parameterized measure value and standard deviation parameter.

[0245] The unnormalized weights are normalized to determine the skinning influence weights corresponding to each grid vertex.

[0246] The linear hybrid skinning algorithm is used to determine the mapped spatial coordinates of each mesh vertex based on the initial spatial coordinates, the rigid body transformation matrix, and the skinning influence weights corresponding to each mesh vertex.

[0247] The assembled device mesh model is determined based on the mapped spatial coordinates of each mesh vertex.

[0248] Each module in the aforementioned medical device assembly simulation device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.

[0249] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 4As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a medical device assembly simulation method. The display unit is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0250] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0251] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0252] Step 1: Obtain a multidimensional blood vessel image, extract the blood vessel lumen region from the multidimensional blood vessel image using an image recognition algorithm, and generate a three-dimensional surface mesh model of the blood vessel lumen region using a surface reconstruction algorithm.

[0253] Step 2: Construct a Voronoi diagram inside the three-dimensional surface mesh model, and determine the maximum inscribed sphere radius within the three-dimensional surface mesh model based on the Voronoi diagram;

[0254] Step 3: Based on the maximum inscribed sphere radius, determine the main geometric center line of the three-dimensional surface mesh model and the branch vessel geometric center lines in the three-dimensional surface mesh model, and perform B-spline curve fitting on the main geometric center line to determine the smooth main curve;

[0255] Step 4: Determine the target skeleton drive line of the medical device to be assembled based on the smooth main curve and the geometric center line of the branch vessels in the three-dimensional surface mesh model;

[0256] Step 5: Based on the target skeleton driving lines, perform deformation mapping on the medical device reference mesh model of the medical device to be assembled to obtain the assembled device mesh model, thereby completing the assembly simulation of the medical device.

[0257] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0258] Step 1: Obtain a multidimensional blood vessel image, extract the blood vessel lumen region from the multidimensional blood vessel image using an image recognition algorithm, and generate a three-dimensional surface mesh model of the blood vessel lumen region using a surface reconstruction algorithm.

[0259] Step 2: Construct a Voronoi diagram inside the three-dimensional surface mesh model, and determine the maximum inscribed sphere radius within the three-dimensional surface mesh model based on the Voronoi diagram;

[0260] Step 3: Based on the maximum inscribed sphere radius, determine the main geometric center line of the three-dimensional surface mesh model and the branch vessel geometric center lines in the three-dimensional surface mesh model, and perform B-spline curve fitting on the main geometric center line to determine the smooth main curve;

[0261] Step 4: Determine the target skeleton drive line of the medical device to be assembled based on the smooth main curve and the geometric center line of the branch vessels in the three-dimensional surface mesh model;

[0262] Step 5: Based on the target skeleton driving lines, perform deformation mapping on the medical device reference mesh model of the medical device to be assembled to obtain the assembled device mesh model, thereby completing the assembly simulation of the medical device.

[0263] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, performs the following steps:

[0264] Step 1: Obtain a multidimensional blood vessel image, extract the blood vessel lumen region from the multidimensional blood vessel image using an image recognition algorithm, and generate a three-dimensional surface mesh model of the blood vessel lumen region using a surface reconstruction algorithm.

[0265] Step 2: Construct a Voronoi diagram inside the three-dimensional surface mesh model, and determine the maximum inscribed sphere radius within the three-dimensional surface mesh model based on the Voronoi diagram;

[0266] Step 3: Based on the maximum inscribed sphere radius, determine the main geometric center line of the three-dimensional surface mesh model and the branch vessel geometric center lines in the three-dimensional surface mesh model, and perform B-spline curve fitting on the main geometric center line to determine the smooth main curve;

[0267] Step 4: Determine the target skeleton drive line of the medical device to be assembled based on the smooth main curve and the geometric center line of the branch vessels in the three-dimensional surface mesh model;

[0268] Step 5: Based on the target skeleton driving lines, perform deformation mapping on the medical device reference mesh model of the medical device to be assembled to obtain the assembled device mesh model, thereby completing the assembly simulation of the medical device.

[0269] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0270] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0271] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0272] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A medical device assembly simulation method, characterized in that, include: A multidimensional blood vessel image is acquired, and the vascular lumen region is extracted from the multidimensional blood vessel image using an image recognition algorithm. A surface reconstruction algorithm is then used to generate a three-dimensional surface mesh model of the vascular lumen region. A Voronoi diagram is constructed inside the three-dimensional surface mesh model, and the maximum inscribed sphere radius within the three-dimensional surface mesh model is determined based on the Voronoi diagram. Based on the maximum inscribed sphere radius, the geometric center line of the main trunk of the three-dimensional surface mesh model and the geometric center line of the branch vessels in the three-dimensional surface mesh model are determined, and the geometric center line of the main trunk is fitted with a B-spline curve to determine a smooth main trunk curve. Based on the smooth main curve and the geometric center lines of the branch vessels in the three-dimensional surface mesh model, the target skeleton drive line of the medical device to be assembled is determined. Based on the target skeleton driving lines, the medical device reference mesh model of the medical device to be assembled is deformed and mapped to obtain the assembled device mesh model, thereby completing the assembly simulation of the medical device.

2. The method according to claim 1, characterized in that, Based on the maximum inscribed sphere radius, the main geometric centerline of the three-dimensional surface mesh model is determined, including: The reciprocal of the square of the maximum inscribed sphere radius is used as a constraint term in the path search cost function, and the entrance endpoint and exit endpoint are determined from the three-dimensional surface mesh model. Based on the entrance endpoint and the exit endpoint, at least one spatial path curve is determined. Based on the constraints of the path search cost function and the spatial path curves, the energy integral of the cost function corresponding to each spatial path curve is determined. The minimum integral value is determined from the energy integral of the cost function, and the spatial path curve corresponding to the minimum integral value is taken as the geometric center line of the main trunk.

3. The method according to claim 1, characterized in that, Based on the smooth main trunk curve and the geometric center lines of the branch vessels in the three-dimensional surface mesh model, the target skeleton drive lines of the medical device to be assembled are determined, including: Determine whether the medical device to be assembled has a branch structure; If the medical device to be assembled does not have a branch structure, then the smooth main curve is used as the target skeleton driving line of the medical device to be assembled. If the medical device to be assembled has a branch structure, the target bifurcation anchor point is determined based on the geometric center line of the branch blood vessel and the smooth trunk curve. Determine the distal branch point of the geometric center line of the branch vessel, and construct a third-order Bézier curve based on the target bifurcation anchor point and the distal branch point; Based on the third-order Bezier curve, the geometric center line of the branch vessels, and the smooth trunk curve, the target skeleton drive line of the medical device to be assembled is determined.

4. The method according to claim 3, characterized in that, Based on the geometric centerline of the branch vessels and the smooth trunk curve, the target bifurcation anchor point is determined, including: Determine the spatial distance from each discrete sampling point on the geometric center line of the branch vessels to the smooth trunk curve, and the nearest projection position of the discrete sampling point to the smooth trunk curve; Based on the spatial distance and the maximum inscribed sphere radius of the smooth trunk curve at the nearest projection position, determine the branch detachment segment after the branch vessel's geometric center line detaches from the lumen region of the trunk vessel; The target sampling point is determined from the discrete sampling points corresponding to the branch break-off segment based on the distance between the discrete sampling points corresponding to the branch break-off segment and the smooth trunk curve. Bezier curve fitting is performed on the target sampling points to determine the branch reference points that characterize the extension position of the branch breakaway segment; Based on the branch reference points, candidate trunk line segments are determined from the smooth trunk curve; Based on the tangent vector of the smooth trunk curve at the candidate trunk position within the candidate trunk segment, and the unit direction vector of the endpoint of the candidate trunk segment pointing to the branch reference point, the orthogonal matching cost is determined, and the target bifurcation anchor point is determined from the candidate trunk positions.

5. The method according to claim 3, characterized in that, Determine the distal branch point of the geometric center line of the branch vessel, and construct a third-order Bézier curve based on the target bifurcation anchor point and the distal branch point, including: The target bifurcation anchor point is taken as the first endpoint, and the distal branch point is taken as the second endpoint; Determine the maximum inscribed sphere radius corresponding to the distal branch point, and take half of the maximum inscribed sphere radius corresponding to the distal branch point as the first extension length. Based on the first extension length, extend the distal branch point along the tangent direction of the geometric center line of the branch vessel at the distal branch point, and take the extension endpoint as the first intermediate control point. Determine the maximum inscribed sphere radius corresponding to the target bifurcation anchor point, and take half of the maximum inscribed sphere radius corresponding to the target bifurcation anchor point as the second extension length. Based on the second extension length, extend the target bifurcation anchor point along the tangent direction of the smooth trunk curve at the target bifurcation anchor point, and take the extension endpoint as the second intermediate control point. A third-order Bézier curve is constructed based on the first endpoint, the second endpoint, the first intermediate control point, and the second intermediate control point.

6. The method according to claim 3, characterized in that, Based on the target skeleton driving lines, deformation mapping is performed on the medical device reference mesh model of the medical device to be assembled to obtain the assembled device mesh model, including: A smooth frame field is constructed based on the target skeleton driving line using a rotation minimization frame algorithm; When the medical device to be assembled has a branch structure, the first normal vector of the blood vessel bifurcation plane is determined by three-dimensional cross product operation between the tangent vector of the smooth trunk curve at the target bifurcation anchor point and the tangent vector at the position closest to the target bifurcation anchor point on the geometric center line of the branch blood vessel. Based on the smooth frame field, determine the second normal vector of the device bifurcation plane of the medical device to be assembled; With minimizing the angle between the first normal vector and the second normal vector as the optimization objective, a target optimization function is constructed. Based on the target optimization function, the optimal rotation angle is determined, and the pre-rotation matrix corresponding to the optimal rotation angle is determined. The pre-rotation matrix is ​​applied to the medical device reference mesh model of the medical device to be assembled to determine the calibrated mesh model; When the medical device to be assembled does not have a branch structure, the reference mesh model of the medical device is used as the calibrated mesh model; The skeleton nodes of the medical device to be assembled are extracted from the medical device reference mesh model. The starting node of the skeleton nodes is mapped to the target skeleton drive line to determine the first mapping node. Based on the first mapping node and node geometric constraints, the other nodes of the skeleton nodes except the starting node are mapped to the target skeleton drive line to determine the coordinates of the nodes after mapping. Based on the mapped coordinates of the nodes and the smoothed frame field, generate the rigid body transformation matrix corresponding to each skeleton node; The assembly-state instrument mesh model is determined using a linear hybrid skinning algorithm based on the calibrated mesh model and the rigid body transformation matrix.

7. The method according to claim 6, characterized in that, The assembly-state device mesh model is determined using a linear hybrid skinning algorithm based on the calibrated mesh model and the rigid body transformation matrix, including: Determine the initial spatial coordinates of each grid vertex in the calibrated grid model; Based on the medical device reference mesh model, an initial straight skeleton is determined, and the skeleton line segment closest to each mesh vertex is determined on the initial straight skeleton; By projecting each grid vertex vertically onto its corresponding skeleton line segment, the local centroid coordinate parameters of the projection point of each grid vertex on the skeleton line segment are obtained. Based on the line segment index of each skeleton line segment and the local centroid coordinate parameters corresponding to each grid vertex, the continuous parameterized measure value of each grid vertex on the initial straight skeleton is determined; The unnormalized weights corresponding to each grid vertex are determined using the Gaussian probability density function, based on the continuous parameterized measure value and standard deviation parameter. The unnormalized weights are normalized to determine the skinning influence weights corresponding to each grid vertex. The linear hybrid skinning algorithm is used to determine the mapped spatial coordinates of each mesh vertex based on the initial spatial coordinates, the rigid body transformation matrix, and the skinning influence weights corresponding to each mesh vertex. The assembled device mesh model is determined based on the mapped spatial coordinates of each mesh vertex.

8. A medical device assembly simulation device, characterized in that, The medical device assembly simulation device includes: The three-dimensional mesh model construction module is used to acquire multi-dimensional blood vessel images, extract the vascular lumen region from the multi-dimensional blood vessel images through image recognition algorithms, and generate a three-dimensional surface mesh model of the vascular lumen region using a surface reconstruction algorithm. The inscribed sphere radius determination module is used to construct a Voronoi diagram inside the three-dimensional surface mesh model and determine the maximum inscribed sphere radius within the three-dimensional surface mesh model based on the Voronoi diagram. The smooth trunk curve determination module is used to determine the geometric center line of the trunk and the geometric center line of the branch vessels in the three-dimensional surface mesh model based on the maximum inscribed sphere radius, and to perform B-spline curve fitting on the geometric center line of the trunk to determine the smooth trunk curve. The target skeleton driving line determination module is used to determine the target skeleton driving line of the medical device to be assembled based on the smooth trunk curve and the geometric center line of the branch blood vessels in the three-dimensional surface mesh model. The device mesh model determination module is used to perform deformation mapping on the medical device reference mesh model of the medical device to be assembled based on the target skeleton driving lines, so as to obtain the assembled device mesh model and thus complete the assembly simulation of the medical device.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images