Artificial intelligence-based endocrown manufacturing method and system

Abstract

An artificial intelligence-based endocrown manufacturing method and system is provided. [Solution] The artificial intelligence-based manufacturing method for endocrowns includes the steps of acquiring an oral CBCT image and an intraoral scan model, synthesizing the oral CBCT image and the intraoral scan model to generate a composite model, inputting the composite model into a trained AI model and generating an endocrown model based on the composite model using the AI ​​model, conducting a simulation strength test on the endocrown model and regenerating the endocrown model if the stress in any region of the endocrown model is greater than a predetermined value, and conducting a simulation trial fitting test on the endocrown model that passes the simulation strength test.

Description

[Technical Field]

[0001] The present application relates to the technical field of dental crown manufacturing, and in particular to an artificial intelligence-based method and system for manufacturing endocrowns. [Background technology]

[0002] In the field of dental prosthodontics, as an important method of tooth restoration, endocrowns play a vital role in repairing dental defects because they can provide retention by utilizing the pulp cavity structure of remaining tooth tissue. Traditional endocrown manufacturing relies primarily on handcrafted manufacturing, which involves a series of manual processes by dentists, including tooth preparation, dental impression taking, dental model injection, and wax pattern creation.

[0003] Conventional methods for manufacturing endocrowns have the following significant drawbacks:

[0004] 1. In the handmade manufacturing process, many steps depend on the personal experience and operational skill of professional dental technicians, making it difficult to achieve precise and quantitative control.

[0005] 2. The handmade manufacturing process is tedious. For example, the wax pattern needs to be repeatedly modified, which lengthens the overall manufacturing cycle and reduces manufacturing efficiency. As a result, patients have to visit medical institutions multiple times, which increases time costs and negatively impacts the patient's medical experience.

[0006] 3. In the absence of precise digital design guidance, the restoration shape, its fit to the pulp cavity, etc. are determined solely by the dentist's subjective judgment, which is prone to human error and results in insufficient adaptation of the restoration to the patient's dental tissue, often requiring numerous intraoperative adjustments and extending the patient's treatment time.

[0007] Therefore, how to solve the above-mentioned drawbacks of the prior art is a problem to be considered and solved in this application. Summary of the Invention [Problem to be solved by the invention]

[0008] The objective of the present application is to provide an artificial intelligence based method and system for manufacturing endocrowns. [Means for solving the problem]

[0009] To achieve the above objectives, the technical solutions adopted in this application are as follows: An endocrown manufacturing method based on artificial intelligence includes: Step 1: acquiring oral CBCT images and intraoral scan models; Step 2: synthesizing the oral CBCT image and the intraoral scan model to generate a synthetic model; Step 3: inputting the synthetic model into a trained AI model, and generating an end-crown model based on the synthetic model by the AI ​​model; Step 4: performing a simulation strength test on the end-crown model, and regenerating the end-crown model if the stress in any region of the end-crown model is greater than a predetermined value; Step 5: conducting a simulated temporary fitting test on the end-crown model that has passed the simulated strength test; Step 6: adjusting the end-crown model based on the results of the simulated temporary fitting test; Step 7: inputting the adjusted end crown model into a printing device and printing and manufacturing a semi-finished end crown; and Step 8, wherein the semi-finished endocrown is subjected to a hardening treatment to produce a finished endocrown.

[0010] CBCT images and intraoral scan models are conventional technologies and are briefly described as follows. CBCT (cone beam computed tomography) imaging is a radiological imaging technology that uses X-ray 3D scans to provide information on hard tissues, such as jawbone morphology, nerve canal pathways, and maxillary sinus location. Intraoral scan models (intraoral scan data), on the other hand, are optical 3D surface scan technologies that can rapidly obtain the surface morphology of tooth crowns and gingiva without radiation, making them suitable for restoration design and orthodontic simulation. These two imaging modalities are synergistically applied through software integration. Combining crown data from intraoral scans with intraosseous structure data acquired via CBCT allows for the generation of a complete 3D model. Two-dimensional radiological imaging techniques, such as periapical radiographs and panoramic radiographs, can also be incorporated.

[0011] The method for training an AI model is as follows:

[0012] Collect three-dimensional jawbone data from CBCT scans and optical data from intraoral scans. Annotation of pulp chamber borders, margin design type (butt-joint / wrap-around), and material selection (e.g., lithium disilicate glass ceramics) by an oral specialist; We use a two-channel 3D CNN to process CBCT voxel data and intraoral scan point cloud data, and fuse features via a cross-attention mechanism. It integrates a mechanical simulation module, calculates stress distribution (e.g., von Mises stress) in real time, and constrains the rationality of the design. Optimize design accuracy and mechanical compatibility by performing supervised learning using past repair examples. Perform reinforcement learning optimization based on clinical success rates to provide manual corrective feedback for complex cases. The simulation strength test is mainly realized by finite element analysis (FEA), but a detailed description is omitted here. If the stress in any region of the end-crown model is greater than a predetermined value, the end-crown model is regenerated. Specifically, this is achieved by updating the surrogate model, annotating the defect area, and then performing calculations.

[0013] The simulated temporary fitting test is realized using virtual temporary fitting technology, which uses augmented reality technology to superimpose a model of the restoration (endocrown model) onto a digital model of the patient's mouth (synthetic model).

[0014] When a semi-finished endocrown is printed and manufactured, it can be printed using a composite resin.

[0015] During the curing process, dual-curing resin cements (e.g., All Ceram Core) are used, which utilize a dual mechanism of light-initiated rapid curing (triggered by light irradiation) and chemically initiated deep curing (continuous reaction in the dark) to fully cure the restoration (endocrown) in areas with insufficient light transparency, providing reliable adhesive strength.

[0016] In the prior art, a series of processes such as tooth preparation, dental impression taking, dental model injection, and wax pattern creation were performed manually by professional dental technicians. In this application, broadly speaking, the manufacturing time of an endocrown can be reduced by generating an endocrown model using a trained AI model and then printing the (semi-finished) endocrown using a printing device.

[0017] The method for manufacturing endocrowns provided by the present application does not rely on professional dental technicians and can achieve precise quantitative control.

[0018] According to the present application, the generated endocrown model can be rapidly manufactured with a 3D printer and restored in a single session, reducing costs for both the dentist and the patient and improving the patient's medical experience.

[0019] Before manufacturing the finished endocrown, the endocrown model is subjected to a simulated strength test and a simulated trial fitting test. This dual testing approach reduces the risk that the finished endocrown will not be suitable for the patient, avoids extended patient care time due to extensive intraoperative adjustments of the finished endocrown, and avoids a shortened service life of the finished endocrown due to lower than expected mechanical strength.

[0020] In dual testing, the simulated strength test is performed before the simulated temporary fitting test, eliminating the need to regenerate the end-crown model and restart the simulated temporary fitting test, thus avoiding extended manufacturing times for the finished end-crown.

[0021] Improvements in endocrown manufacturing methods can reduce manufacturing difficulty and increase manufacturing speed, allowing clinics and other medical facilities to independently manufacture finished endocrowns (especially for simple cases), eliminating the need for multiple patient visits.

[0022] According to the present application, by utilizing artificial intelligence in the design of endocrowns, errors caused by over-reliance on the empirical judgment of professional dental technicians can be avoided.

[0023] This application allows for the customization of endocrowns, which can be individually tailored to the specific condition of the diseased tooth. This allows for excellent retention and sufficient strength for future use while minimizing the removal of healthy tooth structure. Based on this approach, minimally invasive restoration of dental crowns can be achieved. This method preserves approximately 2 mm of the dentin ferrule and eliminates the need for placing a metal post in the root canal, thereby avoiding damage to the root during root canal treatment preparation. This maximizes the preservation of remaining dental tissue, reduces the risk of tooth fracture, and extends the useful life of the tooth.

[0024] According to another technical solution, in step 3, the step of generating an end crown model comprises: Step S1 of generating a holding portion to be held in the dental pulp cavity; Step S2 of generating an occlusion that engages with the opposing tooth and the adjacent tooth; and step S3 of creating a margin portion that engages the remaining tooth structure.

[0025] In step S1, a three-dimensional spiral retention structure is designed. Specifically, by utilizing the principle of biomimetic anchoring, a spiral groove structure with a gradient pitch is designed to ensure pull-out force while increasing interfacial bond strength. Furthermore, a porous reinforcement structure is designed. Specifically, a structure with various supports (e.g., honeycomb, columnar, or tree-like structures) is integrated within the retention portion, reducing the weight of the dental crown while improving its strength.

[0026] In step S2, multidimensional occlusal contact optimization is performed. Specifically, a virtual occlusion analysis system (e.g., Dental Occlusion Analyzer 4.0) is used to generate an optimal occlusal contact point distribution model through a machine learning algorithm, thereby avoiding interference between the occlusion area and other teeth.

[0027] In step S3, the margin contour is generated using 3D U-Net.

[0028] According to another technical solution, the outer wall of the holding portion forms an adhesive surface, and the surface roughness of the adhesive surface is in the range of 0.2 μm to 0.5 μm.

[0029] In this embodiment, the above-mentioned surface roughness refers to the surface roughness of the outer wall of the holding portion.

[0030] Surface roughness affects the mechanical interlocking effect between the adhesive and the substrate (retaining part). If the adhesive surface is not smooth, from a microscopic perspective, the adhesive will penetrate into the uneven structure on the adhesive surface and solidify, creating an anchoring effect and increasing adhesive strength.

[0031] Surface roughness affects the adhesive strength of the adhesive as described below.

[0032] If the surface roughness is too low (<0.2 μm), the surface of the adhesive surface will be too smooth, resulting in insufficient mechanical interlocking force and a significant decrease in adhesive strength. If the surface roughness is too high (>0.5 μm), the adhesive will not be able to fully penetrate the uneven surface structure, which will result in stress concentrations or microcracks, which will in turn reduce the adhesive strength.

[0033] Surface roughness also affects plaque control effectiveness as follows:

[0034] If the surface roughness is too high (>0.5 μm), bacteria can easily adhere to the adhesive surface, promoting plaque formation and increasing the risk of secondary caries. Too low a surface roughness (<0.2 μm) reduces the number of bacterial attachment sites, helping to control the rate of plaque formation.

[0035] In short, when the surface roughness is in the range of 0.2 μm to 0.5 μm, not only is the adhesive strength of the adhesive ensured, but the rate of plaque formation is also suppressed, achieving a balance between adhesive strength and plaque control.

[0036] According to another technical solution, in step S1 of step 3, the surface roughness of each region of the adhesive surface is identified based on a trained morphological feature-surface roughness association model.

[0037] The purpose of this example is to dynamically adjust the surface roughness of each region of the adhesive surface, one possible implementation of which is as follows: Based on the scan data, CAD software is used to analyze the shape features (e.g., grooves, sharp edges, flat areas) to identify the different requirements for adhesive strength and plaque control in each region.

[0038] Regarding the operational methodology of the morphological feature-surface roughness association model, input features are compared with clinical standard values ​​in a database via a deep neural network (e.g., SSA-BiLSTM model) and a recommended roughness range is output.

[0039] The training process of the morphological feature-surface roughness association model refers to the training process of the AI ​​model described above.

[0040] According to another technical solution, in step 4, the step of performing a simulation strength test includes: Step S1: applying a vertical load of magnitude A to the end-crown model; Step S2: applying a diagonal load of magnitude B to the end-crown model; Step S3: acquiring stress change data for each region of the end-crown model; and step S4 of comparing the maximum stress value during the simulated strength test with the tensile strength of the dentin, and regenerating the endocrown if the maximum stress value is greater than the tensile strength of the dentin.

[0041] In this embodiment, all parts of the endocrown model are first generated and then subjected to a simulation strength test together, thereby shortening the manufacturing process of the finished endocrown.

[0042] According to another technical solution, in step 6, the step of adjusting the end crown model includes: adjusting the convergence angle of the holding portion; adjusting the surface roughness of the holding portion; and adjusting the distribution of occlusal contact points on the occlusal portion.

[0043] The convergence angle may be considered as the tilt angle.

[0044] In this embodiment, the adjustment content in step 6 is clarified, and multiple adjustments are made to further ensure the adaptation of the endocrown model to the patient, thereby avoiding the extension of patient treatment due to extensive intraoperative adjustments to the finished endocrown.

[0045] There is further provided an artificial intelligence based endocrown manufacturing system, said system comprising: an acquisition module for acquiring oral CBCT images and intraoral scan models; a processing module that combines the intraoral CBCT image and the intraoral scan model to generate a combined model; a generation module that uses a trained AI model to generate an end-crown model based on the synthetic model; an analysis module that performs a simulation strength test on the end-crown model; and a verification module that performs a simulated temporary fitting test on the end-crown model that has passed the simulated strength test.

[0046] In this embodiment, the relevant explanations can be found in the corresponding parts of the above embodiments. Note that the specific structure of each module is not limited as long as it satisfies the requirements.

[0047] According to another technical solution, the end-crown model comprises: a holding portion that is held in the dental pulp cavity; an occlusal portion that engages with the opposing tooth and the adjacent tooth; and a margin portion that engages the remaining tooth structure.

[0048] In this embodiment, the endocrown model is divided into a holding part, an occlusal part, and a marginal part, so that different schemes can be selected to generate the endocrown model according to this division to meet different needs.

[0049] According to another technical solution, the outer wall of the holding portion forms an adhesive surface, and the surface roughness of the adhesive surface is in the range of 0.2 μm to 0.5 μm.

[0050] When the surface roughness is in the range of 0.2 μm to 0.5 μm, not only is the adhesive strength of the adhesive ensured, but the rate of plaque formation is also suppressed, achieving a balance between adhesive strength and plaque control.

[0051] As used herein, "including," "comprising," "having," and the like are open-ended terms meaning "including but not limited to."

[0052] As used herein, unless otherwise specified, terms generally have their ordinary meaning in the art, the context of this application, and in the particular context. Some terms used to describe this application are explained below or elsewhere in this specification to provide those skilled in the art with additional guidance regarding the description of this application.

[0053] The working principle and beneficial effects of the present application are summarized as follows: In the prior art, a series of processes, such as tooth preparation, dental impression taking, dental model injection, and wax pattern production, were manually performed by professional dental technicians. In this application, broadly speaking, the endocrown manufacturing time can be shortened by using a trained AI model to generate an endocrown model, and then printing the endocrown using a printing device.

[0054] The method for manufacturing endocrowns provided by the present application does not rely on professional dental technicians and can achieve precise quantitative control.

[0055] According to the present application, the generated endocrown model can be rapidly manufactured using a 3D printer and restored in a single visit, reducing costs (e.g., time costs) for both the dentist and the patient and improving the patient's medical experience.

[0056] Before manufacturing the finished endocrown, the endocrown model is subjected to a simulated strength test and a simulated trial fitting test. This dual testing approach reduces the risk that the finished endocrown will not be suitable for the patient, avoids extended patient care time due to extensive intraoperative adjustments of the finished endocrown, and avoids a shortened service life of the finished endocrown due to lower than expected mechanical strength.

[0057] In dual testing, the simulated strength test is performed before the simulated temporary fitting test, eliminating the need to regenerate the end-crown model and restart the simulated temporary fitting test, thus avoiding extended manufacturing times for the finished end-crown.

[0058] Improvements in endocrown manufacturing methods can reduce manufacturing difficulty and increase manufacturing speed, allowing clinics and other medical facilities to independently manufacture finished endocrowns (especially for simple cases), eliminating the need for multiple patient visits.

[0059] According to the present application, by utilizing artificial intelligence in the design of endocrowns, errors caused by over-reliance on the empirical judgment of professional dental technicians can be avoided.

[0060] The present application allows for customization of endocrowns, which can be individually tailored to the specific condition of the tooth, providing excellent retention and sufficient strength for future use while minimizing the removal of healthy tooth structure. [Brief explanation of the drawings]

[0061] [Figure 1] 1 is a schematic flowchart of a method for manufacturing an endocrown according to one embodiment of the present application. [Figure 2] 1 is a schematic configuration diagram of an endocrown manufacturing system according to an embodiment of the present application. FIG. DETAILED DESCRIPTION OF THE INVENTION

[0062] The present application will now be further described with reference to the accompanying drawings and embodiments. The present application will be clearly described below with reference to the drawings and detailed description, and those skilled in the art can change or modify the technology taught in this specification after understanding the embodiments of the present application without departing from the spirit and scope of the present application.

[0063] The terms used herein are used only for the purpose of describing particular embodiments and are not intended to limit the scope of the present application. As used herein, singular forms such as "a," "one," "the," and "said" also include plural forms.

[0064] Referring to Figure 1, the artificial intelligence-based manufacturing method for endocrowns is as follows: Step 1: acquiring oral CBCT images and intraoral scan models; Step 2: synthesizing the oral CBCT image and the intraoral scan model to generate a synthetic model; Step 3: inputting the synthetic model into a trained AI model, and generating an end-crown model based on the synthetic model by the AI ​​model; Step 4: performing a simulation strength test on the end-crown model, and regenerating the end-crown model if the stress in any region of the end-crown model is greater than a predetermined value; Step 5: conducting a simulated temporary fitting test on the end-crown model that has passed the simulated strength test; Step 6: adjusting the end-crown model based on the results of the simulated temporary fitting test; Step 7: inputting the adjusted end crown model into a printing device and printing and manufacturing a semi-finished end crown; and Step 8, wherein the semi-finished endocrown is subjected to a hardening treatment to produce a finished endocrown.

[0065] CBCT images and intraoral scan models are conventional technologies and are briefly described as follows. CBCT (cone beam computed tomography) imaging is a radiological imaging technology that uses X-ray 3D scans to provide information on hard tissues, such as jawbone morphology, nerve canal pathways, and maxillary sinus location. Intraoral scan models (intraoral scan data), on the other hand, are optical 3D surface scan technologies that can rapidly obtain the surface morphology of tooth crowns and gingiva without radiation, making them suitable for restoration design and orthodontic simulation. These two imaging modalities are synergistically applied through software integration. Combining crown data from intraoral scans with intraosseous structure data acquired via CBCT allows for the generation of a complete 3D model. Two-dimensional radiological imaging techniques, such as periapical radiographs and panoramic radiographs, can also be incorporated.

[0066] In some embodiments, the CBCT images are in DICOM format, the intraoral scan models are in STL format, and the synthesis process uses an Iterated Closest Points (ICP) algorithm for comparison.

[0067] This application is based on artificial intelligence using an AI model. The AI ​​model is trained as follows:

[0068] Collect three-dimensional jawbone data from CBCT scans and optical data from intraoral scans. Annotation of pulp chamber borders, margin design type (butt joint / wraparound), and material selection (e.g., lithium disilicate glass ceramics) by an oral specialist. We use a two-channel 3D CNN to process CBCT voxel data and intraoral scan point cloud data, and fuse features via a cross-attention mechanism. It integrates a mechanical simulation module, calculates stress distribution (e.g., von Mises stress) in real time, and constrains the rationality of the design. Optimize design accuracy and mechanical compatibility by performing supervised learning using past repair examples. Perform reinforcement learning optimization based on clinical success rates to provide manual corrective feedback for complex cases. The simulation strength test is mainly realized by finite element analysis (FEA), but a detailed description is omitted here. If the stress in any region of the end-crown model is greater than a predetermined value, the end-crown model is regenerated. Specifically, this is achieved by updating the surrogate model, annotating the defect area, and then performing calculations.

[0069] The simulated temporary fitting test is realized using virtual temporary fitting technology. The technical principle is to use augmented reality technology to superimpose a model of the restoration (endocrown model) onto a digital model of the patient's mouth (synthetic model). If no adjustments to the endocrown model are required, step 6 can be skipped.

[0070] When a semi-finished endocrown is printed and manufactured, it can be printed using a composite resin.

[0071] During the curing process, dual-curing resin cements (e.g., All Ceram Core) are used, which utilize a dual mechanism of light-initiated rapid curing (triggered by light irradiation) and chemically initiated deep curing (continuous reaction in the dark) to fully cure the restoration (endocrown) in areas with insufficient light transparency, providing reliable adhesive strength.

[0072] In the prior art, a series of processes, such as tooth preparation, dental impression taking, dental model injection, and wax pattern creation, were performed manually by professional dental technicians. In this application, broadly speaking, the manufacturing time of endocrowns can be reduced by using a trained AI model to generate endocrown models and then printing the endocrowns using a printing device.

[0073] The method for manufacturing endocrowns provided by the present application does not rely on professional dental technicians and can achieve precise quantitative control.

[0074] According to the present application, the generated endocrown model can be rapidly manufactured using a 3D printer and restored in a single visit, reducing costs (e.g., time costs) for both the dentist and the patient and improving the patient's medical experience.

[0075] Before manufacturing the finished endocrown, the endocrown model is subjected to a simulated strength test and a simulated trial fitting test. This dual testing approach reduces the risk that the finished endocrown will not be suitable for the patient, avoids extended patient care time due to extensive intraoperative adjustments of the finished endocrown, and avoids a shortened service life of the finished endocrown due to lower than expected mechanical strength.

[0076] In dual testing, the simulated strength test is performed before the simulated temporary fitting test, eliminating the need to regenerate the end-crown model and restart the simulated temporary fitting test, thus avoiding extended manufacturing times for the finished end-crown.

[0077] Improvements in endocrown manufacturing methods can reduce manufacturing difficulty and increase manufacturing speed, allowing clinics and other medical facilities to independently manufacture finished endocrowns (especially for simple cases), eliminating the need for multiple patient visits.

[0078] According to the present application, by utilizing artificial intelligence in the design of endocrowns, errors caused by over-reliance on the empirical judgment of professional dental technicians can be avoided.

[0079] This application allows for the customization of endocrowns, which can be individually tailored to the specific condition of the diseased tooth. This allows for excellent retention and sufficient strength for future use while minimizing the removal of healthy tooth structure. Based on this approach, minimally invasive restoration of dental crowns can be achieved. This method preserves approximately 2 mm of the dentin ferrule and eliminates the need for placing a metal post in the root canal, thereby avoiding damage to the root during root canal treatment preparation. This maximizes the preservation of remaining dental tissue, reduces the risk of tooth fracture, and extends the useful life of the tooth.

[0080] In this embodiment, in step 3, the step of generating an end-crown model includes: Step S1 of generating a holding portion to be held in the dental pulp cavity; Step S2 of generating an occlusion that engages with the opposing tooth and the adjacent tooth; and step S3 creating a margin that engages the remaining tooth structure (e.g., a dentin ferrule).

[0081] In step S1, a three-dimensional spiral retention structure is designed. Specifically, by utilizing the principle of biomimetic anchoring, a spiral groove structure with a gradient pitch is designed to ensure pull-out force while increasing interfacial bond strength. Furthermore, a porous reinforcement structure is designed. Specifically, a structure with various supports (e.g., honeycomb, columnar, or tree-like structures) is integrated within the retention portion, reducing the weight of the dental crown while improving its strength.

[0082] In step S2, multidimensional occlusal contact optimization is performed. Specifically, a virtual occlusion analysis system (e.g., Dental Occlusion Analyzer 4.0) is used to generate an optimal occlusal contact point distribution model through a machine learning algorithm, thereby avoiding interference between the occlusion area and other teeth.

[0083] In step S3, the margin contour is generated using 3D U-Net.

[0084] In this embodiment, the outer wall of the holding portion forms an adhesive surface, and the surface roughness of the adhesive surface is in the range of 0.2 μm to 0.5 μm.

[0085] In this embodiment, the above-mentioned surface roughness refers to the surface roughness of the outer wall of the holding portion.

[0086] Surface roughness affects the mechanical interlocking effect between the adhesive and the substrate (retaining part). If the adhesive surface is not smooth, from a microscopic perspective, the adhesive will penetrate into the uneven structure on the adhesive surface and solidify, creating an anchoring effect and increasing adhesive strength.

[0087] Surface roughness affects the adhesive strength of adhesives as follows: If the surface roughness is too low (<0.2 μm), the adhesive surface will be too smooth, resulting in insufficient mechanical interlocking force and a significant decrease in adhesive strength. If the surface roughness is too high (>0.5 μm), the adhesive will not be able to fully penetrate the uneven surface structure, which will result in stress concentrations or microcracks, which will in turn reduce the adhesive strength.

[0088] Surface roughness also affects plaque control as follows: if the surface roughness is too high (>0.5 μm), bacteria can easily adhere to the adhesive surface, promoting plaque formation and increasing the risk of secondary caries. Too low a surface roughness (<0.2 μm) reduces the number of bacterial attachment sites, helping to control the rate of plaque formation.

[0089] In short, when the surface roughness is in the range of 0.2 μm to 0.5 μm, not only is the adhesive strength of the adhesive ensured, but the rate of plaque formation is also suppressed, achieving a balance between adhesive strength and plaque control.

[0090] In this embodiment, in step S1 of step 3, the surface roughness of each region of the adhesive surface is identified based on a trained morphological feature-surface roughness association model.

[0091] The purpose of this example is to dynamically adjust the surface roughness of each region of the adhesive surface, one possible implementation of which is as follows: Based on the scan data, CAD software is used to analyze the shape features (e.g., grooves, sharp edges, flat areas) to identify the different requirements for adhesive strength and plaque control in each region.

[0092] Regarding the operational methodology of the morphological feature-surface roughness association model, the input features are compared with clinical standard values ​​in a database via a deep neural network (e.g., SSA-BiLSTM model) and a recommended roughness range (e.g., for flat areas, a roughness of 0.5 μm is recommended) is output.

[0093] The training process of the morphological feature-surface roughness association model refers to the training process of the AI ​​model described above.

[0094] In this embodiment, in step 4, the step of performing a simulation strength test is Step S1: applying a vertical load of magnitude A to the end-crown model; Step S2: applying a diagonal load of magnitude B to the end-crown model; Step S3: acquiring stress change data for each region of the end-crown model; and step S4 of comparing the maximum stress value during the simulated strength test with the tensile strength of the dentin, and regenerating the endocrown if the maximum stress value is greater than the tensile strength of the dentin.

[0095] In this embodiment, each part of the endocrown model is first generated, and then a simulation strength test is carried out on them all together, thereby shortening the manufacturing process of the finished endocrown. The specific settings of the vertical and oblique loads (e.g., the load magnitude) are adjusted according to actual requirements.

[0096] In some embodiments, A and B are 100N.

[0097] In some embodiments, when applying a diagonal load of magnitude B, the angle is set to 45 degrees.

[0098] In some embodiments, an asymmetric dynamic load spectrum (frequency range: 0.5-30 Hz) is constructed based on electromyographic (EMG) signals and chewing movement trajectory data.

[0099] In some embodiments, a hybrid algorithm combining the penalty function method and the Lagrange multiplier method is used to process the dynamic crown-antagonist contact and simulate stick-slip friction behavior.

[0100] In some embodiments, a saliva viscoelastic lubrication model (generalized Maxwell constitutive model) is introduced to analyze the effects of interfacial energy reduction and capillary action on peripheral microleakage in a moist environment.

[0101] In some embodiments, a thermo-mechanical coupling simulation is performed to simulate the thermal expansion mismatch stress at the ceramic-metal interface caused by thermal cycling (5°C⇔55°C).

[0102] In this embodiment, in step 6, the step of adjusting the end crown model includes: adjusting the convergence angle of the holding portion; adjusting the surface roughness of the holding portion; and adjusting the distribution of occlusal contact points on the occlusal portion.

[0103] The convergence angle may be considered as the tilt angle.

[0104] In this embodiment, the adjustment content in step 6 is clarified, and multiple adjustments are made to further ensure the adaptation of the endocrown model to the patient, thereby avoiding the extension of patient treatment due to (extensive) intraoperative adjustments to the finished endocrown.

[0105] Referring to FIG. 2, there is further provided an artificial intelligence-based endocrown manufacturing system, the system comprising: an acquisition module for acquiring oral CBCT images and intraoral scan models; a processing module that combines the intraoral CBCT image and the intraoral scan model to generate a combined model; a generation module that uses a trained AI model to generate an end-crown model based on the synthetic model; an analysis module that performs a simulation strength test on the end-crown model; and a verification module that performs a simulated temporary fitting test on the end-crown model that has passed the simulated strength test.

[0106] In this embodiment, the relevant explanations can be found in the corresponding parts of the above embodiments. Note that the specific structure of each module is not limited as long as it satisfies the requirements.

[0107] In this embodiment, the end-crown model is a holding portion that is held in the dental pulp cavity; an occlusal portion that engages with the opposing tooth and the adjacent tooth; and a margin portion that engages the remaining tooth structure.

[0108] In this embodiment, the endocrown model is divided into a holding part, an occlusal part, and a marginal part, so that different schemes can be selected to generate the endocrown model according to this division to meet different needs.

[0109] In this embodiment, the outer wall of the holding portion forms an adhesive surface, and the surface roughness of the adhesive surface is in the range of 0.2 μm to 0.5 μm.

[0110] When the surface roughness is in the range of 0.2 μm to 0.5 μm, not only is the adhesive strength of the adhesive ensured, but the rate of plaque formation is also suppressed, achieving a balance between adhesive strength and plaque control.

[0111] The above embodiments are intended to explain the technical concepts and features of the present application, enabling those skilled in the art to understand the contents of the present application and implement the present application accordingly, and do not limit the scope of protection of the present application. Any equivalent modifications or variations made according to the gist of the present application shall be included in the scope of protection of the present application.

Claims

1. Step 1: acquiring oral CBCT images and an intraoral scan model; Step 2: synthesizing the oral CBCT image and the intraoral scan model to generate a synthetic model; Step 3: inputting the synthetic model into a trained AI model, and generating an end-crown model based on the synthetic model by the AI ​​model; Step 4: performing a simulation strength test on the end-crown model, and regenerating the end-crown model if the stress in any region of the end-crown model is greater than a predetermined value; Step 5: conducting a simulated temporary fitting test on the end-crown model that has passed the simulated strength test; Step 6: adjusting the end-crown model based on the results of the simulated temporary fitting test; Step 7: inputting the adjusted end-crown model into a printing device and printing and manufacturing a semi-finished end-crown; Step 8: subjecting the semi-finished endocrown to a hardening treatment to produce a finished endocrown; A method for manufacturing an endocrown based on artificial intelligence, comprising:

2. In step 3, the step of generating an end-crown model comprises: Step S1 of generating a holding portion to be held in the dental pulp cavity; Step S2: generating an occlusion that engages with the opposing tooth and the adjacent tooth; and step S3 of creating a margin portion that engages the remaining tooth structure. The method for manufacturing an endocrown based on artificial intelligence according to claim 1 .

3. The outer wall of the holding portion forms an adhesive surface, and the surface roughness of the adhesive surface is in the range of 0.2 μm to 0.5 μm. The method for manufacturing an endocrown based on artificial intelligence according to claim 2.

4. In step S1 of step 3, the surface roughness of each region of the adhesive surface is identified based on a trained morphological feature-surface roughness association model; The method for manufacturing an endocrown based on artificial intelligence according to claim 3.

5. In step 4, the step of performing a simulation strength test includes: Step S1: applying a vertical load of magnitude A to the end-crown model; Step S2: applying a diagonal load of magnitude B to the end-crown model; Step S3: acquiring stress change data for each region of the end-crown model; and (S4) comparing the maximum stress value during the simulated strength test with the tensile strength of dentin, and regenerating the endocrown if the maximum stress value is greater than the tensile strength of the dentin. The method for manufacturing an endocrown based on artificial intelligence according to claim 1 .

6. In step 6, the step of adjusting the end-crown model comprises: adjusting the convergence angle of the holding portion; adjusting the surface roughness of the holding portion; and adjusting the distribution of occlusal contact points on the occlusal portion. The method for manufacturing an endocrown based on artificial intelligence according to claim 2.

7. an acquisition module for acquiring oral CBCT images and intraoral scan models; a processing module that combines the oral CBCT image and the intraoral scan model to generate a combined model; a generation module that uses a trained AI model to generate an end-crown model based on the synthetic model; an analysis module that performs a simulation strength test on the end-crown model; a verification module that performs a simulated temporary fitting test on the end-crown model that has passed the simulated strength test; An endocrown manufacturing system based on artificial intelligence, comprising:

8. The end-crown model is a holding portion that is held in the dental pulp cavity; an occlusal portion that engages with the opposing tooth and the adjacent tooth; a margin portion that engages the remaining tooth structure; 8. The artificial intelligence-based endocrown manufacturing system according to claim 7.

9. The outer wall of the holding portion forms an adhesive surface, and the surface roughness of the adhesive surface is in the range of 0.2 μm to 0.5 μm.

9. The artificial intelligence-based endocrown manufacturing system according to claim 8.

Images