In-situ printing track planning method

Abstract

According to the in-situ printing track planning method provided by the invention, an integrated point cloud data processing flow is designed after 3D scanning, so that manual intervention is reduced; the LSCM algorithm and the ARAP algorithm are respectively utilized to carry out plane parameterization and dimension reduction on the three-dimensional triangular mesh so as to reduce the implementation difficulty and improve the path planning efficiency, and the two methods are mutually verified so as to ensure the parameterization effect; solving a grid correspondence problem by utilizing vertex topology uniqueness, and designing a dynamic reasoning mechanism, namely, when fingerprints are not completely unique, gradually deducing to global correspondence through limited vertex corresponding information in a triangle to adapt to uncertainty of defective grid information; a defective skull layering path is integrally generated through input bone tissue parameters and curved surface defect grids, and the structural characteristics of an original skull are simulated. The whole process of the method from point cloud preprocessing to path generation is automatic, intervention is reduced, response is quick, adaptation individuation is improved, and skull tissue function recovery is promoted through bionic design.

Description

Technical Field

[0001] The present invention relates to the technical field of tissue engineering, and in particular to an in-situ printing trajectory planning method. Background Art

[0002] Skull defect is a relatively common disease in cranio-maxillofacial surgery. It is mainly seen in trauma, electric shock injury, infection, tumor resection, decompressive craniectomy for intracranial diseases, congenital craniofacial deformities, etc. Skull defect causes serious harm to the patient's physiology and psychology in many aspects. Therefore, skull repair is very necessary.

[0003] Bone tissue engineering technology is a relatively advanced bone defect treatment method in recent years. In essence, it is to composite a growth factor with osteogenic inductive activity and a biodegradable biological scaffold material, and then implant it into the defect site. As the biological scaffold gradually degrades, the newly formed bone tissue gradually replaces the scaffold material and fills the defect site, ultimately achieving the purpose of complete repair and regeneration. Compared with several existing mainstream treatment methods, it does not require autologous bone tissue, and the secondary trauma is low; the biodegradable scaffold material regenerates into completely autologous tissue, which can avoid the possible long-term immune rejection reaction after surgery.

[0004] In-situ printing is a method of directly printing and repairing tissues at the defect site according to the shape and characteristics of the defect site. In the aspect of biological printing for repairing tissue defects, it is a fast and simple solution, which can simplify the surgical process, improve the injury recovery speed, avoid in-vitro culture of artificial tissues, and reduce the operation time at the same time. Extrusion-based bioprinting is a bioprinting form more suitable for in-situ repair of skull defects. For relatively complex skull defect surgery situations, clamping the bioprinting extruder head on the end effector of a multi-axis robotic arm is a better solution, which can handle various complex defect morphologies and defect poses.

[0005] Anatomical studies have shown that bone structure includes three parts. The first part is bone substance, the second part is periosteum, and the third part is bone marrow. For bone substance, there are two forms, namely compact bone and cancellous bone structures; from the structure perspective, it can also be divided into cortical bone and cancellous bone parts, specifically as Figure 1 shown. The porosity of cortical bone is relatively low, between 3% and 30%, the tissue is dense and has a high hardness. The porosity of cancellous bone is relatively large (greater than 30%), the tissue arrangement is loose, and it is mostly composed of rod-shaped or sheet-shaped trabeculae connected by dots and intertwined to form a porous structure, which has good energy absorption capacity. For the skull, it is distributed in the middle layer. Generally speaking, the skull is a spherical shell-like structure, a classic sandwich structure of outer cortical bone - middle cancellous bone - inner cortical bone ( Figure 2) To perform in-situ printing repair on the defective cranial bone surface, its morphological and mechanical properties should be maintained as much as possible. Therefore, it is necessary to imitate the native bone tissue structure and print in layers.

[0006] Curved surface (layered) printing is a 3D printing technology that is opposite to "plane (layered) printing". The print head in plane printing technology mainly moves in 3 axes, usually stacking materials layer by layer on a horizontal plane. Because plane printing is relatively easy to implement, it has a wide range of applications. However, when facing structures to be printed with large surface curvature changes such as spheres and shells, it will produce a staircase effect, rather than an ideal continuous smooth spherical shell surface.

[0007] At present, the related research on in-situ bioprinting repair of cranial defects is still blank. By specifically designing the cranial bone structure, imitating the native bone structure to perform curved surface layering on the skull, and improving the mechanical properties of bioprinted tissues; retaining the multi-porous structure in the middle layer of the bone plate and leaving channels for cell migration, tissue nutrition, and transportation of metabolic wastes has very important clinical application value. Summary of the Invention

[0008] The technical problem to be solved by the present invention is to provide a method for planning in-situ printing trajectories.

[0009] The technical solution adopted by the present invention is as follows:

[0010] A method for planning in-situ printing trajectories, the specific steps are as follows:

[0011] (1) Sample data acquisition

[0012] Obtain the cranial defect range through a structured light 3D scanner, and apply a defect surface repair algorithm designed based on the Visual Studio platform to generate an STL file of the defective surface.

[0013] (2) Curved surface parameterization

[0014] Use the CGAL library in the Visual Studio platform to perform curved surface parameterization on the STL model of the defective surface using the LSCM and / or ARAP algorithm to generate a two-dimensional triangular surface mesh file.

[0015] (3) "Fingerprint" graph isomorphism processing

[0016] Adopt an algorithm for restoring the corresponding relationship between vertices in the case of "fingerprint" graph isomorphism for two-dimensional and three-dimensional triangular meshes based on the MATLAB platform.

[0017] (4) Curved surface path planning algorithm

[0018] Adopt an equidistant path module based on the grid boundary line on a two-dimensional triangular grid on the MATLAB platform to simulate the cortical bone structure; adopt a rounded Hilbert path module within the grid boundary line to simulate the cancellous bone structure.

[0019] Both of the two modules in the above step (4) can adjust the trajectory based on the diameter of the bio - printer extrusion head and the path filling ratio; overall, the morphology of the complete surface path planning can also be adjusted according to the extrusion head diameter and the cranial thickness data.

[0020] Preferably, for the above in - situ printing trajectory planning method, in the step (1), the repair algorithm automatically performs triangular meshing on the point cloud data of the defective bone tissue after pre - processing the point cloud data of the defective bone tissue, and then performs hole repair to output the surface model of the bone defect.

[0021] Preferably, for the above in - situ printing trajectory planning method, in the step (2), the LSCM algorithm mainly focuses on conformal transformation, aiming to find a two - dimensional parameterized coordinate to minimize the angular deformation on the surface triangular grid: the LSCM algorithm optimizes the parameterization process by minimizing the Dirichlet energy E. After derivation, the LSCM is formalized as a least - squares problem, and the core is to solve the linear system such as L g = 0, where L is the discrete Laplace matrix, and the coefficients are determined by the grid topology and edge weights. Then, use a least - squares solver (such as Cholesky decomposition) to solve the mapped two - dimensional coordinates.

[0022] Preferably, for the above in - situ printing trajectory planning method, in the step (2), the ARAP algorithm further controls local rigidity on the basis of the LSCM algorithm, that is, it tries to keep the triangle side lengths and shapes as much as possible and reduce local deformation. It achieves this goal through global optimization: by establishing an energy function E = ∑ (i,j) ω ij ||(u i - u j ) - R ij (v i - v j )|| 2 to measure the rigidity error before and after the triangle mapping. Since the energy function E depends on the transformation matrix R ij , the ARAP algorithm solves the mapped two - dimensional coordinates through multiple iterations of the alternating optimization method.

[0023] For the above in - situ printing trajectory planning method, in the step (3), the "fingerprint" in the graph isomorphism method of "fingerprint" refers to the number of first - order, second - order to higher - order adjacent points of each vertex in the triangular grid.

[0024] Preferably, in the above in-situ printing trajectory planning method, the algorithm for restoring the corresponding relationship between vertices when the triangular meshes are isomorphic in the "fingerprint" map in step (3) is as follows:

[0025] (1) In a three-dimensional triangular mesh, if the "fingerprint" information of a certain point is unique, then the point with this "fingerprint" information in the two-dimensional triangular mesh corresponds uniquely to this point in the three-dimensional mesh;

[0026] (2) If the fingerprint information of each point in not every surface triangular mesh is unique, but it can be ensured that the three vertices in at least one triangle correspond uniquely, then the vertex correspondence information between the two-dimensional and three-dimensional triangular meshes can be established, that is, the vertex connection of "graph isomorphism";

[0027] (3) If there is less unique "fingerprint" information of vertices, but the corresponding relationship is established between some edges (that is, two adjacent vertices), then find the "fingerprint" of the unknown vertex of one or two triangles to which this edge belongs. If this edge connects two vertices, it is necessary to judge whether the "fingerprints" of these two vertices are different. If they are different, a relationship can be established. If they are the same, look for the next group of known edges; if the edges of all triangles have no established corresponding relationship, then find a triangle that only corresponds to one vertex, and judge whether the "fingerprints" of the other two unknown points of this triangle are different. If they are different, find all the triangles containing this known point, find the only group where the fingerprints of the two unknown points also correspond, establish the relationship, and then extend it to all vertices.

[0028] Preferably, in the above in-situ printing trajectory planning method, after obtaining the planar triangular mesh range in step (4), three values are input: the diameter of the extruded hydrogel syringe, the porosity of cancellous bone, and the thickness of the cranial defect. First, taking the boundary line as the reference line and the syringe diameter as the offset, multiple concentric equidistant lines are made to offset inward, and they are connected end to end to form a single-layer cortical bone extrusion path. For cancellous bone, the minimum bounding square covering the triangular mesh is covered by Hilbert curves. The density of the Hilbert curves is determined by the diameter of the extruded hydrogel syringe and the input porosity of cancellous bone. The Hilbert curves with right-angled corners are smoothed into rounded corners, and then only the Hilbert curves within the triangular mesh are retained, and the parts outside the boundary are removed. For the vertices of the broken lines that need to be retained, the intersections of the broken lines and the mesh are found. If a single broken line intersects multiple triangular mesh lines, the intersection order is retained. Then, each broken line is classified as an entering boundary line / an exiting boundary line / an in-boundary line / an out-of-boundary line, and the mesh boundary line between the exiting boundary line and the entering boundary line is inserted. The endpoint information of the retained broken lines is mapped back to the three-dimensional mesh again, and the ratio of the intersection point of the triangular side line in the two-dimensional mesh to the vertex is taken. For the endpoints within the triangle, the centroid coordinates are maintained. By reflecting back to the three-dimensional triangular mesh in this way, a single-layer three-dimensional cortical bone / cancellous bone extrusion path can be obtained. Then, multiple extrusion paths are generated based on the syringe diameter and the cranial defect thickness, and the cortical bone / cancellous bone ratio of the native bone tissue is imitated to automatically allocate the structures of each layer. Then, the heads and tails of each layer are connected to generate a complete curved surface defect repair trajectory.

[0029] The beneficial effects of the present invention are as follows:

[0030] The above in-situ printing trajectory planning method can largely maintain the spatial complexity of the original native structure. The porous structure of the tissue is beneficial to cell adhesion, migration and proliferation, as well as the transportation of nutrients, oxygen and waste. It can be widely applied to the in-situ printing trajectory planning for tissue engineering to repair cranial defects, laying a foundation for the realization of using a cranial tissue engineering scaffold to repair cranial defects. Specifically:

[0031] 1. An integrated point cloud data processing flow is designed after 3D scanning, reducing manual intervention;

[0032] 2. The LSCM and ARAP algorithms are respectively used to perform planar parameterization on the three-dimensional triangular mesh, reducing the dimension to reduce the implementation difficulty and improving the path planning efficiency. These two methods are used to verify each other to ensure the parameterization effect;

[0033] 3. The vertex topology uniqueness is used to solve the mesh correspondence problem. A dynamic reasoning mechanism is designed. When the fingerprint is not completely unique, the global correspondence is gradually deduced through the limited vertex correspondence information within the triangle to adapt to the uncertainty of the defective mesh information;

[0034] 4. Integrate the input bone tissue parameters and the surface defect mesh to generate a layered path for the defective skull, mimicking the structural characteristics of the native skull. The overall process of the method is fully automated from point cloud preprocessing to path generation, reducing intervention, enabling rapid response, enhancing adaptability to individual cases, and promoting the functional recovery of the skull tissue through bionic design. Description of the Drawings

[0035] Figure 1 It is the curved surface layered structure of the cortical bone and cancellous bone of the skull.

[0036] Figure 2 It is the sandwich structure of the skull.

[0037] Figure 3 It is the flowchart of the defective surface repair algorithm.

[0038] Figure 4 It is the flowchart of the LSCM algorithm in the surface parameterization method.

[0039] Figure 5 It is the flowchart of the ARAP algorithm in the surface parameterization method.

[0040] Figure 6 It is to map a 3D mesh to 2D based on the LSCM method.

[0041] Figure 7 It is to map a 3D mesh to 2D based on the ARAP method.

[0042] Figure 8 (a) shows the vertex numbers of the 3D mesh; Figure 8 (b) shows the vertex numbers of the 2D mesh.

[0043] Figure 9 It is an example of the vertex "fingerprint" method.

[0044] Figure 10 It is the flowchart of re - establishing the connection between triangular meshes based on the "fingerprint" graph isomorphism method.

[0045] Figure 11 (a) is the isometric line trajectory mimicking the cortical bone; Figure 11 (b) is the Hilbert line trajectory mimicking the cancellous bone.

[0046] Figure 12 It is a structured light 3D scanner scanning the defective skull.

[0047] Figure 13 It is a skull defect model.

[0048] Figure 14 It is the surface STL model generated by the defective surface repair algorithm.

[0049] Figure 15Simulate cancellous bone tissue with the Hilbert line trajectory generated by applying the surface path algorithm.

[0050] Figure 16 Simulate cortical bone tissue with the isometric line trajectory generated by applying the surface path algorithm.

[0051] Figure 17 (a) is the complete multi-layer surface path; Figure 17 (b) is the sectional view of the multi-layer path. Specific implementation mode

[0052] In order to enable those skilled in the art to better understand the technical solution of the present invention, the technical solution of the present invention will be further described in detail below in conjunction with the accompanying drawings and specific implementation modes.

[0053] The idea of in-situ surface printing adopted by the present invention is as follows: for the upper and lower layers of cortical bone tissue, isometric lines generated based on the boundary line are designed to be densely paved on the surface; for the middle layer of cancellous bone tissue, considering the use of an extrusion-type bioprinting robotic arm, the Hilbert curve is used to imitate the morphology of cancellous bone on the surface. The Hilbert curve is a continuous curve that can traverse all points in the unit square and has the characteristic of being drawn in one stroke, which is beneficial for maintaining the shape of the hydrogel biomaterial during the extrusion process on the surface by the robotic arm and not being damaged by the circular path. The specific implementation mode is as follows:

[0054] Example 1

[0055] An in-situ printing trajectory planning method has the following steps:

[0056] (1) Defect surface acquisition

[0057] Scan the defect bone tissue with a VTOP-Designer desktop structured light scanner (Tianjin Weishen Technology Co., Ltd.) to obtain the defect point cloud data. Automatically perform preprocessing such as rotation, clipping, and filtering on the point cloud data of the defect bone tissue to obtain uniform defect bone tissue point cloud data with only surface features; apply a defect door surface repair algorithm designed based on the Visual Studio platform to automatically perform triangular meshing on the point cloud data, and detect and reconstruct the hole edge features; then automatically locate, surface construct, and surface smooth the meshed holes to quickly and accurately output the surface model of the bone defect and generate the STL file of the defect surface. The flow chart of its repair algorithm is as Figure 3 shown.

[0058] (2) Surface parameterization processing

[0059] Using the CGAL library in the Visual Studio platform, the STL model of the defective surface is processed by the LSCM or ARAP algorithm for surface parameterization to generate a two-dimensional triangular surface mesh file:

[0060] Since the structure of the native skull is a sandwich structure and the overall is a spherical shell-like structure, when using bioprinting to repair it, the traditional plane layering method cannot maintain its original structural strength and cannot ensure a smooth printing process at all angles. Therefore, the STL surface obtained by scanning needs to be used as a single-layer whole for path planning. However, directly performing trajectory planning on a surface in three-dimensional space is time-consuming and inefficient. So, the surface parameterization method is adopted to map the three-dimensional triangular mesh of the spatial surface to a two-dimensional triangular mesh by the LSCM (Least Squares Conformal Mapping) method or the ARAP (As-Rigid-As-Possible Mapping) method.

[0061] For the LSCM algorithm, it should keep the angles of the triangles as the same as possible during the mapping process, while the ARAP algorithm, based on the LSCM, tries to ensure that the triangles are mapped onto the two-dimensional plane without distortion. Taking the ARAP algorithm as an example, the principle of the surface parameterization method is introduced as follows: Since it is necessary to keep the triangles from being distorted as much as possible during the mapping process, the transformation matrix is a rotation matrix, and the corresponding parameterization result is obtained by restricting the transformation matrix.

[0062] Suppose a triangle t in the three-dimensional mesh is represented as The corresponding triangle after mapping is represented as The relationship between these two triangles can be represented by a 2×2 Jacobian matrix J t (u).

[0063] Let L(t) represent the restricted transformation matrix, that is, the rotation transformation matrix. The following energy function can be defined to represent the difference between the possible transformation and the target transformation:

[0064]

[0065] where A t is the area of the triangle. The energy function can also be reconstructed as:

[0066]

[0067] Solve for the minimum value of the energy function and take its derivative

[0068]

[0069] where θ ijis the angle opposite to the side. In the above formula, the unknowns are the vertex u of the mapping on the two-dimensional plane and the transformation matrix L. Therefore, the parameterization algorithm is constructed as an optimization problem as follows:

[0070] (u, t) = argmin(u, t) E(u, t), L t ∈M(4)

[0071] Since L t is restricted to a rotation matrix (isometric transformation matrix) and has the following form:

[0072]

[0073] There are two unknowns in the energy function of Equation (4). Solving it is divided into three steps:

[0074] 1. Locally, fix all u and find the optimal rotation matrix L for each triangle t . Initialize u with the two-dimensional coordinates obtained using the LSCM algorithm. For the following 2×2 matrix:

[0075]

[0076] It can be decomposed by SVD as: J = U∑V T , where

[0077]

[0078] Then the optimal transformation matrix L = UV can be obtained T

[0079] 2. Globally, fix the rotation matrix corresponding to each triangle to solve for the optimal U. Substitute the solved L from the previous step into the linear system for solving the minimum of the above energy equation to solve for U. The left side of its equation is a fixed matrix, and a linear matrix AX = B can be constructed to solve it.

[0080] 3. Repeat the above two steps until the energy change is not obvious.

[0081] When applying these two algorithms to solve the skull defect surface parameterization problem, the results are relatively similar. Their flowcharts are as Figure 4 (LSCM), Figure 5 (ARAP) shown. After the LSCM method, the comparison after mapping the three-dimensional mesh to two dimensions is as Figure 6 shown; after the ARAP method, the comparison after mapping the three-dimensional mesh to two dimensions is as Figure 7 shown.

[0082] (3) Recover the graph isomorphism feature using local topological consistency

[0083] After the mesh is subjected to surface parameterization through an algorithm and mapped to a two-dimensional plane, a problem of loss of vertex correspondence between two triangular meshes will occur. As Figure 8 shown, that is, the sequence numbers should have changed at the corresponding vertex positions, but the relationship between vertices cannot be simply re-established directly between points with relatively small coordinate errors in terms of distance. Because the mapping is approximately a projection process of flattening the surface, for different triangular surfaces, it is difficult to determine the projection direction; in addition, before the conformal mapping method is used, the larger the range of the three-dimensional surface and the greater the curvature, the less closely related the characteristics of the mapped two-dimensional mesh and the three-dimensional mesh are. Therefore, it is necessary to use the topological relationship between the vertices of the triangular mesh to re-find the vertex relationship before and after the mapping. Since the two conformal mapping methods used in this method do not change the topological relationship (the connection relationship between vertices) of the triangular mesh, this invariant relationship is used to establish a connection between the two triangular meshes. For this purpose, the present invention proposes a method of establishing "fingerprint" information for each vertex to find the corresponding point sequence numbers before and after the mapping. As Figure 10 shown, it is a simple triangular mesh relationship. For point 1, the number of its first-order adjacent points, that is, the number of points within the range of a connecting line, is 4. Similarly, it can be deduced that the number of second-order adjacent points is 4 and the third-order is 0. From this, it can be deduced that Figure 10 the "fingerprint" information of point 1 in the figure is 4-4-0; similarly, the fingerprint relationships of all points in this example can be deduced, as shown in Table 1. It can be seen from the table that for the vertices in each triangular mesh, the "fingerprint" information before and after the mapping corresponds to each other. And for Figure 9 this example, except that points 5 and 6 have the same "fingerprint" information, the fingerprint information of the remaining points is unique, and as long as the "fingerprint" information is unique, it is available. For such triangular meshes obtained from skull defect scans and repairs, the triangles are all edge-edge connected, and there will be no situation where point-point connections result in triangular islands in the mesh. And before and after the mapping of such triangular meshes, only by establishing the vertex correspondence relationship between a pair of triangles, the corresponding relationships between all the remaining triangular meshes can be deduced. The flow chart of re-establishing the connection between triangular meshes by the "fingerprint" graph isomorphism method is as Figure 10 shown. Taking the example of being able to establish the relationship between the three vertices of a triangle between two triangular meshes, first input the three-dimensional triangular mesh A and the mapped two-dimensional triangular mesh B, then construct their respective adjacency matrices, and then calculate the number of first-order adjacent points of each vertex until the fourth order or more. Establish an index relationship between the vertices with unique and identical "fingerprints" between A and B. If three vertices can be found corresponding in a row, then based on the topological relationship of the triangle, the relationship between a known point (already forming a triangle) and an unknown point can be derived from two points, and then the corresponding relationships between all unknown triangle pairs can be deduced.

[0084] Table 1 Figure 9The number of n - order connection points of each vertex

[0085]

[0086]

[0087] (4) Curved surface trajectory planning

[0088] An equidistant line path module based on the grid boundary line on a two - dimensional triangular grid is used on the MATLAB platform to simulate the cortical bone structure; a rounded - corner Hilbert line path module within the grid boundary line range is used to simulate the cancellous bone structure:

[0089] After obtaining the range of the planar triangular grid, three values are input: the diameter of the water - gel - extruding syringe, the porosity of the cancellous bone, and the thickness of the cranial defect. First, taking the boundary line as the reference line and the syringe diameter as the offset, multiple circles of equidistant lines are made to offset inward and connected end - to - end to form a single - layer cortical bone extrusion path. For the cancellous bone, the smallest enclosing square of the triangular grid is covered by the Hilbert line, and the density of the Hilbert line is determined by the diameter of the water - gel - extruding syringe and the input porosity of the cancellous bone. The Hilbert line with a right - angled corner is smoothed into a rounded - corner turn, and then only the Hilbert lines within the triangular grid are retained, and the parts outside the boundary are removed. For the vertices of the broken lines that need to be retained, the intersection points of the broken lines and the grid are found. If a single broken line intersects multiple triangular grid lines, the intersection point order is retained. Then, each broken line is classified as an entering - boundary - line / exiting - boundary - line / in - boundary - line / out - of - boundary - line, and the grid boundary line between the exiting - boundary - line and the entering - boundary - line is inserted. The endpoint information of the retained broken lines is remapped back to the three - dimensional grid, and the ratio of the intersection point of the triangle side line in the two - dimensional grid to the vertex is obtained. For the endpoints inside the triangle, the centroid coordinates are maintained, and by this method, the three - dimensional cortical bone / cancellous bone extrusion path of a single layer can be obtained. Then, a multi - layer extrusion path is generated based on the syringe diameter and the cranial defect thickness, and the cortical bone / cancellous bone ratio of the native bone tissue is imitated to automatically allocate the structures of each layer, and then each layer is connected end - to - end to generate a complete curved - surface defect repair trajectory.

[0090] After being able to map the spatial triangular grid to the planar triangular grid, trajectory planning is carried out on the planar triangular grid, which is divided into two types of trajectories. One is the equidistant line trajectory imitating the cortical bone, and the other is the Hilbert line trajectory imitating the cancellous bone. These two trajectories are projected on the planar triangular grid, and the intersection points and special points between the trajectory line and the grid line are retained. Then, these points are remapped back to the three - dimensional grid based on the relative position information of the points on the triangular grid, and the trajectory planning of a single - layer curved surface is completed. On this basis, according to the diameter of the extrusion head, the thickness information of the single - layer curved surface, the porosity information of the native cranial cancellous bone, and the cranial thickness information near the defect range, a multi - layer curved - surface trajectory planning is generated, as Figure 11Shown are two trajectories that mimic cortical bone and cancellous bone. The normal vector obtained by weighted averaging the area with the normal vectors of the small triangles on the triangular surface is offset multiple times based on the thickness of the extrusion head in this direction to obtain the overall extrusion path.

[0091] Example 2

[0092] To verify the feasibility of the method described in Example 1, a cranial defect model was used to evaluate the effect.

[0093] After structural light scanning ( Figure 12 、 Figure 13 ) and processing, the triangular mesh of the surface defect as shown in Figure 14 was obtained. After stratifying it, different types of surface trajectory planning were carried out according to cortical bone and cancellous bone, as shown in Figure 15 、 Figure 16 . On this basis, the triangular mesh was weighted according to the triangle area within the range of the defective cranial surface, and offset by 10 layers with the single-layer thickness of the bioprinting extrusion head (taking 1 mm as an example) to generate a complete multi-layer surface path, as shown in Figure 17 (a). Figure 17 (b) shows a sectional view of the layered structure of the multi-layer path, and the result shows that it is similar to the sandwich structure of the native bone.

[0094] It can be seen from this that the generated cranial repair surface path planning is well-defined, stratified according to the original structure of the defective surface, and can mimic the cortical bone and cancellous bone structures of the native bone tissue as much as possible. The repair is natural. Only three parameters, namely the diameter of the extrusion head, the porosity of the cancellous bone, and the thickness of the cranial defect, need to be input throughout the process, and it only takes 1 - 2 minutes.

[0095] The bionic-designed cortical bone path is dense and provides mechanical support; the cancellous bone is realized by Hilbert lines, optimizing the right-angle corners into arcs, reducing stress concentration and expanding the connectivity of the pores. Its highly connected fractal characteristics can simulate the trabecular network, and the continuous and dead-end-free path design can avoid closed cavities, ensuring that cells, nutrients, and metabolic wastes can migrate freely along the path gaps. The staggered gaps between the curves form multi-level pores, which can accelerate the diffusion of metabolites.

[0096] In summary, the cranial repair surface path planning method can meet the requirements of timely and bionic repair of cranial tissue, and it is completely feasible to be applied to in-situ printing during medical procedures.

[0097] The above-described embodiments are only descriptions of the preferred embodiments of the present invention, and do not limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations and improvements made by those of ordinary engineering and technical personnel in the art to the technical solutions of the present invention shall fall within the protection scope of the present invention.

Claims

1. An in-situ printing trajectory planning method, characterized in that: The specific steps are as follows: (1) Sample data acquisition The skull defect range is obtained through a structured light 3D scanner, and an STL file of the defect surface is generated; (2) Surface parameterization processing The STL model of the defect surface is processed by the LSCM and / or ARAP algorithm for surface parameterization to generate a two-dimensional triangular surface mesh file; (3) Isomorphic processing of "fingerprint" map An algorithm for restoring the corresponding relationship between vertices in the case of isomorphic "fingerprint" maps of two-dimensional and three-dimensional triangular meshes based on the MATLAB platform is adopted; (4) Surface path planning algorithm An equidistant line path module based on the grid boundary line on the two-dimensional triangular mesh is adopted on the MATLAB platform to simulate the cortical bone structure; a rounded Hilbert line path module within the grid boundary line is adopted to simulate the cancellous bone structure.

2. The in-situ printing trajectory planning method according to claim 1, wherein: In the step (1), after preprocessing the point cloud data of the defective bone tissue by a repair algorithm, the point cloud data of the defective bone tissue is automatically triangulated, and then hole repair is performed to output a surface model of the bone defect and generate an STL file of the defect surface.

3. The in-situ printing trajectory planning method according to claim 1, wherein: In step (2), the LSCM algorithm is used to focus on conformal transformation to find a two-dimensional parameterized coordinate to minimize the angular deformation on the surface triangular mesh. The LSCM algorithm optimizes the parameterization process by minimizing the Dirichlet energy E. Through derivation, the LSCM is formulated as a least squares problem, and the core is to solve a linear system such as L q = 0, where L is the discrete Laplacian matrix, and the coefficients are determined by the mesh topology and edge weights. Then, a least squares solver is used to solve the mapped two-dimensional coordinates, thereby realizing the surface parameterization of the STL model of the defective surface.

4. The in-situ printing trajectory planning method according to claim 1, characterized in that: In step (2), the ARAP algorithm further controls local rigidity on the basis of the LSCM algorithm, that is, it tries to keep the triangle side lengths and shapes as much as possible and reduce local deformation. It achieves this goal through global optimization: by establishing an energy function \(E=\sum (i，j) \omega ij ||(u i -u j )-R ij (v i -v j )|| 2 to measure the rigidity error before and after triangle mapping. Since the energy function \(E\) depends on the transformation matrix \(R ij , the ARAP algorithm solves for the two-dimensional coordinates after mapping through multiple iterations using the alternating optimization method.

5. The in-situ printing trajectory planning method according to claim 1, wherein: The algorithm for restoring the corresponding relationship between vertices in the case of isomorphic "fingerprint" maps of triangular meshes in the step (3) is as follows: (1) In the three-dimensional triangular mesh, if the "fingerprint" information of a certain point is unique, then the point with this "fingerprint" information in the two-dimensional triangular mesh corresponds uniquely to this point in the three-dimensional mesh; (2) If the fingerprint information of each point in each surface triangular mesh is not unique, but it can be ensured that the three vertices in at least one triangle correspond uniquely, then the corresponding information of the vertices of the two-dimensional and three-dimensional triangular meshes can be established; (3) If the unique "fingerprint" information of the vertices is less, but the corresponding relationship is established between some edges, then find the "fingerprint" of the unknown vertex of one or two triangles to which this edge belongs. If this edge connects two vertices, it is necessary to judge whether the "fingerprints" of these two vertices are different. If they are different, a relationship can be established. If they are the same, look for the next group of known edges; if the edges of all triangles have no corresponding relationship, then find a triangle that only corresponds to one vertex, and judge whether the "fingerprints" of the other two unknown points of this triangle are different. If they are different, find all the triangles containing this known point, find the only group in which the fingerprints of the two unknown points also correspond, establish a relationship, and then extend it to all vertices.

6. The in-situ printing trajectory planning method according to claim 1, characterized in that: In step (4), after obtaining the range of the planar triangular mesh, input three values: the diameter of the extruded hydrogel syringe, the porosity of the cancellous bone, and the thickness of the skull defect. First, taking the boundary line as the reference line and the syringe diameter as the offset, draw multiple concentric equidistant lines that offset inward, and connect their heads and tails into one line to form a single-layer cortical bone extrusion path. For the cancellous bone, use the Hilbert line to cover the minimum bounding square of the triangular mesh. The density of the Hilbert line is determined by the diameter of the extruded hydrogel syringe and the input porosity of the cancellous bone. Smooth the Hilbert line with a right-angle corner into a rounded corner, and then only retain the Hilbert line within the triangular mesh and remove the part outside the boundary. For the vertices of the broken line that need to be retained, find the intersection points of the broken line and the mesh. If a single broken line intersects multiple triangular mesh lines, retain the intersection order. After that, classify each broken line as an entering boundary line / an exiting boundary line / an in-boundary line / an out-boundary line, and insert the mesh boundary line between the exiting boundary line and the entering boundary line. Map the endpoint information of the retained broken line back to the three-dimensional mesh, and obtain the ratio from the intersection point of the triangle side line in the two-dimensional mesh to the vertex. Keep the centroid coordinates for the endpoints inside the triangle, and use this method to reflect back to the three-dimensional triangular mesh to obtain a single-layer three-dimensional cortical bone / cancellous bone extrusion path. Then, generate a multi-layer extrusion path based on the syringe diameter and the skull defect thickness, and automatically allocate the structure of each layer according to the cortical bone / cancellous bone ratio of the native bone tissue. Then connect the heads and tails of each layer to generate a complete curved surface defect repair trajectory.

Images