Real-time virtual implantation method and device for stent-assisted coils

Abstract

The present invention discloses a real-time virtual implantation device for stent-assisted coils, which includes acquiring a three-dimensional vascular image sequence and constructing a diseased vascular model containing an aneurysm and the parent artery; extracting parent artery parameters according to the diseased vascular model and recommending stent specifications, and generating point cloud data for the stent based on the stent specifications, parent artery parameters, and a pre-designed stent cross-section; acquiring coil parameters, determining the starting point of the coil based on the aneurysm cavity, iteratively generating a set of candidate points for each time step of the coil based on the starting point and the coil parameters, and screening the candidate points according to the energy of each candidate point to determine the unique confirmed point for each time step. The unique confirmed points of all time steps constitute the point cloud data of the coil; generating a stent model based on the point cloud data of the stent and generating a coil model based on the point cloud data of the coil, which can improve the real-time virtual implantation effect of stent-assisted coils and reduce computational consumption.

Description

Technical Field

[0001] The present invention belongs to the technical field of medical simulation, and particularly relates to a real-time virtual implantation method and device for a stent-assisted coil. Background Art

[0002] Currently, the main interventional treatment method for medium and small aneurysms, especially ruptured aneurysms, is to embolize the aneurysm cavity with metal coils, thereby reducing the impact of blood flow on the aneurysm wall, triggering thrombosis in the aneurysm cavity, and ultimately achieving the effect of closing the aneurysm cavity. For wide-neck aneurysms, in order to prevent the herniation of the coils, a stent is usually used to assist the embolization of the coils.

[0003] To analyze the postoperative hemodynamic results and optimize the surgical plan, researchers usually use finite element simulation technology to simulate the process of stent deployment in the parent artery and the process of coil delivery from the microcatheter to the aneurysm cavity, obtain simulation results with different embolization densities and different stents, and then use computational fluid dynamics methods to evaluate the hemodynamic results of different plans, such as blood flow streamlines, velocity isosurfaces, wall shear stress, shear oscillation index, etc.

[0004] However, the disadvantage of the finite element method is that it requires a long calculation time, ranging from several hours to several days depending on the complexity of the model. Moreover, due to overly complex non-linear contacts, the calculation results are very likely to diverge.

[0005] Therefore, the research and development of a real-time and highly robust virtual implantation technology for stent-assisted coils is of great significance for optimizing intracranial interventional surgical plans. Summary of the Invention

[0006] In view of the above, the object of the present invention is to provide a real-time virtual implantation method and device for a stent-assisted coil, which can improve the real-time virtual implantation effect of the stent-assisted coil and reduce the calculation consumption.

[0007] To achieve the above object of the invention, a real-time virtual implantation device for a stent-assisted coil provided by an embodiment includes a memory, a processor, and a computer program stored in the memory and executable on the processor. The feature is that when the processor executes the computer program, the following steps are implemented:

[0008] Image processing: Obtain a three-dimensional vascular image sequence and construct a diseased vascular model including an aneurysm and a parent artery.

[0009] Virtual implantation of the stent: Extract the parameters of the parent artery according to the diseased vascular model, recommend the stent specifications for supporting the parent artery based on the parameters of the parent artery, and generate point cloud data for the stent according to the stent specifications, the parameters of the parent artery, and a pre-designed stent cross-section.

[0010] Separation of the aneurysm cavity: Separate the aneurysm cavity according to the diseased blood vessel model and reconstruct the aneurysm cavity model.

[0011] Virtual embolization of coils: Obtain the coil parameters, determine the starting point of the coil based on the aneurysm cavity model, iterate to generate a set of candidate points for each time step of the coil based on the starting point and the coil parameters, and screen the candidate points according to the energy of each candidate point to determine the unique confirmed point for each time step. The unique confirmed points for all time steps constitute the point cloud data of the coil.

[0012] Model generation: Generate a stent model based on the point cloud data of the stent and generate a coil model based on the point cloud data of the coil.

[0013] Preferably, extracting the parameters of the parent artery according to the diseased blood vessel model includes:

[0014] Confirm the proximal entrance of the parent artery according to the diseased blood vessel model, and starting from the proximal entrance, generate a centerline, a sequence of radii along the line, and a sequence of tangent vectors along the line to all distal exits of the parent artery. Among them, the centerline, the sequence of radii along the line, and the sequence of tangent vectors along the line are collectively referred to as the parameters of the parent artery.

[0015] Preferably, recommending the specification parameters of the stent for supporting the parent artery based on the parameters of the parent artery includes:

[0016] Select two points on the centerline of the parent artery at the expected proximal and distal positions of the stent, and recommend the stent specifications based on these two points. Among them, the stent specifications include the stent length specification and the stent diameter specification.

[0017] For the stent length specification, calculate the length of the centerline between the two points, select the length specification in the stent length specification set that is closest to this length as the recommended stent length specification. At the same time, keep the distal position of the stent and update the proximal position of the stent according to the recommended stent length specification so that the length of the centerline between the distal and proximal ends of the stent is equal to the recommended stent length specification.

[0018] For the stent diameter specification, compare the radii at the two points and select twice the larger radius as the reference value, or calculate the average value of the radius sequence between the two points and use twice the average value as the reference value. Then select the diameter specification in the stent diameter specification set that is closest to the reference value and greater than the reference value as the recommended stent diameter specification.

[0019] Preferably, the stent specification includes the stent length specification. Generating the point cloud data for the stent according to the stent specification, the parameters of the parent artery, and the pre-designed stent cross-section includes:

[0020] The pre-designed stent cross-section has periodically designed intersection points. The stent is discretized at equal intervals in the length direction based on the intersection points and resampled at equal intervals on the center line.

[0021] At each resampling point position, a circle representing the stent cross-section is generated according to the radius along the line and the tangent vector along the line, and the circle is discretized by angle to obtain N points along the circular cross-section contour. N is equal to the number of intersection points on the stent cross-section. The points on each circular cross-section contour are sorted according to the helical structure of the stent, and the obtained point sequence is used as the point cloud data of the stent.

[0022] Preferably, the obtaining of the coil parameters includes:

[0023] Ignoring the primary helical structure of the coil, the coil is equivalent to a circular cross-section filament with only a secondary helical structure and a radius of r, where r is the primary helical radius. The coil is discretized in length to obtain n line elements with a length of l, and at the same time, n + 1 points are obtained. The secondary helical radius R of the coil is equal to the initial curvature radius of the coil. According to the initial curvature radius, the initial rotation angle a between the line elements is obtained as a = l / R.

[0024] Preferably, the determining of the starting point of the coil based on the aneurysm cavity model includes:

[0025] Determine the geometric center of the aneurysm cavity based on the aneurysm cavity model, and randomly select a point on the spherical surface with the geometric center as the center of the sphere as the starting point;

[0026] Or, randomly generate a point located within the aneurysm cavity in the bounding box of the aneurysm as the starting point;

[0027] Or, obtain the starting point determined by an interactive method.

[0028] Preferably, the iteratively generating the candidate point set for each time step of the coil based on the starting point and the coil parameters includes:

[0029] Take the starting point as the only confirmed point in the first time step, randomly generate a confirmation direction for the starting point, and randomly generate a confirmation direction and the only confirmed point in the second time step on the conical surface with the starting point as the vertex and the starting point confirmation direction as the rotation axis;

[0030] For the third time step and above starting from the initial point, first, an initial candidate direction is generated for the current time step, including: determining the rotation axis based on the confirmed directions of the confirmed points in the previous two time steps starting from the current time step, calculating the rotation matrix according to the rotation axis and the initial rotation angle included in the spring coil parameters, and multiplying the rotation matrix by the confirmed direction of the confirmed point in the previous time step starting from the current time step to obtain the initial candidate direction of the current time step; then, a candidate point set is generated based on the initial candidate direction, including: generating candidate points on the solid azimuth angle centered on the initial candidate direction to form a candidate point set, and the selection range of the solid azimuth angle is (0, 2π).

[0031] Preferably, screening the candidate points based on the energy of each candidate point to determine the unique confirmed point for each time step includes:

[0032] Calculating the total energy of each candidate point, and this total energy includes the energy Φ caused by the spatial relationship between the candidate point and the aneurysm cavity sac and the energy Φ caused by the spatial relationship between the candidate point and the known confirmed point col and the energy Φ caused by the spring coil bending deformation due to the candidate direction formed by the candidate point and the previous confirmed point ben ;

[0033] Screening the candidate point set based on the total energy of each candidate point includes: setting the candidate points with Φ sac greater than zero as invalid candidate points, and selecting the candidate point with the lowest total energy as the unique confirmed point from the remaining candidate points. When there are multiple candidate points with the lowest total energy, randomly select one candidate point as the unique confirmed point: when Φ sac of all candidate points in the candidate point set are greater than zero, then delete the candidate point set, roll back to the candidate point set where the unique confirmed point of the previous time step is located, and set the unique confirmed point of the previous time step as an invalid candidate point. At the same time, select the candidate point with the lowest total energy in the candidate point set where the invalid candidate point is located as the new unique confirmed point of the previous time step, and continue to regenerate the candidate point set for the current time step,

[0034] While determining the unique confirmed point of the current time step, the confirmed direction between the unique confirmed point of the current time step and the unique confirmed point of the previous time step is also determined.

[0035] Among them, for the energy Φ sac caused by the spatial relationship between the candidate point and the aneurysm cavity, when the candidate point is located inside the aneurysm cavity and the distance from the cavity is greater than the first helical radius of the spring coil, the energy Φ sac is zero, otherwise the energy Φ sac is greater than zero;

[0036] For the energy Φ caused by the spatial relationship between the candidate point and the known confirmed point col When the distance between the candidate point and the confirmed point is greater than or equal to twice the first helical radius of the coil, then Φ col The energy is zero, otherwise the energy Φ col is greater than zero;

[0037] For the energy Φ caused by the bending deformation of the coil due to the candidate direction formed by the candidate point and the previous confirmed point ben When the candidate direction is the initial candidate direction, the energy Φ ben is zero, otherwise calculate the angle α between the candidate direction and the initial candidate direction, and the energy Φ ben is positively correlated with the angle: Φ ben = Αα Β where A and B are parameters greater than zero.

[0038] To achieve the above invention purpose, the embodiment also provides a real-time virtual implantation device for a stent-assisted coil, including an image processing module, a virtual implantation module for the stent, a tumor cavity separation module, a virtual embolization module for the coil, and a model generation module.

[0039] The image processing module is used to obtain a three-dimensional vascular image sequence and construct a diseased vascular model including an aneurysm and a parent artery.

[0040] The virtual implantation module for the stent is used to extract the parent artery parameters according to the diseased vascular model, recommend the stent specifications for supporting the parent artery based on the parent artery parameters, and generate point cloud data for the stent according to the stent specifications, the parent artery parameters, and the pre-designed stent cross-section.

[0041] The tumor cavity separation module is used to separate the tumor cavity according to the diseased vascular model and reconstruct the aneurysm tumor cavity model.

[0042] The virtual embolization module for the coil is used to obtain the coil parameters, determine the starting point of the coil according to the aneurysm tumor cavity model, iteratively generate a set of candidate points for each time step of the coil based on the starting point and the coil parameters, and screen the candidate points according to the energy of each candidate point to determine the unique confirmed point for each time step. The unique confirmed points for all time steps constitute the point cloud data of the coil.

[0043] The model generation module is used to generate a stent model based on the point cloud data of the stent and generate a coil model based on the point cloud data of the coil.

[0044] To achieve the above invention purpose, the embodiment also provides a real-time virtual implantation method for a stent-assisted coil, including the following steps:

[0045] Image processing: Obtain a three-dimensional vascular image sequence and construct a diseased blood vessel model containing an aneurysm and the parent artery.

[0046] Virtual implantation of the stent: Extract the parameters of the parent artery according to the diseased blood vessel model, recommend the stent specifications for supporting the parent artery based on the parameters of the parent artery, and generate point cloud data for the stent according to the stent specifications, the parameters of the parent artery, and the pre-designed stent cross-section.

[0047] Separation of the aneurysm cavity: Separate the aneurysm cavity according to the diseased blood vessel model and reconstruct the aneurysm cavity model.

[0048] Virtual embolization of the coil: Obtain the coil parameters, determine the starting point of the coil according to the aneurysm cavity model, iteratively generate a set of candidate points for each time step of the coil based on the starting point and the coil parameters, and screen the candidate points according to the energy of each candidate point to determine the unique confirmed point for each time step. The unique confirmed points of all time steps constitute the point cloud data of the coil.

[0049] Model generation: Generate a stent model based on the point cloud data of the stent and generate a coil model based on the point cloud data of the coil.

[0050] Compared with the prior art, the beneficial effects of the present invention at least include:

[0051] During the virtual implantation of the stent, recommend the stent specifications based on the parameters of the parent artery extracted from the diseased blood vessel model, and generate point cloud data for the stent according to the stent specifications, the parameters of the parent artery, and the pre-designed stent cross-section; during the virtual embolization of the coil, iteratively generate a set of candidate points for each time step of the coil based on the starting point and the coil parameters, and screen the candidate points according to the energy of each candidate point to determine the unique confirmed point for each time step. The unique confirmed points of all time steps constitute the point cloud data of the coil; the computational consumption of these two processes is low, the computational efficiency is high, and there is no problem of non-convergence, which greatly improves the real-time virtual implantation effect of the stent-assisted coil. Description of the drawings

[0052] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0053] Figure 1 It is a flowchart of the real-time virtual implantation method of the stent-assisted coil provided by the embodiment:

[0054] Figure 2Schematic diagram of resampling of the discrete points of the stent and the center line provided by the embodiment;

[0055] Figure 3 Schematic diagram of angular discretization of the stent cross-section provided by the embodiment;

[0056] Figure 4 Schematic diagram of generation of the candidate point set provided by the embodiment;

[0057] Figure 5 Schematic diagram of confirmation of the unique confirmation point provided by the embodiment;

[0058] Figure 6 Schematic diagram of fallback of the unique confirmation point provided by the embodiment;

[0059] Figure 7 Effect diagram of the coil in the blood vessel provided by the embodiment;

[0060] Figure 8 Schematic structural diagram of the real-time virtual implantation device of the stent-assisted coil provided by the embodiment. Detailed implementation manners

[0061] To make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific implementation manners described herein are only used to explain the present invention and do not limit the protection scope of the present invention.

[0062] When using finite element simulation to simulate the stent-assisted coil embolization process, a complex preprocessing process is required, and there are disadvantages such as large real-time calculation consumption, long time, and difficult convergence, which limit the application of hemodynamics in optimizing the surgical plan. To solve this technical problem, the real-time virtual implantation method and device of the stent-assisted coil provided by the embodiments of the present invention can achieve the real-time virtual implantation of the stent-assisted coil with low human and material costs and low calculation consumption, improve the real-time virtual implantation effect, and further improve the application value of subsequent hemodynamics in clinical practice.

[0063] The embodiment provides a real-time virtual implantation method for a stent-assisted coil, as Figure 1 shown, including the following steps:

[0064] S110, Image processing.

[0065] In the embodiment, the image processing includes: obtaining a three-dimensional blood vessel image sequence and constructing a diseased blood vessel model including an aneurysm and a parent artery. Among them, the three-dimensional blood vessel image sequence is a sequence arranged in the order of acquisition time of three-dimensional blood vessel images, including but not limited to three-dimensional image sequences of digital subtraction angiography (DSA), CT angiography (CTA), and magnetic resonance angiography (MRA).

[0066] After obtaining the three-dimensional vascular image sequence, after segmenting the three-dimensional vascular image sequence using the threshold method, the level set method, or an artificial intelligence segmentation model (such as 3D Unet), the marching cubes algorithm is used to perform surface reconstruction on the segmentation result to obtain a vascular model. Then, the region of interest is extracted from the vascular model, the aneurysm and the part of the parent artery are retained, and the remaining vascular branches are deleted to obtain a diseased vascular model containing the aneurysm and the parent artery.

[0067] S120, virtual implantation of the stent.

[0068] In the embodiment, the virtual implantation of the stent includes: extracting the parameters of the parent artery according to the diseased vascular model, and recommending the stent specifications for supporting the parent artery based on the parameters of the parent artery. According to the stent specifications, the parameters of the parent artery, and the pre-designed stent cross-section, point cloud data is generated for the stent.

[0069] In a possible implementation manner, extracting the parameters of the parent artery according to the diseased vascular model includes: confirming the proximal entrance of the parent artery in an interactive manner according to the diseased vascular model, and starting from the proximal entrance, using the centerline generation algorithm of the third-party library vmtk and other methods to generate the centerline, the sequence of radii along the line, and the sequence of tangent vectors along the line from the parent artery to all distal exits. Among them, the centerline, the sequence of radii along the line, and the sequence of tangent vectors along the line are collectively referred to as the parameters of the parent artery.

[0070] In a possible implementation manner, recommending the stent specifications for supporting the parent artery based on the parameters of the parent artery includes:

[0071] On the centerline of the parent artery, two points are selected at the expected proximal and distal positions of the stent, and the stent specifications are recommended based on these two points. Among them, the stent specifications include the stent length specification and the stent diameter specification.

[0072] For the stent length specification, calculate the length of the centerline between the two points, and select the length specification in the stent length specification set that is closest to this length as the recommended stent length specification. At the same time, keeping the distal position of the stent, update the proximal position of the stent according to the recommended stent length specification, so that the length of the centerline between the distal and proximal ends of the stent is equal to the recommended stent length specification.

[0073] For the stent diameter specification, two methods are used to confirm the reference value, that is, including comparing the radii at the two points and selecting 2 times the larger radius as the reference value, or calculating the average value of the radius sequence between the two points and using 2 times the average value as the reference value. Then, after obtaining the reference value, select the diameter specification in the stent diameter specification set that is closest to the reference value and greater than the reference value as the recommended stent diameter specification.

[0074] In a possible implementation, point cloud data is generated for the stent based on the stent specifications, the parameters of the parent artery with the aneurysm, and the pre-designed stent cross-section, including:

[0075] First, the pre-designed stent cross-section has periodically designed intersection points. The stent is discretized at equal intervals in the length direction according to the intersection points and resampled at equal intervals on the center line, as Figure 2 shown.

[0076] Then, at each resampled point position, a circle representing the stent cross-section is generated according to the radius along the line and the tangent vector along the line (such as the dashed circle in Figure 3 ), and the circle is discretized by angle to obtain N points along the contour of the circular cross-section (such as the black dots on the dashed circle in Figure 3 ). N is equal to the number of intersection points on the stent cross-section. The points on each circular cross-section contour are sorted according to the helical structure of the stent, which can realize the virtual braiding of the stent. The obtained point sequence is used as the point cloud data of the stent, which is convenient for the subsequent generation and visualization of the stent model.

[0077] S130, aneurysm cavity separation.

[0078] In the embodiment, the aneurysm cavity separation includes: separating the aneurysm cavity according to the diseased blood vessel model and reconstructing the aneurysm cavity model. In one implementation, a sequence of binary images of the diseased blood vessel is generated using the diseased blood vessel model. A healthy blood vessel model without an aneurysm is generated using the sequence of radii along the parent artery, and a sequence of binary images of the healthy blood vessel is generated using the healthy blood vessel model. The sequence of binary images of the diseased blood vessel is subtracted from the sequence of binary images of the healthy blood vessel to obtain a sequence of binary images of the aneurysm cavity. The marching cubes algorithm is used to perform surface reconstruction on the sequence of binary images of the aneurysm cavity to obtain the aneurysm cavity model.

[0079] S140, virtual embolization of the coil.

[0080] Each coil can be regarded as being swept by a series of ordered points with the same radius. The virtual embolization process of the coil is essentially a process of probing, adjusting, and confirming a series of ordered points.

[0081] In the embodiment, the virtual embolization of the coil includes: obtaining the coil parameters, determining the starting point of the coil according to the aneurysm cavity model, iteratively generating a set of candidate points for each time step of the coil based on the starting point and the coil parameters, and screening the candidate points according to the energy of each candidate point to determine the unique confirmed point for each time step. The unique confirmed points for all time steps constitute the point cloud data of the coil.

[0082] In one embodiment, obtaining the coil parameters includes: ignoring the primary helical structure of the coil, and equivalenting the coil to a circular cross-section filament with only a secondary helical structure and a radius of r, where r is the primary helical radius. Discretize the length of the coil to obtain n line elements with a length of l, and at the same time obtain n + 1 points. The secondary helical radius R of the coil is equal to the initial curvature radius of the coil, and the initial rotation angle a between the line elements is obtained according to the initial curvature radius, a = l / R.

[0083] In one embodiment, determining the starting point of the coil based on the aneurysm cavity model includes: (a) determining the geometric center of the aneurysm cavity according to the aneurysm cavity model, and randomly selecting a point on the spherical surface with this geometric center as the center of the sphere as the starting point. Among them, the spherical radius can be determined by the aneurysm size, such as one-fourth of the equivalent radius of the aneurysm, or can be determined by the microcatheter radius, such as 5 times the microcatheter radius.

[0084] (b) Alternatively, randomly generate a point located within the aneurysm cavity in the bounding box of the aneurysm. Specifically, directly randomly generate a three-dimensional coordinate in the bounding box of the aneurysm, and further determine whether the randomly generated point is located within the aneurysm cavity. If it is not within the cavity, regenerate it until the randomly generated point is located within the aneurysm cavity, and then this point is the starting point.

[0085] (c) Alternatively, obtain the starting point determined by an interactive method. Specifically, it is manually determined by the user's graphical interface through an interactive method.

[0086] In the embodiment, based on the starting point and the coil parameters, iterate to generate a set of candidate points for each time step of the coil, and screen the candidate points according to the energy of each candidate point to determine the unique confirmed point for each time step. This process is an iterative loop process, that is, the unique confirmed point of the current time step is determined according to the unique confirmed point of the historical time step.

[0087] In one possible embodiment, iterating based on the starting point and the coil parameters to generate a set of candidate points for each time step of the coil includes:

[0088] Take the starting point as the unique confirmed point of the first time step, randomly generate a confirmation direction for the starting point, and randomly generate the confirmation direction and the unique confirmed point of the second time step on the conical surface with the starting point as the vertex and the starting point confirmation direction as the rotation axis;

[0089] For the third time step and above starting from the initial point, first, generate an initial candidate direction for the current time step, including: determining the rotation axis based on the confirmed directions of the confirmed points in the first two time steps starting from the current time step, calculating the rotation matrix according to the rotation axis and the initial rotation angle included in the spring coil parameters, and multiplying the rotation matrix by the confirmed direction of the confirmed point in the previous time step starting from the current time step to obtain the initial candidate direction for the current time step; then, generate a set of candidate points based on the initial candidate direction, including: generating candidate points on the solid azimuth angle centered on the initial candidate direction to form a set of candidate points, and the selection range of the solid azimuth angle is (0, 2π), as Figure 4 shown.

[0090] In a possible implementation, screen the candidate points according to the energy of each candidate point to determine the unique confirmed point for each time step, including:

[0091] First, calculate the total energy of each candidate point, and the total energy includes the energy Φ sac caused by the spatial relationship between the candidate point and the aneurysm cavity, the energy Φ col caused by the spatial relationship between the candidate point and the known confirmed point, and the energy Φ ben of the spring coil bending deformation caused by the candidate direction formed by the candidate point and the previous confirmed point. It is expressed by the formula: Φ tot =Φ sac +Φ col +Φ ben ;

[0092] Among them, for the energy Φ sac caused by the spatial relationship between the candidate point and the aneurysm cavity, when the candidate point is located inside the aneurysm cavity and the distance from the cavity is greater than the first helical radius of the spring coil, the energy Φ sac is zero, otherwise the energy Φ sac is greater than zero;

[0093] For the energy Φ col caused by the spatial relationship between the candidate point and the known confirmed point, when the distance between the candidate point and the confirmed point is greater than or equal to twice the first helical radius of the spring coil, the Φ col energy is zero, otherwise the energy Φ col is greater than zero;

[0094] For the energy Φ ben of the spring coil bending deformation caused by the candidate direction formed by the candidate point and the previous confirmed point, when the candidate direction is the initial candidate direction, the energy Φ ben is zero, otherwise calculate the included angle α between the candidate direction and the initial candidate direction, and the magnitude of the energy Φ ben is positively correlated with the included angle: Φ ben =ΑαΒ where A and B are parameters greater than zero, the value range of A is from 0 to 1, and the value range of B is from 1 to 2.

[0095] Then, the candidate point set is screened according to the total energy of each candidate point, including: setting the candidate points with Φ sac greater than zero as invalid candidate points, as Figure 5 shown, and selecting the candidate point with the lowest total energy from the remaining candidate points as the only confirmed point. When there are multiple candidate points with the lowest total energy, randomly select one candidate point as the only confirmed point, and add the determined only confirmed point to the sequence of confirmed points: as Figure 6 shown, when Φ sac of all candidate points in the candidate point set are greater than zero, then delete the candidate point set, roll back to the candidate point set where the only confirmed point of the previous time step is located, and set the only confirmed point of the previous time step as an invalid candidate point. At the same time, select the candidate point with the lowest total energy in the candidate point set where the invalid candidate point is located as the new only confirmed point of the previous time step, and continue to generate a candidate point set for the current time step. While determining the only confirmed point of the current time step, the confirmation direction between the only confirmed point of the current time step and the only confirmed point of the previous time step is also determined.

[0096] It should be noted that in the case where all candidate points in the candidate point set where the confirmed point is set as an invalid candidate point are invalid candidate points, further rollback is performed until candidate points that meet the requirements are obtained.

[0097] After determining the only confirmed points of all time steps, the only confirmed points of all time steps constitute the point cloud data of the spring coil, and this point cloud data is used for the subsequent generation and visualization of the spring coil model.

[0098] S150, model generation.

[0099] In the embodiment, the model generation includes generating a stent model based on the point cloud data of the stent and generating a spring coil model based on the point cloud data of the spring coil, specifically including: generating a cylinder representing the stent according to the point cloud data sequence of the stent and the stent diameter, and performing a closing process on both ends of each stent. Generating a cylinder representing the spring coil according to the point cloud data sequence of the point spring coil and the spring coil radius, then performing a closing process on both ends of the stent, and finally exporting the stent and the spring coil in stl format. The effects of the stent and the spring coil in the blood vessel are as Figure 7 shown.

[0100] The real-time virtual implantation method of the stent-assisted coil provided by the above embodiments greatly improves the calculation efficiency, has no convergence problem, and the generated coil result avoids intrusion with the aneurysm wall. At the same time, the obtained stent model and coil model can not only be used for real-time surgical planning in clinical practice, but also be used to evaluate the blood flow changes after stent-assisted coil embolization, optimize the surgical plan, and thus improve the surgical effect.

[0101] The embodiment also provides a real-time virtual implantation device for stent-assisted coils, including a memory, a processor, and a computer program stored in the memory and executable on the processor. Wherein, when the processor executes the computer program, the real-time virtual implantation method of the stent-assisted coil is implemented, specifically including the following steps:

[0102] Step 1, image processing: Obtain a three-dimensional vascular image sequence and construct a diseased vascular model including an aneurysm and a parent artery.

[0103] Step 2, virtual implantation of the stent: Extract the parent artery parameters according to the diseased vascular model, recommend the stent specifications for supporting the parent artery based on the parent artery parameters, and generate point cloud data for the stent according to the stent specifications, the parent artery parameters, and the pre-designed stent cross-section.

[0104] Step 3, aneurysm cavity separation: Separate the aneurysm cavity according to the diseased vascular model and reconstruct the aneurysm cavity model.

[0105] Step 4, virtual embolization of the coil: Obtain the coil parameters, determine the starting point of the coil according to the aneurysm cavity model, iteratively generate a set of candidate points for each time step of the coil based on the starting point and the coil parameters, and screen the candidate points according to the energy of each candidate point to determine the unique confirmed point for each time step. The unique confirmed points of all time steps constitute the point cloud data of the coil.

[0106] Step 5, model generation: Generate a stent model based on the point cloud data of the stent and generate a coil model based on the point cloud data of the coil.

[0107] In the embodiment, the memory can be a volatile memory at the proximal end, such as RAM, or a non-volatile memory, such as ROM, FLASH, floppy disk, mechanical hard disk, etc., or a remote storage cloud. The processor can be a central processing unit (CPU), a microprocessor (MPU), a digital signal processor (DSP), or a field programmable gate array (FPGA), that is, the steps of the real-time virtual implantation method of the stent-assisted coil can be implemented through these processors.

[0108] The embodiment also provides a real-time virtual implantation device for stent-assisted coils, such as Figure 8As shown in the figure, it includes an image processing module 810, a virtual stent implantation module 820, a tumor cavity separation module 830, a virtual coil embolization module 840 for coils, and a model generation module 850.

[0109] Among them, the image processing module 810 is used to obtain a three-dimensional vascular image sequence and construct a diseased blood vessel model including an aneurysm and a parent artery.

[0110] The virtual stent implantation module 820 of the stent is used to extract the parameters of the parent artery according to the diseased blood vessel model, recommend the stent specifications for supporting the parent artery based on the parameters of the parent artery, and generate point cloud data for the stent according to the stent specifications, the parameters of the parent artery, and the pre-designed stent cross-section.

[0111] The tumor cavity separation module 830 is used to separate the tumor cavity according to the diseased blood vessel model and reconstruct the aneurysm tumor cavity model.

[0112] The virtual embolization module 840 for coils is used to obtain the coil parameters, determine the starting point of the coil according to the aneurysm tumor cavity model, iteratively generate a set of candidate points for each time step of the coil based on the starting point and the coil parameters, and screen the candidate points according to the energy of each candidate point to determine the unique confirmed point for each time step. The unique confirmed points for all time steps constitute the point cloud data of the coil.

[0113] The model generation module 850 is used to generate a stent model based on the point cloud data of the stent and generate a coil model based on the point cloud data of the coil.

[0114] It should be noted that when the real-time virtual implantation device for stent-assisted coil performs real-time virtual implantation of stent-assisted coil, the above-mentioned examples of the functional modules should be used for illustration. The above functions can be allocated to different functional modules according to needs, that is, the internal structure of the terminal or server is divided into different functional modules to complete all or part of the functions described above. In addition, the real-time virtual implantation device for stent-assisted coil provided in the above embodiments and the embodiments of the real-time virtual implantation method for stent-assisted coil belong to the same concept. The specific implementation process is detailed in the embodiments of the real-time virtual implantation method for stent-assisted coil, which will not be elaborated here.

[0115] The above specific embodiments have described in detail the technical solutions and beneficial effects of the present invention. It should be understood that the above is only the most preferred embodiment of the present invention and is not used to limit the present invention. Any modifications, supplements, equivalent replacements, etc. made within the scope of the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A real-time virtual implantation device for stent-assisted coils, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, the following steps are implemented: Image processing: Obtain a three-dimensional vascular image sequence and construct a diseased blood vessel model including an aneurysm and the parent artery; Virtual implantation of the stent: Extract the parent artery parameters according to the diseased blood vessel model, recommend the stent specifications for supporting the parent artery based on the parent artery parameters, and generate point cloud data for the stent according to the stent specifications, the parent artery parameters, and the pre-designed stent cross-section; Lumen separation: Separate the aneurysm lumen according to the diseased blood vessel model and reconstruct the aneurysm lumen model; Virtual embolization of the coil: Obtain the coil parameters, determine the starting point of the coil according to the aneurysm lumen model, iteratively generate a set of candidate points for each time step of the coil based on the starting point and the coil parameters, and screen the candidate points according to the energy of each candidate point to determine the unique confirmed point for each time step. The unique confirmed points of all time steps constitute the point cloud data of the coil; Model generation: Generate a stent model based on the point cloud data of the stent and generate a coil model based on the point cloud data of the coil; Among them, iteratively generating a set of candidate points for each time step of the coil based on the starting point and the coil parameters includes: Taking the starting point as the unique confirmed point of the first time step, randomly generating a confirmation direction for the starting point, and randomly generating a confirmation direction and a unique confirmed point of the second time step on the conical surface with the starting point as the vertex and the starting point confirmation direction as the rotation axis; For the third time step and above starting from the initial point, first, generate an initial candidate direction for the current time step, including: determining the rotation axis according to the confirmed directions of the confirmed points of the previous two time steps starting from the current time step, calculating the rotation matrix according to the rotation axis and the initial rotation angle included in the coil parameters, and multiplying the rotation matrix by the confirmation direction of the confirmed point of the previous time step starting from the current time step to obtain the initial candidate direction of the current time step; then, generate a set of candidate points according to the initial candidate direction, including: generating candidate points on the solid azimuth angle centered on the initial candidate direction to form a set of candidate points, and the selection range of the solid azimuth angle is (0, 2π).

2. The real-time virtual implantation device for the stent-assisted coil according to claim 1, characterized in that, The extracting the parent artery parameters according to the diseased blood vessel model includes: Confirm the proximal entrance of the parent artery according to the diseased blood vessel model, and starting from the proximal entrance, generate the centerline, the sequence of radii along the line, and the sequence of tangent vectors along the line from the parent artery to all distal exits. Among them, the centerline, the sequence of radii along the line, and the sequence of tangent vectors along the line are collectively referred to as the parent artery parameters.

3. The real-time virtual implantation device of the stent-assisted coil according to claim 2, characterized in that, The recommending the specification parameters of the stent for supporting the parent artery based on the parent artery parameters includes: Select two points on the centerline of the parent artery at the expected proximal and distal positions of the stent, and recommend the stent specifications based on these two points. Among them, the stent specifications include the stent length specification and the stent diameter specification; For the stent length specification, calculate the length of the center line between two points, select the length specification in the stent length specification set that is closest to this length as the recommended stent length specification. At the same time, keep the distal position of the stent and update the proximal position of the stent according to the recommended stent length specification, so that the length of the center line between the distal end and the proximal end of the stent is equal to the recommended stent length specification; For the stent diameter specification, compare the radii at two points and select twice the larger radius as the reference value, or calculate the average value of the radius sequence between the two points and use twice the average value as the reference value. Then select the diameter specification in the stent diameter specification set that is closest to and greater than the reference value as the recommended stent diameter specification.

4. The real-time virtual implantation device for the stent-assisted coil according to claim 2, characterized in that, The stent specification includes the stent length specification. Generating point cloud data for the stent based on the stent specification, the aneurysm-bearing artery parameters, and the pre-designed stent cross-section includes: The pre-designed stent cross-section has periodically designed intersection points. Discretize the stent at equal intervals in the length direction according to the intersection points and resample at equal intervals on the center line; At each resampling point position, generate a circle representing the stent cross-section according to the radius along the line and the tangent vector along the line, and discretize the circle by angle to obtain N points along the contour of the circular cross-section. N is equal to the number of intersection points on the stent cross-section. Sort the points on each circular cross-section contour according to the helical structure of the stent, and the obtained point sequence is used as the point cloud data of the stent.

5. The real-time virtual implantation device for the stent-assisted coil according to claim 1, characterized in that, The obtaining of the coil parameters includes: Ignore the primary helical structure of the coil, and equivalent the coil to a circular cross-section filament with only a secondary helical structure and a radius of r, where r is the primary helical radius. Discretize the coil in length to obtain n line elements with a length of l, and at the same time obtain n + 1 points. The secondary helical radius R of the coil is equal to the initial curvature radius of the coil. According to the initial curvature radius, obtain the initial rotation angle a = l / R between the line elements.

6. The real-time virtual implantation device of the stent-assisted coil according to claim 1, wherein The determining of the starting point of the coil based on the aneurysm cavity model includes: Determine the geometric center of the aneurysm cavity based on the aneurysm cavity model, and randomly select a point on the sphere with this geometric center as the center as the starting point; Or, randomly generate a point located within the aneurysm cavity in the bounding box of the aneurysm as the starting point; Or, obtain the starting point determined by an interactive method.

7. The real-time virtual implantation device for stent-assisted coils according to claim 1, characterized in that The screening of candidate points based on the energy of each candidate point to determine the unique confirmed point at each time step includes: Calculate the total energy of each candidate point, which includes the energy Φ caused by the spatial relationship between the candidate point and the aneurysm cavity sac , the energy Φ caused by the spatial relationship between the candidate point and the known confirmed point col , the energy Φ of the spring coil bending deformation caused by the candidate direction formed by the candidate point and the previous confirmed point ben ; Screen the candidate point set according to the total energy of each candidate point, including: setting the candidate points where Φ sac is greater than zero as invalid candidate points, and selecting the candidate point with the lowest total energy from the remaining candidate points as the only confirmed point. When there are multiple candidate points with the lowest total energy, randomly select one candidate point as the only confirmed point: when Φ sac of all candidate points in the candidate point set are greater than zero, then delete the candidate point set, roll back to the candidate point set where the only confirmed point of the previous time step is located, and set the only confirmed point of the previous time step as an invalid candidate point. At the same time, select the candidate point with the lowest total energy in the candidate point set where the invalid candidate point is located as the new only confirmed point of the previous time step, and continue to regenerate the candidate point set for the current time step; When determining the unique confirmed point at the current time step, also determine the confirmation direction between the unique confirmed point at the current time step and the unique confirmed point at the previous time step.

8. The real-time virtual implantation device for stent-assisted coils according to claim 7, characterized in that, For the energy Φ caused by the spatial relationship between the candidate point and the aneurysm cavity sac , when the candidate point is located within the aneurysm cavity and the distance from the cavity is greater than the first helical radius of the coil, the energy Φ sac is zero, otherwise the energy Φ sac is greater than zero; Regarding the energy Φ caused by the spatial relationship between the candidate point and the known confirmed point col , when the distance between the candidate point and the confirmed point is greater than or equal to twice the first helical radius of the spring coil, then Φ col The energy is zero, otherwise the energy Φ col is greater than zero; The energy Φ caused by the bending deformation of the coil spring due to the candidate direction formed by the candidate point and the previous confirmed point ben , when the candidate direction is the initial candidate direction, the energy Φ ben is zero. Otherwise, calculate the included angle α between the candidate direction and the initial candidate direction, and the magnitude of the energy Φ ben has a positive correlation with the included angle: Φ ben = Αα Β , where A and B are parameters greater than zero.

9. A real-time virtual implantation device for a stent-assisted coil, characterized in that, Including an image processing module, a virtual implantation module of the stent, a cavity separation module, a virtual embolization module of the coil, and a model generation module, The image processing module is used to obtain a three-dimensional vascular image sequence and construct a diseased blood vessel model including an aneurysm and an aneurysm-bearing artery; The virtual implantation module of the stent is used to extract the aneurysm-bearing artery parameters according to the diseased blood vessel model, recommend the stent specification for supporting the aneurysm-bearing artery based on the aneurysm-bearing artery parameters, and generate point cloud data for the stent based on the stent specification, the aneurysm-bearing artery parameters, and the pre-designed stent cross-section; The aneurysm cavity separation module is used to separate the aneurysm cavity according to the diseased blood vessel model and reconstruct the aneurysm cavity model; The virtual embolization module of the coil is used to obtain coil parameters, determine the starting point of the coil according to the aneurysm cavity model, iteratively generate a set of candidate points for each time step of the coil based on the starting point and the coil parameters, and screen the candidate points according to the energy of each candidate point to determine the unique confirmed point for each time step. The unique confirmed points of all time steps constitute the point cloud data of the coil; The model generation module is used to generate a stent model based on the point cloud data of the stent and generate a coil model based on the point cloud data of the coil; Among them, iteratively generating a set of candidate points for each time step of the coil based on the starting point and the coil parameters includes: Taking the starting point as the unique confirmed point of the first time step, randomly generating a confirmation direction for the starting point, and randomly generating the confirmation direction and the unique confirmed point of the second time step on the conical surface with the starting point as the vertex and the starting point confirmation direction as the rotation axis; For the third time step and above starting from the initial point, first, generate an initial candidate direction for the current time step, including: determining the rotation axis according to the confirmed directions of the confirmed points of the previous two time steps starting from the current time step, calculating the rotation matrix according to the rotation axis and the initial rotation angle included in the coil parameters, and multiplying the rotation matrix by the confirmation direction of the confirmed point of the previous time step starting from the current time step to obtain the initial candidate direction of the current time step; then, generate a set of candidate points according to the initial candidate direction, including: generating candidate points on the solid azimuth angle centered on the initial candidate direction to form a set of candidate points, and the selection range of the solid azimuth angle is (0, 2π).

10. A real-time virtual implantation method for stent-assisted coils, characterized in that, It includes the following steps: Image processing: Obtain a three-dimensional blood vessel image sequence and construct a diseased blood vessel model including an aneurysm and the parent artery; Virtual implantation of the stent: Extract the parent artery parameters according to the diseased blood vessel model, recommend the stent specifications for supporting the parent artery based on the parent artery parameters, and generate point cloud data for the stent according to the stent specifications, the parent artery parameters, and the pre-designed stent cross-section; Aneurysm cavity separation: Separate the aneurysm cavity according to the diseased blood vessel model and reconstruct the aneurysm cavity model; Virtual embolization of the coil: Obtain coil parameters, determine the starting point of the coil according to the aneurysm cavity model, iteratively generate a set of candidate points for each time step of the coil based on the starting point and the coil parameters, and screen the candidate points according to the energy of each candidate point to determine the unique confirmed point for each time step. The unique confirmed points of all time steps constitute the point cloud data of the coil; Model generation: Generate a stent model based on the point cloud data of the stent and generate a coil model based on the point cloud data of the coil; Among them, iteratively generating a set of candidate points for each time step of the coil based on the starting point and the coil parameters includes: Taking the starting point as the unique confirmed point of the first time step, randomly generating a confirmation direction for the starting point, and randomly generating the confirmation direction and the unique confirmed point of the second time step on the conical surface with the starting point as the vertex and the starting point confirmation direction as the rotation axis; For the third time step and above starting from the initial point, first, generate an initial candidate direction for the current time step, including: determining the rotation axis based on the confirmed directions of the confirmed points in the previous two time steps starting from the current time step, calculating the rotation matrix according to the rotation axis and the initial rotation angle included in the spring coil parameters, and multiplying the rotation matrix by the confirmed direction of the confirmed point in the previous time step starting from the current time step to obtain the initial candidate direction for the current time step; then, generate a candidate point set based on the initial candidate direction, including: generating candidate points on the solid azimuth angle centered on the initial candidate direction to form a candidate point set, and the selection range of the solid azimuth angle is (0, 2π).

Images