A costal cartilage graft composite surface microstructure design method for cartilage defect repair

Abstract

The application discloses a rib cartilage graft composite surface microstructure design method for cartilage defect repair, which comprises the following steps: firstly, constructing a surface microstructure equivalent representative element database; then, establishing a mapping relationship among equivalent material parameters, microstructure type parameters and geometric parameters based on a multilayer perceptron (MLP); then, exploring optimal equivalent mechanical parameters of a graft interface microstructure; and finally, realizing intelligent design of the microstructure for the mechanical microenvironment reconstruction of a cartilage graft interface.

Description

Technical Field

[0001] This invention relates to a cartilage graft design method, specifically to a composite surface microstructure design method for rib cartilage grafts for cartilage defect repair. Background Technology

[0002] Osteoarthritis is one of the leading causes of disability among chronic degenerative diseases worldwide, with focal cartilage defects being a significant contributing factor. In my country, over 5 million people suffer from knee cartilage damage annually, and without timely intervention, more than 30% will progress to osteoarthritis within 10 years. Repairing cartilage defects is crucial for slowing the progression of osteoarthritis and improving patients' quality of life.

[0003] Cartilage tissue is avascular, nerveless, and lymphatic, with low cell density, resulting in significantly insufficient self-repair capabilities. Current mainstream surgical techniques each have limitations: microfractures, which involve puncturing subchondral bone to induce bone marrow stem cell migration to the defect site, are low-cost and simple to perform, but are only suitable for small defects and tend to produce fibrocartilage with poor mechanical properties, leading to a high long-term degradation rate; autologous chondrocyte transplantation, which involves expanding autologous chondrocytes in vitro and then reimplanting them, avoids immune rejection but is costly, prone to cell dedifferentiation during in vitro expansion, and struggles to reconstruct the natural layered structure of articular cartilage, resulting in poor mechanical properties; allogeneic cartilage transplantation is limited by donor shortages and immunogenicity constraints; mosaic inlay, through the combined implantation of multiple grafts, can repair large defects, but the interfaces are prone to granulation tissue and fibrocartilage formation, resulting in poor integration quality.

[0004] Autologous rib cartilage transplantation, which has emerged in recent years, has become an important alternative due to its abundant donors, lack of immunogenicity, similar physicochemical composition to articular cartilage, and ability to repair large-area defects. However, articular cartilage exhibits a layered biomechanical modulus, and there are differences in biomechanical modulus between rib cartilage and articular cartilage. This can easily lead to stress concentration and shielding at the transplantation interface, deteriorating the interfacial microenvironment and resulting in poor integration, thus raising questions about long-term efficacy. Furthermore, the ability to reconstruct the collagen fiber network between the graft and the host cartilage has not been reported.

[0005] Therefore, in order to solve the problems of interfacial mechanical mismatch and microenvironment deterioration in current rib cartilage transplantation, it is necessary to develop a surface treatment method for rib cartilage grafts to regulate the interfacial mechanical microenvironment and promote the structural integration of the graft and the host cartilage. Summary of the Invention

[0006] To address the issues of stress concentration and shielding at the transplant interface caused by the differences in mechanical properties such as elastic modulus and permeability between rib cartilage and articular cartilage, and to the problems arising from the fact that traditional rib cartilage transplantation surgery relies solely on endogenous cell proliferation and secretion for tissue healing, resulting in poor lateral integration between the graft and the host cartilage and a tendency for graft degeneration in the long term, this invention provides a method for designing a composite surface microstructure for rib cartilage grafts for cartilage defect repair.

[0007] The objective of this invention is achieved through the following technical solution:

[0008] A method for designing the composite surface microstructure of rib cartilage grafts for cartilage defect repair includes the following steps:

[0009] Step 1: Construct a database of Representational Elements (RVEs) for surface microstructures:

[0010] Step 1.1, Surface microstructure parameter sampling:

[0011] (1) For microstructure geometric parameters (diameter) / width ,depth ,cycle The reasonable boundaries of the microstructure geometric parameters are determined using the equivalent medium theory and processing constraints, and random sampling is performed. The specific steps are as follows:

[0012] (a) Based on the limitations of the processing technology and the conditions for the validity of the equivalent medium theory, the range of geometric parameters of the microstructure is defined and a preliminary sampling set is generated using Latin hypercube sampling (LHS);

[0013] (b) On the initial sampling set, calculate the partial derivatives of the equivalent Young's modulus and equivalent permeability with respect to the structural geometric parameters based on the equivalent medium theory;

[0014] (c) Random sampling is added to the neighborhood of sample points whose partial derivative magnitude is greater than a certain threshold to increase the sample density;

[0015] (d) Based on the geometric validity conditions of the microstructure and the equivalent material performance requirements, sample points that do not meet the requirements are deleted to obtain an intermediate sample set. The geometric validity condition is the diameter. / width Less than period The equivalent material performance requirement is the equivalent Young's modulus. Not less than the Young's modulus of the host cartilage, equivalent permeability Not greater than the permeability of the host cartilage;

[0016] (2) For microstructure type parameters, if they exist For each type of microstructure, the following approach is adopted. Encode using a 3D unit vector, where the last bit of the vector is used to identify the valid or invalid state of the microstructure;

[0017] (3) Incorporate the microstructure type parameters into the intermediate sample set to construct an effective geometric parameter sample set for the surface microstructure;

[0018] Step 1.2: Obtain the equivalent representative element (RVE) material parameters:

[0019] (1) Based on the periodicity of surface microstructure The samples were grouped, and the largest feature size in each group was selected to construct a finite element model; the RVE height was fixed as the depth. The length and width of the plane are gradually increased to 3 times the length and width of the plane for convergence analysis. When the relative change rate of the equivalent stiffness matrix and the equivalent permeability matrix of adjacent iteration steps is lower than the preset threshold, the boundary effect of RVE is determined to be negligible. The current plane size is the length and width of the equivalent representative element of the period group.

[0020] (2) Traverse the effective geometric parameter sample set, carry out porous media coupling simulation based on COMSOL Multiphysics, measure the RVE equivalent stiffness matrix and equivalent permeability matrix, and obtain an effective dataset of equivalent material properties;

[0021] (3) In the effective dataset, obtain the maximum and minimum values ​​of the 27 parameters of the equivalent material properties (anisotropic equivalent stiffness matrix: 21 parameters; anisotropic equivalent permeability matrix: 6 parameters) respectively to obtain the effective material parameter range. Outside the effective range, obtain sample points containing invalid material parameters by random number generation and set their microstructure geometric parameters to all 0 values.

[0022] (4) Merge the valid and invalid datasets to finally obtain the surface microstructure RVE structural parameters-equivalent material properties dataset;

[0023] Step 2: Construct and train a surface microstructure reverse design model based on a multilayer perceptron (MLP):

[0024] Step 2.1: The surface microstructure reverse design model is a multilayer perceptron model constrained by physical information, consisting of an input layer, hidden layers, and an output layer. All layers are fully connected, with an equivalent stiffness matrix. With equivalent permeability matrix As input, the output is the geometric continuity parameters of the microstructure and the discrete parameters of the structure type;

[0025] Step 2.2: Train the surface microstructure reverse design model based on the dataset obtained in Step 1;

[0026] Step 3: Obtain the optimal interface equivalent mechanical parameters:

[0027] Step 3.1, Construction of the joint-scale finite element model:

[0028] A finite element analysis model of joint-scale transplanted cartilage was constructed based on COMSOL Multiphysics. The transplantation interface was modeled as a thin layer of homogeneous medium and divided into N layers according to the depth from the joint surface. Each layer was independently assigned material properties, and the loading condition was set as a static Hertz load.

[0029] Step 3.2, Initial sampling of parameter space:

[0030] For each microstructure, samples were taken from the surface microstructure RVE structural parameters-equivalent material properties dataset, and M sets of test points were constructed using random matching.

[0031] Step 3.3, Optimization of equivalent material parameters:

[0032] (1) Traverse the test points and use finite element analysis to calculate the weighted average of the solid phase stress gradient and liquid phase pressure gradient along the normal of the transplanted interface in the neighborhood of the interface layer.

[0033] (2) Select the test point with the smallest average value, use random sampling in the neighborhood of the equivalent material parameters of the point, and obtain M sets of test points. Input the test points into the neural network constructed in step 2 and delete the test points whose prediction type is invalid.

[0034] (3) Repeat the above steps until the change in the average value is less than a certain threshold, thereby determining the optimal equivalent material parameters that enable a smooth transition between solid-phase stress and liquid-phase pressure at the transplantation interface.

[0035] Step 4: Implement reverse design of micro / nano structures based on multilayer perceptron (MLP):

[0036] Based on the optimal equivalent material parameters obtained in step 3, the surface microstructure reverse design model based on multilayer perceptron (MLP) constructed in step 2 is used to reverse design the required microstructure type and geometric size for each partition, forming a microstructure design scheme. Finally, microstructures at the hundred-nanometer scale are linearly superimposed on the formed microstructure design scheme to promote the contact guidance effect on cells.

[0037] Compared to existing technologies, this invention innovatively introduces surface micro / nano structures into cartilage repair surgery, effectively solving problems such as poor lateral integration between the graft and host cartilage, stress concentration, and degeneration that arise from traditional surgery relying solely on biological self-healing. The specific beneficial effects are mainly reflected in the following aspects:

[0038] (1) By utilizing surface micro-nano structures to reduce the equivalent stiffness of the transplantation interface layer and increase the equivalent permeability, a smooth transition between solid-phase stress and liquid-phase pressure at the transplantation interface is achieved, and an interfacial mechanical microenvironment suitable for cartilage repair is constructed. This solves the stress concentration and stress shielding phenomenon between the graft and the host cartilage in traditional joint repair surgery, avoids excessive load on cells leading to apoptosis, and improves the long-term stability of the graft.

[0039] (2) By creating cell-scale surface morphology through surface micro-nano structures, endogenous mesenchymal stem cells and chondrocyte progenitor cells are promoted to adhere, grow and generate pericellular matrix at the transplantation interface through contact guidance effect, which effectively promotes biological healing at the transplantation interface. Attached Figure Description

[0040] Figure 1 Flowchart for the design of composite surface microstructures of rib cartilage grafts for cartilage defect rehabilitation;

[0041] Figure 2 Schematic diagram of the micro-nano structure on the surface of a rib cartilage graft;

[0042] Figure 3 Flowchart for constructing a structural parameter sample set;

[0043] Figure 4 Flowchart for constructing the RVE structure parameter-equivalent material property database for surface microstructures;

[0044] Figure 5 Reverse design of neural network structure diagrams for surface microstructures;

[0045] Figure 6 This is a schematic diagram of the finite element model and boundary conditions for transplanted cartilage. Detailed Implementation

[0046] The technical solution of the present invention will be further described below with reference to the accompanying drawings, but it is not limited thereto. Any modifications or equivalent substitutions to the technical solution of the present invention that do not depart from the spirit and scope of the technical solution of the present invention should be covered within the protection scope of the present invention.

[0047] This invention provides a method for designing composite surface microstructures of rib cartilage grafts for cartilage defect repair. The method first constructs an equivalent representative metadata database of surface microstructures, then establishes a mapping relationship between equivalent material parameters and microstructure type and geometric parameters based on multilayer perceptron (MLP), subsequently explores the optimal equivalent mechanical parameters of the transplant interface microstructure, and finally achieves intelligent microstructure design for reconstructing the mechanical microenvironment of the cartilage transplant interface. Figure 1 As shown, the specific steps are as follows:

[0048] Step 1: Construct a database of Representational Elements (RVEs) for surface microstructures:

[0049] Step 1.1, Surface microstructure parameter sampling:

[0050] To avoid the inefficiency of uniform sampling in high-dimensional parameter spaces, the reasonable boundaries of microstructure parameters are determined using equivalent medium theory and processing constraints, and random sampling is performed.

[0051] The types of microstructure parameters that need to be determined include structure type parameters (discrete, one-hot encoded) and structure geometry parameters (continuously varying).

[0052] For structure type parameters, for Various types of microstructures (microgrooves, micropillars, micropores, etc.) are employed. The 3D unit vector is encoded, with the last bit used to encode whether it is valid or invalid.

[0053] For structural geometric parameters (diameter) / width ,depth ,cycle Based on the limitations of the processing technology and the conditions for the validity of the equivalent medium theory, after defining the range of geometric parameters of the microstructure, a preliminary sampling set is generated using Latin hypercube sampling (LHS). .

[0054] In the initial sampling set The above calculations are based on the isotropic equivalent medium theory, which calculates the equivalent Young's modulus (the ratio of normal stress to normal strain in a certain direction) and the partial derivatives of the equivalent permeability with respect to the structural geometric parameters. ,like The magnitude of the partial derivative is greater than a certain threshold, i.e. or If the sample density is increased, 10 to 20 random sampling points are added to its neighborhood.

[0055] According to the geometric condition (diameter) / width Less than period Sample points that do not meet the requirements are deleted based on the equivalent material performance requirements (equivalent Young's modulus not less than Young's modulus of host cartilage, equivalent permeability not greater than permeability of host cartilage).

[0056] Finally, the structure type parameter (one-hot encoding) is applied. Merge into intermediate sample set Obtain a valid parameter sample set .

[0057] Through the above sampling, an effective geometric parameter sample set of surface microstructure is constructed, and the sample size can be controlled between 2000 and 3000 sets.

[0058] Step 1.2: Obtain the equivalent representative element (RVE) material parameters:

[0059] The equivalent representative element is set as a cuboid with a square base, and its material parameters include: equivalent stiffness matrix. Equivalent penetration matrix Its geometric dimensions include: length and width. Gao Wei .

[0060] The samples were grouped based on the periodicity P of the surface microstructure. Select A finite element model is constructed using the largest feature size among the sample groups. The height of the Relative Value Environment (RVE) is fixed to the depth. Three times that of the initial RVE geometry, set the initial RVE geometry to be... .

[0061] Convergence analysis was performed by progressively increasing the length and width of the plane. RVE finite element simulations were run using COMSOL multiphysics to measure the material's equivalent mechanical parameters. When the relative rates of change of the equivalent stiffness matrix and equivalent permeability matrix in adjacent iterations are both below a preset threshold, i.e.:

[0062]

[0063] and

[0064]

[0065] When established, the boundary effect of RVE is deemed negligible, and the current equivalent representative element's geometric dimensions ( This represents the geometric dimensions of the equivalent representative element of the periodic group. If the above criteria are not met, then the length and width dimensions of the plane are increased.

[0066]

[0067] Proceed to the next simulation iteration.

[0068] By traversing the effective geometric parameter sample set, a porous medium coupling simulation was carried out based on COMSOL Multiphysics to measure the equivalent stiffness matrix and equivalent permeability matrix of RVE, and obtain an effective dataset of equivalent material properties.

[0069] In the valid dataset, the maximum and minimum values ​​of 27 parameters of equivalent material properties (anisotropic equivalent stiffness matrix: 21 parameters; anisotropic equivalent permeability matrix: 6 parameters) were obtained to determine the effective range of material parameters. Outside the valid range, sample points containing invalid material parameters were obtained by random number generation, and their structural geometric parameters were set to all zero values.

[0070] By merging valid and invalid datasets, the final dataset of surface microstructure RVE structural parameters-equivalent material properties is obtained.

[0071] Step 2: Construct and train a surface microstructure reverse design model based on a multilayer perceptron (MLP):

[0072] Step 2.1: The surface microstructure reverse design model is a multilayer perceptron model constrained by physical information, consisting of an input layer, hidden layers, and an output layer, with fully connected layers. This model uses an equivalent stiffness matrix... (Contains 21 independent parameters, corresponding to a 21-dimensional vector, used to characterize the general anisotropic case) and the equivalent permeability matrix (Contains 6 independent parameters, corresponding to a 6-dimensional vector, used to characterize anisotropic permeability) as input, outputting the geometrical continuity parameters of the microstructure (diameter) ,depth ,cycle ( ) and discrete parameters of structure type.

[0073] Step 2.2: Train the surface microstructure reverse design model based on the dataset obtained in Step 1, and select appropriate hyperparameters (learning rate, dropout) to improve training efficiency and prevent overfitting.

[0074] Step 3: Obtain the optimal interface equivalent mechanical parameters:

[0075] Step 3.1, Construction of the joint-scale finite element model:

[0076] like Figure 6 As shown, a finite element analysis model of joint-scale transplanted cartilage was constructed based on COMSOL Multiphysics. The transplantation interface was modeled as a thin layer of homogeneous medium and divided into N layers according to the depth from the joint surface. Each layer was independently assigned material properties (stiffness matrix and permeability matrix), and the loading condition was set as a static Hertz load.

[0077] Step 3.2, Initial sampling of parameter space:

[0078] For each microstructure, 20-30 sample points are collected from the effective parameter set of RVE equivalent material properties, and M test points are constructed using random matching.

[0079] Step 3.3, Optimization of equivalent material parameters:

[0080] By iterating through the test points, the weighted average of the solid-phase stress gradient and liquid-phase pressure gradient along the normal direction of the transplanted interface is calculated using finite element analysis within the neighborhood of the interface layer. The test point with the smallest average value is selected, and random sampling is performed within the neighborhood of the equivalent material parameters of that point to obtain M sets of test points. These test points are then input into the neural network constructed in step two, and test points with invalid prediction types are deleted. The above steps are repeated until the change in the average value is less than a certain threshold, thereby determining the optimal equivalent material parameters that enable a smooth transition between the solid-phase stress and liquid-phase pressure at the transplanted interface.

[0081] Step 4: Implement reverse design of micro / nano structures based on multilayer perceptron (MLP):

[0082] Based on the optimal equivalent material parameters obtained in step 3, the surface microstructure reverse design model based on multilayer perceptron (MLP) constructed in step 2 is used to reverse design the required microstructure type and geometry for each partition, forming a microstructure design scheme. Finally, microstructures at the hundred-nanometer scale are linearly superimposed on the formed microstructure design scheme, that is, hundred-nanometer scale structures are added to the planar parts (e.g., the original plane of the unprocessed microstructure, the bottom of the trench, the bottom of the pore, and the top of the pillar) to promote the contact guidance effect on cells.

[0083] Example:

[0084] To repair large-area cartilage defects in the weight-bearing area of ​​the femoral condyle of the knee joint (defect area > 2 cm). 2 Taking an autologous sixth rib cartilage (2 mm deep) as an example, the micro-nano surface matching the mechanical modulus of the host cartilage was constructed by applying the micro-nano structure reverse design method based on multilayer perceptron described in this invention.

[0085] Step 1: Construct a database of Representational Elements (RVEs) for surface microstructures:

[0086] Step 1.1 Structural parameter sampling:

[0087] Based on the host cartilage's mechanical properties and processing constraints, the boundaries of microstructural parameters were initially determined:

[0088] (a) Period : 5μm~100 μm;

[0089] (b) Width W (micropore / microchannel) or diameter (Microcolumn): 0.5 μm ~ 100 μm (must meet the following requirements) < or < );

[0090] (c) Depth : 5 μm~100 μm.

[0091] Latin hypercube sampling was used to generate 4000 initial samples within the boundary.

[0092] For structural geometric parameters (diameter) / width ,depth ,cycle According to the equivalent medium theory, the equivalent Young's modulus along the surface normal direction is... Equivalent penetration rate The approximate formulas are as follows (taking square micropores as an example):

[0093] (1)

[0094] (2)

[0095] in:

[0096] (3)

[0097] in, , The original elastic modulus and permeability of the material.

[0098] Calculate the partial derivatives of formulas (1) and (2) with respect to the structural parameters at the initial sample points. The magnitude of the partial derivatives should be greater than 20 kPa / μm or 1 × 10⁻⁶. -15 [m 4 For each sample point of [[(N·s)] / μm], 15 additional random samples are taken in its neighborhood.

[0099] Remove samples that do not meet the following constraints:

[0100] (a) ≥ 0.38 MPa: Ensures stress gradient transition at the transplantation interface.

[0101] (b) ≤ 5×10 -14 m 4 / (N·s): Prevents excessive penetration that could lead to rapid local hydraulic pressure loss.

[0102] (c) ≥ 0.5μm or ≥ 0.5μm: Laser processing linewidth limit:

[0103] (d) < or < Avoid geometric invalidity.

[0104] This yields an intermediate sample set. Subsequently, the structural type parameters (one-hot encoding) are incorporated into the sample points to obtain the final sample set, covering three structural types: micropores, microgrooves, and micropillars.

[0105] Step 1.2: Obtain the equivalent representative element (RVE) material parameters:

[0106] The sample set is divided into three groups based on the period P:

[0107] Group G1 (Short Cycle): = 5~20 μm, RVE planar dimensions from 2 ×2 Start iterating;

[0108] Group G2 (Mid-cycle): = 20~60 μm, RVE planar size from 1.5 ×1.5 Start iterating;

[0109] Group G3 (Large Cycle): = 60~100 μm, RVE planar dimensions from 1.2 ×1.2 Start iterating.

[0110] The RVE height is fixed as the microstructure depth. Three times the size of the sample point, and select the sample point with the largest structural size within the group (e.g., width). Maximum or depth (Maximum), constructing a solid-liquid coupling model in COMSOL Multiphysics:

[0111] Porous elastic domain: Young's modulus of the costal cartilage solid-phase skeleton =5.6 MPa, Poisson's ratio =0.1, penetration rate =1.1×10 -15 m 4 / (N·s);

[0112] Laminar flow domain: interstitial fluid dynamic viscosity μ = 0.894 Pa·s.

[0113] Boundary conditions: Periodic boundary conditions are applied to the porous elastic domain, and the interface between the laminar flow domain and the porous medium domain is set with porous medium flow and laminar flow coupling conditions.

[0114] Set up load groups, apply unit strain and unit pressure gradient in each group, calculate the volume average stress and flow velocity in different directions, and thus obtain the equivalent stiffness matrix and equivalent permeability matrix.

[0115] By progressively increasing the RVE plane size (increasing by 0.5 each time). When the equivalent stiffness matrix of adjacent iteration steps (21 independent components) and equivalent permeability matrix When the relative change rate of all 6 independent components is less than 5%, the RVE is considered valid. Record the RVE size at this time as the RVE unit size for that group.

[0116] By traversing the structural parameter sample points and using the corresponding RVE element size, the above solid-liquid coupling model is constructed, and the equivalent stiffness matrix and equivalent permeability matrix are obtained through testing. This allows for the construction of an RVE database mapping "structural parameters - equivalent performance".

[0117] Step 2: Construct and train a surface microstructure reverse design model based on a multilayer perceptron (MLP):

[0118] Constructing a multilayer perceptron (MLP) reverse design model ( Figure 5 ):

[0119] Input layer: Receives the normalized target equivalent stiffness matrix C (21-dimensional) and permeability matrix K (6-dimensional), for a total of 27 features;

[0120] Hidden layers: 5 fully connected layers, with 512 neurons in the first and second layers, 256 neurons in the third and fourth layers, and 128 neurons in the fifth layer. The Tanh activation function is used, and Dropout (rate=0.2) is introduced to prevent overfitting.

[0121] Output layer: branched into two independent headers:

[0122] Regression head: Outputs 3D continuous geometric parameters (W', H', P'), which are then denormalized to obtain the actual dimensions;

[0123] Classification head: Outputs 4-dimensional uniquely heated codes (micropore [1,0,0,0], microgroove [0,1,0,0], micropillar [0,0,1,0], invalid [0,0,0,1]).

[0124] Training uses a composite loss function:

[0125]

[0126] Among them, the first item Geometric parameters (width) ,depth ,cycle The mean square error of ) in the second term; The cross-entropy loss is used to classify the structure type; the third term is a geometric hard constraint penalty term (ensuring that the period is greater than the width). =0.5, =10. Using the Adam optimizer (initial learning rate 1×10⁻⁶). -3 ), train for 200 epochs.

[0127] Step 3: Obtain the optimal interface equivalent mechanical parameters:

[0128] Step 3.1, Construction of the joint-scale finite element model:

[0129] Construct a two-dimensional axisymmetric finite element model of transplanted cartilage.

[0130] Geometric parameters: cartilage thickness hc = 2mm, costal cartilage region: Host cartilage region: Porting interface area: The transplant interface area is divided into 4 regions according to depth (thickness h=0.5mm).

[0131] Mesh parameters: Use COMSOL free trihedral mesh, minimum mesh size 0.02mm, maximum mesh size 0.5mm.

[0132] Material constitutive model:

[0133] Articular cartilage: an isotropic porous elastic constitutive model; Poisson's ratio, Young's modulus, and permeability are determined by the following formulas:

[0134]

[0135]

[0136]

[0137] in, This is the distance from the joint surface. The cartilage compression modulus is taken as 0.38 MPa. The initial permeability is taken as 2E-16 m^4 / (N·s).

[0138] Costal cartilage: isotropic porous elastic constitutive structure, Poisson's ratio Young's modulus penetration rate .

[0139] Transplantation interface region: anisotropic porous elastic constitutive model, composed of stiffness matrix and penetration matrix definition.

[0140] Boundary conditions: such as Figure 6 As shown, the formula for the boundary line load is as follows:

[0141]

[0142] in, The load is set at 2000N, which is within the normal physiological load range for the human body. The contact radius is represented and calculated as follows:

[0143]

[0144] in, The equivalent modulus of the contralateral cartilage is taken as 2 MPa.

[0145] Step 3.2, Initial sampling of interface material parameters:

[0146] In the equivalent material parameter effective dataset constructed in step 1, 30 sample points are randomly selected from each of the three microstructures—micropores, microgrooves, and micropillars—for a total of 90 sample points. Four sample points are then randomly selected from these 90 sample points and assigned to four domains of the transplantation interface, generating 30 assignment schemes. (A 4×27 matrix, with each row representing the material parameters assigned to the corresponding transplantation interface subdomain).

[0147] Step 3.3, Optimization of equivalent material parameters:

[0148] Perform transient finite element analysis on each set of assignment schemes and extract the interface mechanical response at t=120s:

[0149] (a) Definition of the test domain: The entire transplant interface layer (300 μm wide) and its adjacent 200 μm regions of the cartilage body, totaling 800 μm wide and 2 mm high, denoted as... .

[0150] (b) Gradient calculation: In Calculation of solid phase von Mises stress With pore fluid pressure along Domain average of directional gradient , .

[0151] (c) Construction of the objective function: ,in , Priority is given to ensuring the stress continuity of the solid skeleton.

[0152] (d) Initial screening: Selecting those that... The five sets of parameters with the smallest values ​​are used as preliminary optimal candidate solutions. .

[0153] (e) Sample augmentation: Let... For each In its neighborhood Ten random sample points are selected from the sample data, for a total of 50 sample points. These are then input into the neural network constructed in step 2, and sample points that are identified by the network as having invalid structures (the last digit of the structure type parameter > 0.8) are filtered out. This yields the test point set for the next iteration.

[0154] (f) Parameter optimization: Repeat (a)-(e) iteratively, if in the th... In the round of iteration, If the iteration terminates, then the iteration ends.

[0155] This allows us to obtain the optimal equivalent material parameters that enable a smooth transition between stress and pressure at the transplantation interface.

[0156] Step 4: Implement reverse design of micro / nano structures based on multilayer perceptron (MLP):

[0157] Input the target performance of each layer obtained in step 3 into the reverse design model constructed in step 2 to obtain a deeply dependent layered design scheme.

[0158] Microgrooves 300 nm to 500 nm wide and 1 μm to 3 μm deep are superimposed on the obtained microstructures to induce intracellular mechanotransduction effects and promote cell growth, secretion, and differentiation.

[0159] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A method for designing the microstructure of a composite surface of rib cartilage grafts for cartilage defect repair, characterized in that... The method includes the following steps: Step 1: Construct a surface microstructure equivalent representation (RVE) database: Step 1.1, Surface microstructure parameter sampling: (1) For the geometric parameters of the microstructure, the reasonable boundary of the geometric parameters of the microstructure is determined by using the equivalent medium theory and processing constraints, and random sampling is performed; (2) For microstructure type parameters, if they exist For each type of microstructure, the following approach is adopted. Encode using a 3D unit vector, where the last bit of the vector is used to identify the valid or invalid state of the microstructure; (3) Incorporate the microstructure type parameters into the intermediate sample set to construct an effective geometric parameter sample set for the surface microstructure; Step 1.2: Obtain RVE material parameters: (1) Based on the periodicity of surface microstructure The samples were grouped, and the largest feature size in each group was selected to construct a finite element model; the RVE height was fixed as the depth. The length and width of the plane are gradually increased to 3 times the length and width of the plane for convergence analysis. When the relative change rate of the equivalent stiffness matrix and the equivalent permeability matrix of adjacent iteration steps is lower than the preset threshold, the boundary effect of RVE is determined to be negligible. The current plane size is the length and width of the equivalent representative element of the period group. (2) Traverse the effective geometric parameter sample set, carry out porous media coupling simulation based on COMSOL Multiphysics, measure the RVE equivalent stiffness matrix and equivalent permeability matrix, and obtain an effective dataset of equivalent material properties; (3) In the effective dataset, obtain the maximum and minimum values ​​of the 27 parameters of the equivalent material properties to obtain the effective material parameter range. Outside the effective range, obtain sample points containing invalid material parameters by random number generation and set their microstructure geometric parameters to all 0 values. (4) Merge the valid and invalid datasets to finally obtain the surface microstructure RVE structural parameters-equivalent material properties dataset; Step 2: Construct and train a surface microstructure reverse design model based on a multilayer perceptron: Step 2.1: The surface microstructure reverse design model is a multilayer perceptron model constrained by physical information, consisting of an input layer, hidden layers, and an output layer. All layers are fully connected, with an equivalent stiffness matrix. With equivalent permeability matrix As input, the output is the geometric continuity parameters of the microstructure and the discrete parameters of the structure type; Step 2.2: Train the surface microstructure reverse design model based on the dataset obtained in Step 1; Step 3: Obtain the optimal interface equivalent mechanical parameters: Step 3.1, Construction of the joint-scale finite element model: A finite element analysis model of joint-scale transplanted cartilage was constructed based on COMSOL Multiphysics. The transplantation interface was modeled as a thin layer of homogeneous medium and divided into N layers according to the depth from the joint surface. Each layer was independently assigned material properties, and the loading condition was set as a static Hertz load. Step 3.2, Initial sampling of parameter space: For each microstructure, samples were taken from the surface microstructure RVE structural parameters-equivalent material properties dataset, and M sets of test points were constructed using random matching. Step 3.3, Optimization of equivalent material parameters: (1) Traverse the test points and use finite element analysis to calculate the weighted average of the solid phase stress gradient and liquid phase pressure gradient along the normal of the transplanted interface in the neighborhood of the interface layer. (2) Select the test point with the smallest average value, use random sampling in the neighborhood of the equivalent material parameters of the point, and obtain M sets of test points. Input the test points into the neural network constructed in step 2 and delete the test points whose prediction type is invalid. (3) Repeat the above steps until the change in the average value is less than a certain threshold, thereby determining the optimal equivalent material parameters that enable a smooth transition between solid-phase stress and liquid-phase pressure at the transplantation interface. Step 4: Implement reverse design of micro / nano structures based on multilayer perceptrons: Based on the optimal equivalent material parameters obtained in step 3, the surface microstructure reverse design model based on multilayer sensor constructed in step 2 is used to reverse design the required microstructure type and geometric size for each partition, forming a microstructure design scheme. Finally, microstructures at the hundred-nanometer scale are linearly superimposed on the formed microstructure design scheme to promote the contact guidance effect on cells.

2. The method for designing the composite surface microstructure of rib cartilage grafts for cartilage defect repair according to claim 1, characterized in that... In step 1.1, the microstructure geometric parameters include diameter. / width ,depth ,cycle .

3. The method for designing the composite surface microstructure of rib cartilage grafts for cartilage defect repair according to claim 1, characterized in that... In step 1.1, the specific steps of (1) are as follows: (a) Based on the limitations of the processing technology and the conditions for the validity of the equivalent medium theory, the range of geometric parameters of the microstructure is defined and a preliminary sampling set is generated by Latin hypercube sampling; (b) On the initial sampling set, calculate the partial derivatives of the equivalent Young's modulus and equivalent permeability with respect to the structural geometric parameters based on the equivalent medium theory; (c) Random sampling is added to the neighborhood of sample points whose partial derivative magnitude is greater than a certain threshold to increase the sample density; (d) Based on the conditions for the establishment of microstructure geometry and the requirements for equivalent material properties, delete the sample points that do not meet the requirements to obtain the intermediate sample set.

4. The method for designing the composite surface microstructure of rib cartilage grafts for cartilage defect repair according to claim 3, characterized in that... The condition for the geometry to hold is the diameter. / width Less than period The equivalent material performance requirement is the equivalent Young's modulus. Not less than the Young's modulus of the host cartilage, equivalent permeability Not greater than the permeability of the host cartilage.

5. The method for designing the composite surface microstructure of rib cartilage grafts for cartilage defect repair according to claim 1, characterized in that... In step 1.2, the 27 parameters include the anisotropic equivalent stiffness matrix (21 parameters) and the anisotropic equivalent permeability matrix (6 parameters).

Images