A three-dimensional curved surface unfolding method for implant manufacturing, electronic device, storage medium and program product

Abstract

Embodiments of the present application provide a three-dimensional curved surface unfolding method for implant manufacturing, an electronic device, a storage medium and a program product. The method comprises: obtaining a three-dimensional curved surface model of a target biological tissue structure; extracting feature information for constraining an unfolding form from the three-dimensional curved surface model; mapping a surface mesh of the three-dimensional curved surface model to an equal-area two-dimensional plane based on the feature information and a surface area of the three-dimensional curved surface model; and generating a two-dimensional unfolding graph for implant manufacturing based on the two-dimensional plane. By obtaining the three-dimensional curved surface model of the target biological tissue structure, extracting the feature information for constraining the unfolding form, and mapping the surface mesh to the equal-area two-dimensional plane based on the feature information and the surface area, a two-dimensional unfolding graph for implant manufacturing is finally generated, accurate conversion from a three-dimensional complex curved surface to a two-dimensional plane is realized, and the problem of inaccurate implant blank size caused by area distortion of a traditional unfolding method is solved.

Description

Technical Field

[0001] This application relates to the field of medical technology, and more specifically, to a three-dimensional surface unfolding method, electronic device, storage medium, and program product for implant manufacturing. Background Technology

[0002] With the integration of medicine and digital manufacturing, designing personalized implants based on individual patient anatomy has become a clinical trend in orthopedics, oral and maxillofacial surgery, and plastic surgery. For scenarios such as bone defect repair and soft and hard tissue reconstruction, the morphological fit of the implant directly affects surgical outcomes and long-term prognosis.

[0003] In the digital design process of personalized implants, accurately unfolding a three-dimensional curved surface model into a two-dimensional plane is a crucial step. For example, in the repair of skull defects, the three-dimensional curved surface of the skull needs to be flattened to design the cutting template for the repair mesh; in maxillofacial bone reconstruction, complex curved surfaces such as the zygomatic bone and maxilla need to be unfolded to plan the shape of the osteotomy guide or reconstruction titanium plate.

[0004] However, in related technologies, the mapping and conversion from three-dimensional curved surfaces to two-dimensional planes has defects such as area distortion and morphological distortion, resulting in large deviations in the size of the implant blank and poor fit with the defect area. Repeated adjustments are required during the operation, which affects the efficiency of the operation and the repair effect.

[0005] In summary, there is a current need for a three-dimensional surface unfolding method that can meet the precise design requirements of personalized implants in multiple fields such as orthopedics and oral and maxillofacial surgery. Summary of the Invention

[0006] The purpose of this application is to provide a three-dimensional surface unfolding method, electronic device, storage medium, and program product for implant manufacturing, so as to achieve the technical effect of accurately converting complex three-dimensional surfaces into two-dimensional planes and improving the dimensional accuracy of implant blanks.

[0007] The first aspect of this application provides a three-dimensional surface unfolding method for implant manufacturing, the method comprising: Obtain a three-dimensional surface model of the target biological tissue structure; Extract feature information for constraining the unfolded shape from the three-dimensional surface model; Based on the feature information and the surface area of ​​the three-dimensional surface model, the surface mesh of the three-dimensional surface model is mapped to a two-dimensional plane with equal area; Based on the two-dimensional plane, a two-dimensional unfolded diagram for implant manufacturing is generated.

[0008] In the above implementation process, by obtaining a three-dimensional surface model of the target biological tissue structure, extracting feature information for constraining the unfolding morphology, and mapping the surface mesh to a two-dimensional plane of equal area based on the feature information and surface area, a two-dimensional unfolding diagram for implant manufacturing is finally generated, realizing the accurate conversion from a three-dimensional complex surface to a two-dimensional plane, and solving the problem of inaccurate implant blank size caused by area distortion in traditional unfolding methods.

[0009] Furthermore, the acquisition of the three-dimensional surface model of the target biological tissue structure includes: Acquire tomographic image data; the tomographic image data is obtained by performing medical imaging scans on a region including the target biological tissue structure. The tomographic image data is segmented to extract the contour information of the target biological tissue; Based on the contour information, a mesh model composed of polygonal patches is generated using a 3D reconstruction algorithm; The mesh model is defined as the three-dimensional surface model.

[0010] In the above implementation process, by acquiring tomographic image data, performing image segmentation to extract contour information, and generating a mesh model composed of polygonal patches based on the contour information through a three-dimensional reconstruction algorithm, a link is established from the original medical image data to the high-precision three-dimensional surface model, ensuring that the three-dimensional surface model can correctly represent the surface morphology of the target biological tissue structure.

[0011] Further, the step of extracting feature information from the three-dimensional surface model to constrain the unfolded shape includes: The three-dimensional surface model is input into the trained recognition model to obtain feature information, which includes at least one of bone defect boundary, anatomical axis and contour line.

[0012] In the above implementation process, by inputting the three-dimensional surface model into the trained recognition model, at least one of the bone defect boundary, anatomical axis and contour line is automatically identified and extracted as feature information, thus achieving efficient extraction of key anatomical features and avoiding the subjectivity problem of manual annotation.

[0013] Further, the step of mapping the surface mesh of the three-dimensional surface model to a two-dimensional plane of equal area based on the feature information and the surface area of ​​the three-dimensional surface model includes: Construct an energy function, which includes an area conservation term to maintain the consistency of the area before and after unfolding, and a constraint term to map the feature information to the corresponding position or shape of the two-dimensional plane. The energy function is optimized using an iterative algorithm, and the coordinates of each node in the surface mesh on the two-dimensional plane are adjusted. The iteration algorithm stops when it meets the preset convergence condition, so that the error rate between the total area of ​​the two-dimensional plane and the surface area of ​​the three-dimensional curved surface model is lower than the first preset threshold.

[0014] In the above implementation process, by constructing an energy function that includes area conservation terms and feature constraint terms, and using an iterative algorithm to optimize the solution, under the premise of satisfying the mapping of feature information to the corresponding position or shape of the two-dimensional plane, the error rate between the total area of ​​the two-dimensional plane and the surface area of ​​the three-dimensional curved surface model is lower than the first preset threshold, thus achieving high-precision equal area mapping and ensuring that the unfolded two-dimensional plane is conserved with the original three-dimensional curved surface on the global area scale.

[0015] Further, generating a two-dimensional unfolded diagram for implant manufacturing based on the two-dimensional plane includes: The region in the two-dimensional plane corresponding to the feature information is subjected to local mesh densification processing; Based on the original arc length of the feature information on the three-dimensional surface model, the positions of the mesh nodes in the encrypted region are adjusted so that the percentage difference between the length of the curve corresponding to the feature information on the two-dimensional plane and the original arc length is lower than a second preset threshold, thereby obtaining the two-dimensional unfolded diagram; the percentage difference is determined according to the ratio of the difference between the arc length of the spatial curve of the feature information on the three-dimensional surface model and the length of the corresponding planar curve on the two-dimensional plane.

[0016] In the above implementation process, by performing local mesh densification on the region corresponding to the feature information in the two-dimensional plane, and adjusting the position of the mesh nodes in the densified region based on the original arc length of the feature information on the three-dimensional curved surface model, the percentage difference between the length of the feature curve on the two-dimensional plane and the original arc length is lower than the second preset threshold. This effectively suppresses the morphological distortion of key anatomical regions such as root bifurcation and bone defect edges during the unfolding process, ensuring that the two-dimensional unfolded diagram retains the important morphological features of the three-dimensional curved surface, and improving the fitting accuracy between the implant and the defect area.

[0017] Furthermore, the method also includes: Obtain a local shape adjustment instruction for the two-dimensional unfolded diagram; the local shape adjustment instruction includes deformation parameters and the target area to be adjusted in the two-dimensional unfolded diagram; While keeping the total area change of the two-dimensional unfolded diagram below a third preset threshold, the grid node coordinates of the target region are stretched or shrunk based on the deformation parameters to obtain a locally adjusted two-dimensional unfolded diagram.

[0018] In the above implementation process, by obtaining local shape adjustment instructions containing deformation parameters and target area, and performing stretching or shrinking calculations on the grid node coordinates of the target area while keeping the total area change of the two-dimensional unfolded diagram below the third preset threshold, clinicians are given fine-tuning authority, allowing optimization of local morphology based on intraoperative observation or special case needs. At the same time, area monitoring ensures that local adjustments do not disrupt the global area conservation, thus achieving a combination of algorithm calculation and clinical expert experience.

[0019] Furthermore, the two-dimensional unfolded diagram is a digital graphic file, which includes outlines and at least one positioning reference mark; the method further includes: The digital graphic file is transmitted to the cutting device, so that the cutting device determines the processing origin according to the positioning reference mark, generates a cutting path according to the contour line, and cuts the implant blank according to the cutting path to obtain an implant that matches the shape of the three-dimensional curved surface model.

[0020] In the above implementation process, by generating a digital graphic file containing contour lines and at least one positioning reference mark from the two-dimensional unfolded drawing, and transmitting it to the cutting equipment so that it can determine the processing origin according to the positioning reference mark, generate the cutting path according to the contour line and perform cutting, the connection from three-dimensional design to finished implant is realized, improving the manufacturing efficiency and precision of personalized implants, and ensuring that the final implant is adapted to the shape of the three-dimensional curved surface model.

[0021] A second aspect of this application provides an electronic device, the electronic device comprising: processor; Memory used to store processor-executable instructions; Wherein, when the processor invokes the executable instructions, it implements any of the methods described in the first aspect.

[0022] A third aspect of this application provides a computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the steps of any of the methods described in the first aspect.

[0023] A fourth aspect of this application provides a computer program product, the computer program product including a computer program, which, when executed by a processor, implements any of the methods described in the first aspect. Attached Figure Description

[0024] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 A flowchart illustrating a three-dimensional surface unfolding method for implant manufacturing provided in this application embodiment; Figure 2 This is a structural block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0026] The technical solutions in the embodiments of this application will now be described with reference to the accompanying drawings.

[0027] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0028] Taking the oral cavity as a starting point, the repair of alveolar bone defects caused by periodontitis is a challenging problem in oral clinical practice. Guided tissue regeneration achieves tissue regeneration by isolating different cell populations through a barrier membrane, the core of which lies in the precise adaptation of the barrier membrane to the morphology of the bone defect. In clinical practice, commercially available rectangular barrier membranes are often used, which require the dentist to manually trim them according to the contour of the bone defect during the operation. This has the following drawbacks: First, it is highly dependent on the dentist's experience and judgment, resulting in low accuracy in morphological adaptation and a tendency for incomplete coverage or redundancy; second, the manual measurement and trimming process is time-consuming, increasing the risk of intraoperative infection; and third, traditional barrier membranes cannot match the curved surface features of complex bone defects such as root bifurcation lesions and subosseous pockets, affecting the regeneration effect.

[0029] Currently, the development of digital technology has made personalized barrier membrane design possible, but the mapping and conversion from three-dimensional curved surfaces to two-dimensional planes faces technical bottlenecks: conventional unfolding methods are prone to area distortion or morphological deformation. For example, simple stretching and unfolding can cause shape changes in critical areas of bone defects, while isometric mapping, although maintaining shape, cannot guarantee area conservation; the surface reconstruction technology of the Rowan PACS system can only achieve visual unfolding and cannot simultaneously meet the dual requirements of area and morphological fidelity. Therefore, designing a three-dimensional curved surface unfolding scheme that can take into account both area conservation and anatomical feature preservation is of great significance for promoting the digital and precise fabrication of periodontal barrier membranes.

[0030] This application provides a three-dimensional surface unfolding method for implant manufacturing, which can be applied to the repair of periodontal bone defects in the oral cavity. The implementation process and effects are as follows: A three-dimensional surface model of periodontal bone defects and surrounding tissues is reconstructed based on CBCT image data. The model includes the contour of the bone defect edge, root morphology, and anatomical landmarks such as the alveolar ridge crest. The bone defect boundary curve, root axis, and alveolar ridge crest line are extracted from the three-dimensional surface model as feature information constraining the unfolded morphology, constructing a topological association system of feature points, curves, and surfaces. Based on the feature information and the surface area of ​​the three-dimensional surface model, an equal-area mapping algorithm is used to map the surface mesh of the three-dimensional surface model to a two-dimensional plane of equal area. Energy minimization iterative solutions are used to control the error rate between the total area of ​​the two-dimensional plane and the original surface area within a preset range. Local mesh refinement and node position optimization are performed on the regions in the two-dimensional plane corresponding to the feature information to ensure that the morphological distortion rate of key features such as the bone defect boundary and root bifurcation is lower than a preset threshold after unfolding. A digital graphic file containing contour lines and positioning reference marks is generated, which can be directly transferred to a cutting device to create a personalized periodontal barrier membrane precisely adapted to the bone defect morphology.

[0031] The method described in this application is applied to the repair of periodontal bone defects in the oral cavity. It can solve the problems of insufficient precision and poor morphological adaptability in the traditional barrier membrane design by manual trimming. It can achieve efficient conversion from three-dimensional bone defect morphology to two-dimensional trimming template, and provide technical support for the digital production of personalized periodontal barrier membranes, thereby improving the clinical accuracy and operational efficiency of guided tissue regeneration.

[0032] To address any of the problems mentioned above, embodiments of this application provide a three-dimensional surface unfolding method for implant manufacturing, referring to... Figure 1 , Figure 1 This is a flowchart illustrating a three-dimensional surface unfolding method for implant manufacturing, provided in an embodiment of this application.

[0033] In this embodiment, the method includes: Step S10: Obtain a three-dimensional surface model of the target biological tissue structure; Step S20: Extract feature information from the three-dimensional surface model to constrain the unfolded shape; Step S30: Based on the feature information and the surface area of ​​the three-dimensional surface model, map the surface mesh of the three-dimensional surface model into a two-dimensional plane of equal area; Step S40: Based on the two-dimensional plane, generate a two-dimensional unfolded diagram for implant manufacturing.

[0034] It should be noted that the target biological tissue structure is a biological anatomical structure with a curved shape that requires the design of a conforming implant, including but not limited to hard tissues such as oral and maxillofacial bones, skull, ribs, articular surfaces, and pelvis, as well as soft tissue surfaces such as cartilage.

[0035] 3D surface model: A digital surface representation generated by medical image segmentation and 3D reconstruction techniques, consisting of polygonal facets (usually triangles). Its accuracy is determined by the number of facets, vertex density, and reconstruction algorithm.

[0036] Feature information refers to anatomical landmarks or curves extracted from the three-dimensional surface model to constrain the geometry of the unfolding process, including defect edges, anatomical axes, and physiological contour lines. Its function is similar to control points in map projection, ensuring that the unfolded result maintains the clinically required anatomical correspondence.

[0037] It should be understood that "equal area" in this application means the area is the same or similar, that is, the total area of ​​the unfolded two-dimensional plane can be mathematically equal to or similar to the surface area of ​​the original three-dimensional curved surface model. The so-called similar area means that the area error rate is controlled within a preset threshold range. This preset threshold can be set according to the clinical precision requirements. For example, in the case of periodontal bone defect repair, the preset threshold can be 2%. Within this error range, the collagen barrier membrane blank can accurately fit the bone defect area without secondary trimming, meeting the requirements of clinical use.

[0038] Implant manufacturing: This includes digital forming processes such as subtractive manufacturing (laser cutting, CNC milling, waterjet cutting) and additive manufacturing (3D printing, laser cladding). Two-dimensional unfolded diagrams serve as input files for the manufacturing process, driving processing equipment to generate physical implants that conform to the three-dimensional curved surface shape.

[0039] For example, the method provided in this embodiment can be applied to the design of personalized titanium mesh in mandibular segmental defect reconstruction surgery. In this scenario, firstly, a three-dimensional curved surface model of the target biological tissue structure is obtained; the target biological tissue structure is the defect area of ​​the patient's mandibular ameloblastoma after surgery. Mandibular image data is obtained through cone-beam CT scanning, and after three-dimensional reconstruction, a three-dimensional curved surface mesh model of the mandible composed of triangular facets is obtained. This model represents the anatomical structures such as the buccal and lingual contours of the mandible, the alveolar ridge morphology, and the course of the mandibular canal. Secondly, feature information for constraining the unfolded shape is extracted from the three-dimensional curved surface model, such as the lower edge contour line of the mandible, the anterior and posterior edge boundary lines of the defect, and the symmetry reference line of the healthy side as feature information. Among them, the anterior and posterior edge boundary lines of the defect are the reference for the subsequent titanium mesh cutting edge, the lower edge contour line of the mandible is used to prevent the shape of the restoration from being distorted in the direction of gravity, and the symmetry reference line of the healthy side serves as the morphological reference for restoring the symmetry of the patient's face, ensuring that the two-dimensional unfolded diagram is consistent with the anatomical shape of the patient's healthy side. Then, based on the feature information and the surface area of ​​the three-dimensional surface model, the surface mesh of the three-dimensional surface model is mapped to a two-dimensional plane of equal area. For example, the surface area of ​​the three-dimensional surface model of the mandibular defect area is 12.6 cm². 2 Using the anterior and posterior edge boundaries of the defect as fixed boundary constraints and the lower edge contour of the mandible as directional constraints, an equal-area mapping algorithm is employed to map the surface mesh of the 3D curved surface model one-to-one to a 2D plane, ensuring that the mesh topology remains unchanged before and after mapping. Finally, based on the 2D plane, a 2D unfolded diagram for implant fabrication is generated. The mapped 2D plane is processed into a DXF format 2D unfolded diagram containing trimming boundary lines, positioning reference marks, and chamfer lines of the prosthesis edges. This unfolded diagram can be directly imported into a CNC milling machine or a metal 3D printer for cutting or directly molding personalized titanium mesh prostheses.

[0040] Exemplarily, the method provided in this embodiment can also be applied to the design of titanium mesh prostheses in large-area skull defect repair surgery. In this scenario, firstly, a three-dimensional curved surface model of the target biological tissue structure is obtained. The target biological tissue structure is the skull defect area after decompression surgery of the temporoparietal bone. The patient's skull is scanned in thin slices using 64-slice spiral CT with a slice thickness of 0.625 mm to obtain DICOM format tomographic image data. Bone tissue is extracted by threshold segmentation algorithm, and a three-dimensional curved surface model of the outer plate of the skull is reconstructed by moving cube algorithm. The model includes the morphology of the bone defect edge, the physiological curvature of the skull, and anatomical markers such as the temporal line. Then, feature information for constraining the unfolded shape is extracted from the three-dimensional curved surface model. In this scenario, with the contralateral skull mirror data as a reference, the following feature information is identified and extracted using the curved surface registration algorithm: (1) the closed curve of the bone defect boundary, as the final edge of the titanium mesh cutting; (2) the midsagittal projection line of the skull, as the morphological benchmark for left and right symmetry; (3) the projection curve of the superior temporal crest, as the positioning benchmark to prevent the titanium mesh from rotating and misaligning. Then, based on the feature information and the surface area of ​​the three-dimensional surface model, the surface mesh of the three-dimensional surface model is mapped to a two-dimensional plane of equal area. The surface area of ​​the three-dimensional surface model of the skull defect area is 28.3 cm². 2 Using the bone defect boundary curve as a fixed constraint and the midsagittal projection line as an axial constraint, an energy function is constructed and iteratively optimized using the gradient descent method to map the irregular cranial surface into a two-dimensional plane of equal area. Finally, based on the two-dimensional plane, a two-dimensional unfolded diagram for implant manufacturing is generated. In this scenario, a two-dimensional unfolded diagram containing the bone defect contour line, the superior temporal crest positioning marker, and the location of the reserved screw holes is generated, exported as a general three-dimensional manufacturing format file, and used to drive a laser selective melting forming device to directly print the titanium alloy cranial prosthesis.

[0041] In this embodiment, by acquiring a three-dimensional curved surface model of the target biological tissue structure, extracting feature information for constraining the unfolding morphology, and mapping the surface mesh to a two-dimensional plane of equal area based on the feature information and surface area, a two-dimensional unfolding diagram for implant manufacturing is finally generated, realizing the accurate conversion from a three-dimensional complex curved surface to a two-dimensional plane, and solving the problem of inaccurate implant blank size caused by area distortion in traditional unfolding methods.

[0042] Based on any of the above embodiments, step S10 includes: Acquire tomographic image data; the tomographic image data is obtained by performing medical imaging scans on a region including the target biological tissue structure. The tomographic image data is segmented to extract the contour information of the target biological tissue; Based on the contour information, a mesh model composed of polygonal patches is generated using a 3D reconstruction algorithm; The mesh model is defined as the three-dimensional surface model.

[0043] It should be noted that tomographic imaging data refers to a continuous sequence of transverse, coronal, or sagittal images of the human body acquired through medical imaging equipment (CT, CBCT, MRI, etc.). Its parameters include slice thickness, slice spacing, pixel size, and grayscale resolution.

[0044] Image segmentation: The image processing procedure that separates target tissue from the surrounding background in medical images. Optionally, image segmentation includes thresholding, region growing, level set methods, or deep learning methods.

[0045] Contour information: The two-dimensional boundary curve of the target tissue on each tomographic image. The continuity, closure, and accuracy of the inter-layer correspondence of the contour affect the quality of the 3D reconstruction.

[0046] 3D reconstruction algorithms are mathematical methods for converting two-dimensional contour sequences into three-dimensional surface meshes. These algorithms include the moving cube method, Poisson reconstruction method, and Delaunay triangulation method. Among them, the moving cube method generates triangular patches by extracting the isosurfaces of voxel elements.

[0047] Polygonal facets: The basic units that make up a 3D mesh model, usually triangles. The number of facets, their shape quality (aspect ratio, interior angles), and topological connections together determine the computational stability and visual realism of the model.

[0048] For example, taking segmental defects of the mandible in oral and maxillofacial surgery as an example: First, tomographic image data is acquired. For instance, dual-source CT is used to scan the patient's mandibular region, with scanning parameters set as follows: tube voltage 120kV, tube current 200mAs, slice thickness 0.6mm, and reconstruction interval 0.3mm, obtaining a continuous tomographic image sequence containing the mandibular body, ramus, and condyle. The tomographic image data is obtained by medical imaging scanning of the region including the target biological tissue structure. Then, the DICOM format tomographic images are imported into medical image processing software, the bone tissue grayscale threshold is set to 226-1800HU, and the mandible is automatically segmented using a region growing algorithm; for areas where tumor invasion causes discontinuity in the bone cortex, the defect edge contour is supplemented by manual layer-by-layer tracing; after segmentation, a two-dimensional contour line of the outer surface of the mandible is generated on each layer, with a contour line spacing of 0.6mm, and adjacent layers are kept continuous by linear interpolation. Then, the moving cube algorithm was used to reconstruct the 2D contour lines into 3D isosurfaces, generating an initial triangular mesh model. The Laplacian smoothing algorithm was applied to remove step artifacts, preserving the original morphology of the defect edges. A mesh simplification algorithm was used to control the number of model patches to around 15,000, reducing the computational burden while maintaining anatomical details. The optimized 3D mandibular mesh model had an average surface error ≤0.15mm and a maximum defect edge error ≤0.25mm, meeting the accuracy requirements for clinical implant design. This model not only includes the morphology of the defect area but also preserves the trajectory of the inferior alveolar nerve canal, providing anatomical information for feature extraction.

[0049] For example, taking the repair of skull base bone defects in neurofibromatosis type I as an example: A 16-year-old female patient with neurofibromatosis type I underwent a thin-slice CT scan of the skull, with the scanning range from the top of the skull to the lower edge of the mandible, a slice thickness of 0.5 mm, and a slice interval of 0.3 mm, obtaining high-resolution image data including complex deformities such as defects of the greater wing of the sphenoid bone and temporal lobe herniation into the orbit. In this scenario, due to the complete absence of the greater wing of the sphenoid bone, traditional threshold segmentation could not directly extract the defect boundary. Therefore, a mirror segmentation method was adopted, using the midsagittal plane as the plane of symmetry, to mirror the morphology of the healthy side of the sphenoid bone onto the affected side, and the virtual anatomical contour of the defect area was determined by an image registration algorithm; at the same time, based on adjacent bony landmarks such as the squamous part of the temporal bone and the zygomatic process of the frontal bone, the defect edge was manually corrected to obtain a closed skull base defect contour line. Then, the segmented two-dimensional contour sequence was imported into Geomagic Wrap software, and a high-precision three-dimensional surface model was generated using the Poisson surface reconstruction algorithm. For the intricate structures such as the medial and lateral plates of the pterygoid process of the sphenoid bone, a local mesh refinement strategy was adopted to achieve a minimum feature resolution of 0.3 mm in the model. The resulting three-dimensional curved surface model of the skull base accurately presents the three-dimensional morphology of the greater pterygoid defect and the surrounding bony supporting structures, laying the foundation for the precise deployment of the subsequent titanium mesh restoration.

[0050] In this embodiment, by acquiring tomographic image data, performing image segmentation to extract contour information, and generating a mesh model composed of polygonal patches based on the contour information using a three-dimensional reconstruction algorithm, a link is established from the original medical image data to a high-precision three-dimensional surface model, ensuring that the three-dimensional surface model can correctly represent the surface morphology of the target biological tissue structure.

[0051] Based on any of the above embodiments, the step of extracting feature information for constraining the unfolded shape from the three-dimensional surface model includes: The three-dimensional surface model is input into the trained recognition model to obtain feature information, which includes at least one of bone defect boundary, anatomical axis and contour line.

[0052] It should be noted that the recognition model is a deep learning network trained on labeled data, capable of automatically identifying and locating anatomical features from a 3D mesh. Unlike traditional image segmentation networks, this embodiment employs a neural network architecture that can directly process 3D geometric data, avoiding the information loss caused by rendering 3D surfaces as 2D views.

[0053] Optionally, the recognition model is a 3D deep learning network based on an improved PointCNN architecture, which can directly process unstructured 3D mesh data. The model uses the vertex coordinates and normal vectors of the 3D surface model as input features, extracts local geometric features through dynamic graph convolutional layers, and fuses global contextual information through a self-attention mechanism.

[0054] Optionally, the recognition model is a three-dimensional deep learning network based on the PointNet++ architecture, which uses the vertex coordinates, curvature, and shape diameter function of the orbital surface mesh as input features to automatically identify fracture boundaries and key anatomical landmarks of the orbital wall.

[0055] The feature information includes, but is not limited to, three types of geometric constraints with clear anatomical significance that play a decisive role in the unfolded shape: the boundary of the bone defect determines the cutting contour of the barrier membrane; the root axis prevents the root surface area from twisting during the flattening process; and the alveolar ridge crest line serves as a horizontal reference to ensure the correspondence between the two-dimensional drawing and the intraoral anatomical orientation.

[0056] Bone defect boundary: The continuous curve at the junction of the periodontal bone defect area and healthy bone tissue serves as the anatomical basis for the cutting edge of the barrier membrane. The accurate extraction of this boundary directly determines whether the barrier membrane coverage is sufficient.

[0057] Anatomical axis: A straight line or curve that characterizes the spatial orientation of the tooth root. During unfolding, it acts as a rigid constraint to prevent non-uniform stretching of the curved area corresponding to the tooth root.

[0058] Outline: refers to other anatomical curves that have morphological reference significance besides the boundary of the defect, such as the alveolar ridge crest line, the lower border line of the mandible, and the infraorbital border line.

[0059] In this embodiment, by inputting a three-dimensional curved surface model into a trained recognition model, at least one of the following—bone defect boundary, anatomical axis, and contour line—is automatically identified and extracted as feature information, achieving efficient extraction of key anatomical features and avoiding the subjectivity problem of manual annotation.

[0060] Based on any of the above embodiments, the step of mapping the surface mesh of the three-dimensional surface model to a two-dimensional plane of equal area based on the feature information and the surface area of ​​the three-dimensional surface model includes: Construct an energy function, which includes an area conservation term to maintain the consistency of the area before and after unfolding, and a constraint term to map the feature information to the corresponding position or shape of the two-dimensional plane. The energy function is optimized using an iterative algorithm, and the coordinates of each node in the surface mesh on the two-dimensional plane are adjusted. The iteration algorithm stops when it meets the preset convergence condition, so that the error rate between the total area of ​​the two-dimensional plane and the surface area of ​​the three-dimensional curved surface model is lower than the first preset threshold.

[0061] It should be noted that the energy function is a mathematical model that transforms the surface unfolding problem into a numerical optimization problem. A lower energy function value indicates better area conservation and a higher degree of satisfaction of feature constraints in the unfolded result. By minimizing the energy function, a two-dimensional unfolding result that simultaneously satisfies area conservation and feature preservation can be obtained.

[0062] Area conservation term: A sub-term in the energy function used to penalize changes in area before and after expansion. It is generally expressed in the form of a ratio of area of ​​patches, a logarithmic ratio of areas, or the square of the area difference.

[0063] Constraints: Sub-terms in the energy function used to penalize feature information for deviating from the target position or shape.

[0064] Preset convergence criteria: Criteria for terminating iterative optimization. Preset convergence criteria include energy function change threshold, node displacement threshold, or maximum number of iterations. This embodiment does not impose any restrictions on these criteria.

[0065] First preset threshold: The clinically acceptable upper limit of the area error rate. Optionally, the first preset threshold can be set to 2%, within which the barrier membrane or titanium mesh blank can achieve satisfactory adhesion without intraoperative secondary trimming.

[0066] For example, consider the reconstruction of a defect in the body of the mandible: First, construct the energy function in the following form: E=α·Earea +β·E constraint ; Among them, E area The area conservation term is defined as the sum of the squares of the ratios of the areas of all triangular faces before and after unfolding; E constraint The constraints include feature curve position constraints (mapping the defect boundary to its corresponding position in the two-dimensional plane) and feature curve direction constraints (maintaining the horizontal orientation of the lower edge of the mandible in the two-dimensional plane). The weighting coefficients α and β are set to 0.75 and 0.25, respectively, with area conservation as the primary optimization objective. The energy function includes an area conservation term to maintain area consistency before and after unfolding, and constraint terms to map the feature information to its corresponding position or shape in the two-dimensional plane.

[0067] Next, the energy function was iteratively optimized using the L-BFGS quasi-Newton method. During initialization, conformal mapping was used to roughly unfold the 3D mandibular mesh onto a 2D plane. In each iteration, the gradient of the energy function with respect to the 2D node coordinates was calculated, and the node positions were updated along the gradient descent direction. To avoid mesh flipping, local remapping was performed after each iteration to ensure that the normals of all triangular faces were consistent. The iteration process was repeated 180 times.

[0068] In this scenario, the preset convergence conditions are: (1) the relative change in energy function between two adjacent iterations is less than 1e-5; (2) the maximum node displacement is less than 0.001mm. When iterating to the 156th iteration, the energy function decline curve tends to flatten, both convergence conditions are met simultaneously, and the iteration terminates. The first preset threshold is set to 2.0%. The calculated surface area of ​​the three-dimensional curved surface model of the mandibular defect area is 16.8cm. 2 The total area of ​​the unfolded two-dimensional plane is 16.95 cm². 2 The area error rate was 0.89%, which is lower than the preset threshold. Meanwhile, the length retention errors of the anterior and posterior edge boundary curves of the defect were 0.3% and 0.5%, respectively, and the lower edge contour line of the mandible remained approximately horizontal in the two-dimensional plane (angle deviation 2.1°), verifying the effectiveness of the constraint terms.

[0069] For example, consider the repair of a large area of ​​skull defect: The surface area of ​​the three-dimensional curved model of the skull defect area is 32.5 cm². 2 When constructing the energy function, except for the area conservation term E... area and boundary constraint term E constraint In addition, a regularization term E was added. smooth This is used to suppress excessive local distortion of the mesh during the optimization process. The three weights are set to α=0.6, β=0.3, and γ=0.1, respectively.

[0070] The iterative algorithm uses the conjugate gradient method, with an initial step size of 0.01, which is adaptively adjusted. The convergence conditions are set as follows: (1) the gradient norm is less than 1e-4; (2) the area error change is less than 0.005mm. 2 The iterations converged after a total of 210 iterations.

[0071] The total area of ​​the unfolded two-dimensional plane is 32.98 cm². 2 The area error rate was 1.48%, which is lower than the first preset threshold (i.e., 2.0%). The length error of the skull defect boundary curve before and after unfolding was 1.2%, and the length error of the temporal line projection curve before and after unfolding was 0.9%, both of which meet the accuracy requirements.

[0072] In this embodiment, by constructing an energy function that includes area conservation terms and feature constraint terms, and using an iterative algorithm to optimize the solution, under the premise of satisfying the mapping of feature information to the corresponding position or shape of the two-dimensional plane, the error rate between the total area of ​​the two-dimensional plane and the surface area of ​​the three-dimensional curved surface model is lower than the first preset threshold, thus achieving high-precision equal area mapping and ensuring that the unfolded two-dimensional plane is conserved with the original three-dimensional curved surface on the global area scale.

[0073] Based on any of the above embodiments, generating a two-dimensional unfolded diagram for implant manufacturing based on the two-dimensional plane includes: The region in the two-dimensional plane corresponding to the feature information is subjected to local mesh densification processing; Based on the original arc length of the feature information on the three-dimensional surface model, the positions of the mesh nodes in the encrypted region are adjusted so that the percentage difference between the length of the curve corresponding to the feature information on the two-dimensional plane and the original arc length is lower than a second preset threshold, thereby obtaining the two-dimensional unfolded diagram; the percentage difference is determined according to the ratio of the difference between the arc length of the spatial curve of the feature information on the three-dimensional surface model and the length of the corresponding planar curve on the two-dimensional plane.

[0074] It should be noted that local mesh refinement is a technique that increases the mesh density for specific anatomical regions in a 2D unfolded plot to improve the geometric representation of those regions and the accuracy of subsequent deformation control. The selection of the refined region coincides with the distribution of feature information.

[0075] Original arc length: The spatial path length of the characteristic curve on the 3D surface model, serving as a reference value for evaluating whether stretching or compression occurs during the unfolding process. The spatial arc length is calculated by accumulating the 3D Euclidean distances between adjacent vertices.

[0076] Percentage of difference: The relative rate of change of the length of the characteristic curve before and after unfolding. The smaller the percentage of difference, the better the length preservation of the characteristic curve and the lower the degree of morphological distortion.

[0077] Second preset threshold: The upper limit of the percentage difference in characteristic curve length. Optionally, the second preset threshold is set to 3.0%, within which doctors cannot perceive significant morphological differences.

[0078] For example, periodontal furcation lesions will be used as an example for illustration: The 3D surface model depicts a grade III furcation lesion on the buccal side of the mandibular first molar, with the defect extending into the furcation dome region. This region is prone to mesh stretching distortion during 2D unfolding due to drastic curvature changes. The model focuses on the furcation region (the characteristic area has a projected area of ​​approximately 0.35 cm²). 2 The 2D mesh was locally refined from an initial global size of 0.5mm to 0.15mm × 0.15mm, resulting in approximately 320 refined sub-mesh elements. For the bone defect boundary (original 3D perimeter 12.2mm), four layers of refined transition zones were extended along both the inner and outer sides of the boundary line, covering a 0.8cm radius around the defect. A 0.25mm refined mesh was used for the root surface area, while the ordinary alveolar bone area maintained a standard 0.6mm mesh density. A red-green quadtree subdivision strategy was employed for mesh refinement to ensure a smooth transition in mesh size at the boundary between refined and unrefined areas.

[0079] Then, using the characteristic curve at the top of the root bifurcation dome (original spatial arc length 5.3 mm) and the boundary curve of the bone defect (original spatial arc length 12.2 mm) as length references, the Laplace mesh deformation algorithm is used to locally relax and reposition the node coordinates in the encrypted area. The specific steps are: (1) take the original arc length of the three-dimensional characteristic curve as the target length; (2) establish the Laplace coordinate constraints of the two-dimensional mesh nodes; (3) with the goal of minimizing the length error, solve the node displacement vector through a sparse linear system; (4) iteratively adjust until the length error converges.

[0080] The second preset threshold was set to 3.0%. After optimization, the length of the characteristic curve at the top of the root bifurcation dome on the two-dimensional plane was 5.45 mm, with a difference percentage of 2.8%; the perimeter of the bone defect boundary curve on the two-dimensional plane was 12.48 mm, with a difference percentage of 2.3%; the length of the mesial root axis remained completely consistent before and after unfolding (14.0 mm), with a difference percentage of 0%. The length difference percentage of all characteristic curves was less than 3.0%, meeting the conformal optimization requirements.

[0081] The percentage difference is determined based on the ratio of the difference between the arc length of the spatial curve on the three-dimensional surface model and the length of the corresponding planar curve on the two-dimensional plane, according to the feature information. After optimization, the visual similarity between the root bifurcation dome shape and the course of the bone defect edge in the two-dimensional unfolded image and the three-dimensional model is ≥96%.

[0082] For example, consider the repair of orbitozygomatic complex fractures: The infraorbital margin and zygomatic alveolar ridge are key areas for morphological restoration, exhibiting significant curvature changes. The mesh size was increased to 0.2mm × 0.2mm in a 1.5cm strip along the infraorbital margin and to 0.25mm × 0.25mm in the raised area on the zygomatic bone surface. Then, based on the original arc length of the feature information on the 3D surface model, the positions of the mesh nodes in the refined region are adjusted. In this scenario, using the complete contour line of the infraorbital rim (original arc length 52.3 mm) and the crest line of the zygomatic alveolar ridge (original arc length 28.7 mm) as length benchmarks, the node positions are optimized using a nonlinear least squares method. The optimization objective function includes both arc length conservation terms and curvature preservation terms. The optimized two-dimensional length of the infraorbital rim contour line is 53.1 mm, with a difference percentage of 1.5%; the two-dimensional length of the zygomatic alveolar ridge crest line is 29.2 mm, with a difference percentage of 1.7%. Both are below the second preset threshold (i.e., 3.0%), ensuring precise alignment of the restoration edge with the bone window.

[0083] In this embodiment, by performing local mesh densification on the region corresponding to the feature information in the two-dimensional plane, and adjusting the position of the mesh nodes in the densified region based on the original arc length of the feature information on the three-dimensional curved surface model, the percentage difference between the length of the feature curve on the two-dimensional plane and the original arc length is lower than the second preset threshold. This effectively suppresses the morphological distortion of key anatomical regions such as root bifurcation and bone defect edges during the unfolding process, ensuring that the two-dimensional unfolded diagram retains the important morphological features of the three-dimensional curved surface, and improving the fitting accuracy between the implant and the defect area.

[0084] Based on any of the above embodiments, the method further includes: Obtain a local shape adjustment instruction for the two-dimensional unfolded diagram; the local shape adjustment instruction includes deformation parameters and the target area to be adjusted in the two-dimensional unfolded diagram; While keeping the total area change of the two-dimensional unfolded diagram below a third preset threshold, the grid node coordinates of the target region are stretched or shrunk based on the deformation parameters to obtain a locally adjusted two-dimensional unfolded diagram.

[0085] It should be noted that the local shape adjustment command is an operation command used by clinicians to interactively modify the automatically generated 2D unfolded diagram through a graphical user interface. This command includes: the area to be adjusted (indicating its spatial location) and deformation parameters (indicating the adjustment range and method).

[0086] Deformation parameters: numerical indicators that quantify the intensity of local adjustment, and can be expressed in various forms such as tensile coefficient, contraction coefficient, bending angle, and displacement vector. The reasonable range of deformation parameters is constrained by the threshold of total area change.

[0087] Target region: The continuous mesh region selected for local deformation in the 2D unfolded diagram. The region boundary is maintained with C1 continuity through curve fitting to avoid visually abrupt geometric changes after adjustment.

[0088] The third preset threshold: the maximum allowable change in total area during interactive adjustments. The significance of setting this threshold is that local morphological optimization must not violate the global area conservation. Optionally, it can be set to 1.0%, allowing fine-tuning for specific clinical needs, but controlling the total area drift within 1%.

[0089] For example, consider the clinical need for intraoperative fine-tuning of cranial titanium mesh prostheses: The cranial titanium mesh prosthesis, generated based on a 2D unfolded image, underwent equal-area mapping and shape-preserving optimization during the preoperative design phase. Intraoperatively, a local gap of approximately 1.5mm was found between the posterior edge of the titanium mesh and the edge of the bone window. Since re-scanning the design was not possible during surgery, the surgeon accessed the interactive adjustment module in the operating room via a mobile terminal. Local shape adjustment instructions for the 2D unfolded image were obtained: the surgeon opened the patient's 2D unfolded image on a tablet, selected an arc-shaped area approximately 2cm long at the posterior edge of the titanium mesh, and input a displacement vector of "move forward 1.5mm". While keeping the total area change of the 2D unfolded image below a third preset threshold (i.e., 1.0%), the mesh node coordinates of the target area were stretched based on deformation parameters. A mesh deformation algorithm based on radial basis functions was used to smoothly diffuse the local displacement to the surrounding area, ensuring boundary continuity. Real-time calculation showed a total area change of 0.12%, below the threshold. The adjusted 2D unfolded image was transmitted in real-time to a laser engraving machine in the operating room, where a finely adjusted titanium mesh was re-cut on-site and implanted to achieve zero-gap fit with the edge of the bone window.

[0090] In this embodiment, by acquiring local shape adjustment instructions containing deformation parameters and target area, and performing stretching or shrinking calculations on the grid node coordinates of the target area while keeping the total area change of the two-dimensional unfolded diagram below a third preset threshold, clinicians are given fine-tuning authority, allowing optimization of local morphology based on intraoperative observation or special case needs. At the same time, area monitoring ensures that local adjustments do not disrupt global area conservation, thus combining algorithm calculation with clinical expert experience.

[0091] Based on any of the above embodiments, the two-dimensional unfolded drawing is a digital graphic file, which includes outlines and at least one positioning reference mark; the method further includes: The digital graphic file is transmitted to the cutting device, so that the cutting device determines the processing origin according to the positioning reference mark, generates a cutting path according to the contour line, and cuts the implant blank according to the cutting path to obtain an implant that matches the shape of the three-dimensional curved surface model.

[0092] It should be noted that digital graphic files are files that carry geometric information and process attributes of two-dimensional unfolded diagrams. Common formats include DXF, DWG, PDF, SVG, and STEP.

[0093] Contour line: A closed curve that defines the cutting boundary of the implant in a two-dimensional unfolded diagram. The accuracy of the contour line determines the final fit between the implant and the boundary of the three-dimensional curved surface.

[0094] Positioning reference mark: A special graphic element with a specific geometric shape (which can be a combination of circles, crosses or L-shapes) added to the two-dimensional unfolded drawing, used by the processing equipment to automatically identify the workpiece placement position, orientation and scaling ratio.

[0095] Cutting equipment: CNC machining equipment that can automatically cut planar materials based on digital graphic files, including but not limited to laser engraving machines, CNC milling machines, waterjet cutting machines, ultrasonic cutting knives, and die-cutting machines.

[0096] Implants: Medical device products that are ultimately implanted into the human body and conform to the morphology and structure of the target biological tissue. Optionally, implant types include a variety of personalized implantable repair products such as oral barrier membranes, cranial titanium mesh, jawbone reconstruction plates, orbital prostheses, orthopedic bone plates, and joint prosthesis positioning guides.

[0097] For example, the laser engraving fabrication of a personalized periodontal collagen barrier membrane will be used as an example for illustration: The two-dimensional unfolded image of the periodontal barrier membrane, after iso-area mapping, conformal optimization, and interactive adjustments, is exported as a DXF format digital graphic file. This file contains the following key layers and primitives: (1) Contour layer: Stores the two-dimensional projection of the boundary curve of bone defect, using closed polyline primitives, with a line width logic value of 0.1mm, the curve is composed of 156 fitting nodes, and the maximum chord height error is 0.02mm; (2) Positioning reference layer: Store three circular markers with a diameter of 0.5 mm. The center coordinates of the circle correspond to the two-dimensional mapping positions of the highest point of the alveolar crest, the root bifurcation vertex and the lowest point of the defect edge in the three-dimensional model, respectively. The markers are distributed in an asymmetrical L-shape with spacings of 8.2 mm, 6.5 mm and 10.3 mm, respectively, to ensure unique alignment. (3) Marker layer: Stores text information such as patient ID, surgery date, barrier membrane thickness (0.3mm) and material (type I collagen), encoded in the form of QR code.

[0098] The DXF file is sent to the laser engraving workstation in the dental digital processing center via the hospital's intranet. The equipment control software reads the DXF file, automatically parses the layer information, and maps it to the corresponding processing parameters. Then, the laser engraving machine performs the following automated steps: (1) Coordinate system alignment: The CCD vision system automatically identifies the three positioning reference marks in the DXF file, takes the geometric center of the first mark point as the XY plane machining origin (0,0), and uses the line connecting the second mark point and the origin to determine the positive direction of the X axis, thus completing the rigid registration between the workpiece coordinate system and the machine coordinate system. (2) Path planning: Automatically generate continuous cutting trajectory along the contour line multi-segment line, automatically insert deceleration control at corners, and use micro-segment fitting for curve segments with a fitting tolerance of 0.01mm; (3) Process parameter mapping: Automatically set laser power (30W for collagen film material), scanning speed (50mm / s), scanning interval (0.05mm), auxiliary air pressure (0.2MPa) and cooling delay (2s) according to the layer attributes; (4) Cutting execution: The dry collagen barrier membrane blank with a thickness of 0.3mm and a size of 100mm×80mm is adsorbed and fixed on the worktable, and the equipment performs a single penetration cut along the planned path; (5) Quality inspection: After cutting, the online optical inspection instrument automatically measures the outline size of the barrier film and the position of the positioning mark. The deviation from the DXF design value is ≤0.05mm, which is considered qualified.

[0099] Ultimately, a personalized periodontal barrier membrane was obtained that precisely matches the three-dimensional curved surface morphology of the buccal bone defect of the mandibular first molar.

[0100] In this embodiment, by generating a digital graphic file containing contour lines and at least one positioning reference mark from a two-dimensional unfolded drawing, and transmitting it to the cutting device so that it can determine the processing origin based on the positioning reference mark, generate the cutting path based on the contour line, and perform cutting, the process from three-dimensional design to finished implant is realized, improving the manufacturing efficiency and precision of personalized implants and ensuring that the final implant is adapted to the shape of the three-dimensional curved surface model.

[0101] Furthermore, based on the same concept as the three-dimensional surface unfolding method for implant manufacturing described above, this application also provides an equal-area conformal two-dimensional unfolding method for three-dimensional curved surfaces of periodontal bone defects, which can achieve accurate conversion of complex three-dimensional curved surfaces to two-dimensional planes, providing technical support for the fabrication of personalized barrier membranes, and is applicable to the design and fabrication of personalized periodontal tissue regeneration barrier membranes.

[0102] In this embodiment, taking the case of "buccal bone defect of the mandibular first molar" as an example, the method for equal-area conformal two-dimensional unfolding of the three-dimensional curved surface of the periodontal bone defect includes: 1. Acquisition of three-dimensional curved surface model: Image data of the periodontal region of the patient is acquired by cone-beam CT (CBCT) scanning. After noise reduction and segmentation processing, a three-dimensional curved surface model of the periodontal bone defect and surrounding tissues is reconstructed. The model includes the edge contour of the bone defect, root surface morphology and alveolar bone anatomical landmarks. Case matching example (buccal bone defect of mandibular first molar): Key parameters of the reconstructed model:

[0103] Model validation: Point cloud comparison confirmed that the reconstructed model matched the original CBCT image with a degree of ≥95%, and the deviation of the bone defect edge contour was ≤0.2mm, meeting the accuracy requirements for subsequent unfolding.

[0104] 2. Feature Constraint Extraction: Based on a deep learning model, feature recognition is performed on the three-dimensional surface model, and the bone defect boundary curve, tooth root axis, and alveolar ridge crest line are extracted as mapping constraint benchmarks to construct a topological association system of feature points-curves-surfaces; Case matching example (feature extraction of mandibular first molar): Extraction results and topological association:

[0105] Extraction accuracy verification: The deviation between the feature curve and the manual annotation by the expert is ≤0.3mm, which meets the accuracy requirements of the mapping constraint.

[0106] 3. Equal Area Mapping Calculation: The Lambert equal area projection algorithm is adopted, and the feature constraints extracted in step 2 are used as boundary conditions to establish a one-to-one correspondence between the three-dimensional curved surface mesh and the two-dimensional planar mesh. The energy minimization iterative solution is used to ensure that the error rate between the total area of ​​the unfolded two-dimensional plane and the area of ​​the original three-dimensional curved surface is ≤2%. Case matching example (major first molar mapping calculation): Key calculation data and results:

[0107] 4. Conformal optimization: For the irregular shape of periodontal bone defects, multi-scale mesh generation technology is used to refine the mesh in key areas such as the root bifurcation zone and subosseous pocket, ensuring that the characteristic curve distortion rate of the unfolded two-dimensional plane is ≤3%; Case adaptation example (orthodontic optimization of mandibular first molar): Optimization and distortion verification:

[0108] Overall verification after optimization: The morphology of all anatomical features (root bifurcation, bone defect edges) in the 2D unfolded diagram has a similarity of ≥97% with the 3D model, which meets the requirements of personalized barrier membrane design.

[0109] 5. Output of 2D unfolding results: Generate a 2D unfolded diagram containing feature marker lines, clipping boundaries and positioning reference points, and simultaneously output area verification report and morphological similarity evaluation data. The 2D unfolded diagram can be directly adapted to laser engraving process parameters.

[0110] Case fitting example (outline of mandibular first molar): Output results and process compatibility:

[0111] Clinical application: Doctors use the positioning reference points on the unfolded diagram to precisely attach the sculpted barrier membrane to the buccal bone defect area of ​​the patient's mandibular first molar, thus achieving precise operation of guided tissue regeneration.

[0112] The three-dimensional surface model reconstruction in step 1 includes: performing tomographic segmentation of CBCT images with a slice thickness of 0.1-0.3 mm, constructing a polygonal mesh model using the moving cube algorithm, and smoothing the surface to ensure that the number of triangular facets on the model surface reaches 5,000-20,000, thus ensuring that the detail resolution of bone defects is ≥0.05 mm.

[0113] The deep learning model described in step 2 adopts an improved U-Net architecture, using 300 annotated three-dimensional models of periodontal bone defects as the training set. The feature recognition accuracy is ≥92%, and the positioning error of the extracted feature curve is ≤0.1mm.

[0114] Step 3, the energy minimization iterative solution, includes: constructing the area-conserving energy function E. area and boundary constraint energy function E constraint The algorithm is optimized iteratively using the gradient descent method. The iteration terminates when the area error between two consecutive calculations is ≤0.01mm. 2 The number of iterations should be controlled between 50 and 200.

[0115] The multi-scale meshing technique described in step 4 includes: using a fine mesh with a side length of 0.2-0.3 mm in areas with bone defect depth > 3 mm, and using a coarse mesh with a side length of 0.5-0.8 mm in areas with normal bone tissue. The mesh transition areas are smoothly connected through gradient adjustment.

[0116] In addition, it also includes a dynamic correction module: receiving adjustment instructions input by clinicians, stretching or shrinking local areas of the two-dimensional unfolded diagram, displaying the area change in real time during the adjustment process, and ensuring that the total area difference after adjustment is ≤1%.

[0117] This application also provides a digital design system for periodontal barrier membranes, which integrates the above-mentioned three-dimensional surface unfolding method and includes an image processing module, a feature extraction module, a mapping calculation module, an optimization output module, and a doctor interaction module. It can output DXF or SVG format files adapted to laser engraving equipment.

[0118] Specifically, 1. Acquisition of three-dimensional curved surface model: CBCT images of the patient's periodontal region are acquired with an image resolution of not less than 0.1 mm. Bone tissue and soft tissue are separated by a threshold segmentation algorithm. The three-dimensional curved surface model is reconstructed using a moving cube algorithm. After optimization by Poisson surface reconstruction, redundant triangular facets are removed, and the details of bone defect edges and root surfaces are preserved. 2. Feature Constraint Extraction: Based on the pre-trained U-Net deep learning model, key features such as the boundary curve of bone defects, the root midline, and the alveolar ridge crest line are automatically identified and extracted. The feature curves are fitted by the RANSAC algorithm, a coordinate system based on the root is established, and the coordinates of the feature points are standardized. 3. Equal Area Mapping Calculation: The Lambert equal area projection algorithm is introduced, using the characteristic curve as a fixed boundary to construct the topological relation matrix of the 3D surface mesh. The energy function is defined as E=αE. area +βE constraint (Where α and β are weighting coefficients, α=0.7, β=0.3), the area change rate of each triangular facet is controlled within ±1% by using the conjugate gradient method iterative solution, and the total area difference is ≤2%; 4. Conformity Preservation Optimization: A multi-scale meshing strategy is employed to refine the mesh in complex areas such as root bifurcation lesions and subosseous pockets. Mesh node positions are adjusted using Bézier curves to ensure that the curvature change rate of the unfolded feature curves is ≤3%. Boundary locking technology is used to prevent morphological distortion in the model boundary regions. 5. Output and Interactive Adjustment: Generates a 2D unfolded diagram containing feature markers, cutting lines, and positioning holes, outputting a file format compatible with laser engraving equipment. Doctors can adjust local areas through the interactive interface; the system calculates area changes in real time and prompts for the adjustment range, ensuring personalized clinical needs.

[0119] In the specific implementation: 1. Data acquisition and model reconstruction: Patients with periodontal bone pocket defects were selected, and CBCT scans were used to obtain image data with a slice thickness of 0.15mm. Software was used for segmentation, setting the bone tissue threshold to 226-3071 HU, and a three-dimensional curved surface model was reconstructed. After optimization with Geomagic Wrap, the number of triangular patches was approximately 12,000, and the model accuracy error was ≤0.08mm. 2. Feature extraction verification: The U-Net model was trained using 300 labeled periodontal bone defect models. Feature extraction was performed on 20 models in the test set. The average boundary curve positioning error was 0.07mm, and the feature recognition accuracy was 94.3%, meeting clinical accuracy requirements. 3. Mapping algorithm testing: Unfolding tests were conducted on 5 different types of bone defect models (infrabone pockets, bifurcation lesions, etc.). The average area error rate of the equal-area mapping algorithm was 1.2%, significantly lower than the traditional stretching unfolding method (8.7%); the average characteristic curve distortion rate was 2.1%, better than the simple isoangular mapping method (4.8%). 4. Clinical compatibility verification: The unfolding results were imported into a laser engraving machine to create a polylactic acid barrier membrane. The membrane was applied in 10 clinical cases. The average fit score between the barrier membrane and the bone defect was 4.6 points (out of 5). The operation time was shortened by 35% compared with the traditional method, and the bone regeneration rate was increased by 28% compared with the control group 6 months after the operation.

[0120] In this embodiment, the uniform-area conformal unfolding of the periodontal bone defect surface is achieved, reducing the area error rate and morphological distortion rate. Integrating deep learning and anatomical feature constraints, it adapts to complex defect types such as subosseous pockets and root bifurcation lesions, demonstrating strong universality. The output file directly interfaces with laser engraving technology, enabling precise digital fabrication of the barrier membrane, reducing the difficulty of operation for dentists and shortening surgical time. It can be integrated into periodontal barrier membrane design systems, providing technical support for digital oral healthcare and promoting the standardization and precision development of guided tissue regeneration techniques.

[0121] Based on the methods described in any of the above embodiments, this application also provides, as follows: Figure 2 The diagram shows the structure of an electronic device. Figure 2 At the hardware level, the electronic device includes a processor, an internal bus, a network interface, memory, and non-volatile memory, and may also include other hardware required for business operations. The processor reads the corresponding computer program from the non-volatile memory into memory and then runs it to implement the methods described in any of the above embodiments.

[0122] Based on the methods described in any of the above embodiments, this application also provides a computer storage medium storing a computer program, which, when executed by a processor, can be used to perform the methods described in any of the above embodiments.

[0123] Based on the methods described in any of the above embodiments, this application also provides a computer program product, which includes one or more computer programs or instructions. The computer program or instructions may be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. When executed by a processor, the computer program implements the methods described in any of the above embodiments.

[0124] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0125] In addition, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0126] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0127] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application. It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0128] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0129] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

Claims

1. A three-dimensional surface unfolding method for implant manufacturing, characterized in that, The method includes: Obtain a three-dimensional surface model of the target biological tissue structure; Extract feature information for constraining the unfolded shape from the three-dimensional surface model; Based on the feature information and the surface area of ​​the three-dimensional surface model, the surface mesh of the three-dimensional surface model is mapped to a two-dimensional plane with equal area; Based on the two-dimensional plane, a two-dimensional unfolded diagram for implant manufacturing is generated.

2. The method according to claim 1, characterized in that, The acquisition of the three-dimensional surface model of the target biological tissue structure includes: Acquire tomographic image data; the tomographic image data is obtained by performing medical imaging scans on a region including the target biological tissue structure. The tomographic image data is segmented to extract the contour information of the target biological tissue; Based on the contour information, a mesh model composed of polygonal patches is generated using a 3D reconstruction algorithm; The mesh model is defined as the three-dimensional surface model.

3. The method according to claim 1, characterized in that, The extraction of feature information from the three-dimensional surface model to constrain the unfolded shape includes: The three-dimensional surface model is input into the trained recognition model to obtain feature information, which includes at least one of bone defect boundary, anatomical axis and contour line.

4. The method according to claim 1, characterized in that, The step of mapping the surface mesh of the three-dimensional surface model to a two-dimensional plane of equal area based on the feature information and the surface area of ​​the three-dimensional surface model includes: Construct an energy function, which includes an area conservation term to maintain the consistency of the area before and after unfolding, and a constraint term to map the feature information to the corresponding position or shape of the two-dimensional plane. The energy function is optimized using an iterative algorithm, and the coordinates of each node in the surface mesh on the two-dimensional plane are adjusted. The iteration algorithm stops when it meets the preset convergence condition, so that the error rate between the total area of ​​the two-dimensional plane and the surface area of ​​the three-dimensional curved surface model is lower than the first preset threshold.

5. The method according to claim 1, characterized in that, The step of generating a two-dimensional unfolded diagram for implant manufacturing based on the two-dimensional plane includes: The region in the two-dimensional plane corresponding to the feature information is subjected to local mesh densification processing; Based on the original arc length of the feature information on the three-dimensional surface model, the positions of the mesh nodes in the encrypted region are adjusted so that the percentage difference between the length of the curve corresponding to the feature information on the two-dimensional plane and the original arc length is lower than a second preset threshold, thereby obtaining the two-dimensional unfolded diagram; the percentage difference is determined according to the ratio of the difference between the arc length of the spatial curve of the feature information on the three-dimensional surface model and the length of the corresponding planar curve on the two-dimensional plane.

6. The method according to claim 1 or 5, characterized in that, The method further includes: Obtain a local shape adjustment instruction for the two-dimensional unfolded diagram; the local shape adjustment instruction includes deformation parameters and the target area to be adjusted in the two-dimensional unfolded diagram; While keeping the total area change of the two-dimensional unfolded diagram below a third preset threshold, the grid node coordinates of the target region are stretched or shrunk based on the deformation parameters to obtain a locally adjusted two-dimensional unfolded diagram.

7. The method according to claim 1, characterized in that, The two-dimensional unfolded diagram is a digital graphic file, which includes outlines and at least one positioning reference mark; the method further includes: The digital graphic file is transmitted to the cutting device, so that the cutting device determines the processing origin according to the positioning reference mark, generates a cutting path according to the contour line, and cuts the implant blank according to the cutting path to obtain an implant that matches the shape of the three-dimensional curved surface model.

8. An electronic device, characterized in that, The electronic device includes: processor; Memory used to store processor-executable instructions; Wherein, when the processor invokes the executable instructions, it implements the method described in any one of claims 1-7.

9. A computer-readable storage medium, characterized in that, It stores computer instructions that, when executed by a processor, implement the steps of any of the methods described in claims 1-7.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method described in any one of claims 1-7.

Images