Intelligent diagnosis and repair scheme design method of alveolar bone defect based on augmented reality

Abstract

The present invention discloses an augmented reality-based intelligent diagnosis and repair scheme design method for alveolar bone defects, including S1, collecting cone beam computed tomography image data of the patient's alveolar bone; S2, inputting the image data into a pre-trained convolutional neural network model; S3, performing morphological analysis on the identified defect area; S4, automatically generating a repair scheme for a specific defect type based on the output of the convolutional neural network model and the suggestion of a clinical decision support system; S5, using augmented reality technology to intuitively display the generated repair scheme in a real environment; S6, providing an interactive design experience in an augmented reality environment; S7, after confirming the final repair scheme, automatically generating detailed execution instructions for clinical use. The present invention uses augmented reality technology, so that doctors and patients can intuitively view the defect area and the expected effect after its repair in a three-dimensional view in a real environment, helping patients better understand the treatment plan.

Description

Technical Field

[0001] The present invention relates to the technical field of medical data analysis, and in particular to an augmented reality-based intelligent diagnosis and repair scheme design method for alveolar bone defects. Background Art

[0002] In modern dental treatment, the diagnosis and repair of alveolar bone defects is a common and complex problem, and the treatment effect is directly related to the patient's oral health and quality of life. The traditional treatment process of alveolar bone defects relies on medical imaging data such as X-rays and CT scans. Doctors need to judge the type and extent of the defect based on these two-dimensional or three-dimensional imaging data and their own clinical experience, and then design a repair plan. However, this traditional method has many obvious defects.

[0003] First, although traditional imaging diagnostic tools can provide structural information, they are still limited in identifying subtle defects and assessing the degree and type of defects. Two-dimensional images are difficult to fully display complex three-dimensional defects, and three-dimensional CT scanning equipment is expensive and has a relatively large radiation dose to patients. In addition, the interpretation of images depends on the doctor's experience and subjective judgment, and the diagnostic results of different doctors may vary, which poses a considerable challenge to the selection and optimization of treatment plans.

[0004] Secondly, the design of treatment plans lacks personalization. Existing technologies are limited in their ability to deal with individual differences, and often use standardized treatment methods that ignore the individual biomechanical properties of the patient's specific defect area. This may lead to poor treatment results, such as low implant success rates or incomplete functional recovery after restoration.

[0005] Therefore, how to provide an intelligent diagnosis and repair scheme design method for alveolar bone defects based on augmented reality is a problem that technical personnel in this field urgently need to solve. Summary of the invention

[0006] One purpose of the present invention is to propose an augmented reality-based intelligent diagnosis and repair scheme design method for alveolar bone defects. The present invention uses augmented reality technology to enable doctors and patients to intuitively view the defect area and its expected effect after repair in a three-dimensional view in a real environment, helping patients better understand the treatment plan.

[0007] According to an embodiment of the present invention, an augmented reality-based intelligent diagnosis and repair scheme design method for alveolar bone defects includes the following steps:

[0008] S1, collecting cone beam computed tomography image data of the patient's alveolar bone;

[0009] S2, inputting the image data into a pre-trained convolutional neural network model, using the convolutional neural network model to analyze the image data, and automatically identifying and locating the defect area in the alveolar bone;

[0010] S3, perform morphological analysis on the identified defect area and use a structure-sensitive codec network to perform a multi-dimensional assessment of the degree and type of the defect;

[0011] S4. Automatically generate restoration plans for specific defect types based on the output of the convolutional neural network model and the recommendations of the clinical decision support system, including the design of implant position and bone augmentation morphology;

[0012] S5. Use augmented reality technology to intuitively display the generated restoration plan in a real environment, so that doctors and patients can view the defect area and the expected effect after restoration in a three-dimensional view;

[0013] S6. Provides an interactive design experience in an augmented reality environment, allowing doctors to adjust implant positions and bone augmentation designs in real time and view the impact of the adjustments instantly.

[0014] S7. After confirming the final repair plan, detailed execution instructions are automatically generated for clinical use.

[0015] Optionally, the S1 includes the following steps:

[0016] S11. Configure cone beam computed tomography equipment to collect alveolar bone images, identify and adjust the corresponding patient's scanning parameters through the automated patient recognition system. The scanning parameters include the voltage and current of the radiation source, exposure time and scanning angle, and adjust the radiation dose received by the patient:

[0017]

[0018] Among them, V opt and I opt are the optimized voltage and current, V base and I base is the basic voltage and current, H patient is the height of the patient's head, H standard is the standard head height, D defect is the defect depth, D standard is the standard defect depth;

[0019] S12. Use the patient positioning system to automatically adjust the patient's head position to the optimal imaging area. The patient positioning system automatically corrects any slight movement of the patient through a real-time image feedback mechanism to eliminate image blur caused by incorrect or unstable posture of the patient:

[0020] θ opt=θ current +Δθ;

[0021]

[0022] Among them, θ opt is the optimized posture angle, θ current is the currently detected attitude angle, e(t) is the real-time attitude error, is the error change rate, K p and K d are the proportional and derivative control parameters;

[0023] S13. Implement enhanced image acquisition protocol. Add auxiliary customized scanning path based on standard cone beam computed tomography scanning. According to the specific defect area of ​​the patient's alveolar bone, optimize the scanning sequence through software algorithm, increase the coverage depth and detail resolution of the scanning for specific areas, and obtain more detailed defect area images:

[0024]

[0025] Among them, S opt is the optimized scanning sequence, S base is the basic scanning sequence, A defect is the defect area, A total is the total scan area;

[0026] S14, applying automatic image quality enhancement technology to evaluate the quality of the obtained image after image acquisition is completed, and automatically perform signal enhancement and noise reduction processing;

[0027] S15. After completing image acquisition, all image data are standardized, the images are converted into a format suitable for deep learning model processing, and the data is compressed and stored.

[0028] Optionally, S2 includes the following steps:

[0029] S21, receiving the standardized processed alveolar bone cone beam computed tomography image data;

[0030] S22, inputting the received image data into a pre-trained improved convolutional neural network model;

[0031] S23, using an improved convolutional neural network model to perform feature extraction and analysis, wherein the improved convolutional neural network model includes an adaptive feature learning layer, and controls the depth and width of feature extraction by dynamically adjusting learning parameters;

[0032] S24, performing multiple iterative learning through adaptive forward propagation and back propagation algorithms until the improved convolutional neural network model reaches a predetermined recognition accuracy or an upper limit of the number of iterations, so that each input cone beam computed tomography image data can be located in the defect area;

[0033] S25, outputting the defect area information in the identified and located alveolar bone:

[0034] O=λ·D out (N(I))+(1-λ)·I;

[0035] Where λ represents the balance factor, I represents the input image, the original or preprocessed image data, N(I) represents the noise reduction function, and D out Represents the output of a depthwise separable convolutional layer.

[0036] Optionally, the improved convolutional neural network model specifically includes:

[0037] The spatial pyramid pooling layer based on the attention mechanism is integrated into the improved convolutional neural network model. The size of the feature map is dynamically adjusted by adjusting the pooling ratio γ to capture different scale features from local to global:

[0038]

[0039] Among them, γ represents the pooling ratio, which is dynamically adjusted according to the size of the alveolar bone area, L is the loss function, which usually uses the cross entropy loss and is suitable for classification tasks, f(x i ; θ,γ) represents the input x i The network output under the model parameters θ and pooling ratio γ, η is the learning rate, which controls the update step size;

[0040] Each convolutional layer in the improved convolutional neural network model is followed by an adaptively adjusted batch normalization layer. The normalization layer automatically adjusts the feature distribution after convolution using a trainable scaling factor δ and shift factor ξ:

[0041]

[0042] Among them, g(z j ; δ,ξ) represents the normalized and transformed output, M is the number of samples in the batch;

[0043] Adaptive gating units are introduced to adjust the information flow of each layer of the improved convolutional neural network model, and the output of the previous layer is used to dynamically determine the amount of feature information required by the current layer;

[0044] G new =σ(W g ·H old +bg );

[0045] Among them, G new represents the output of the gated unit, which controls the degree of information transmission. σ is the Sigmoid activation function, which ensures that the output is between 0 and 1 and controls the amount of information transmission. W g and b g are the weights and biases of the gating unit;

[0046] The depthwise separable convolutional layer in the improved convolutional neural network model decomposes the standard convolution operation into a spatial convolution followed by a depthwise convolution.

[0047]

[0048] in, represents the result of applying the lth spatial convolution kernel to input I, represents the operation of applying the lth depthwise convolution kernel, where L is the number of depthwise convolution kernels.

[0049] Optionally, S3 includes the following steps:

[0050] S31, receiving output defect area information in the alveolar bone;

[0051] S32, inputting the received defect area information into a structure-sensitive encoding and decoding network for analyzing the morphology and degree of alveolar bone defects, including a multi-layer encoding and decoding structure to extract the features of the defect area and perform in-depth multi-dimensional analysis:

[0052]

[0053] Among them, x represents the original defect area input data, μ and σ 2 Represent the mean and variance of the data respectively;

[0054] S33. Through the encoding process, the network extracts the features of the defective area layer by layer, and uses the deep residual network structure to enhance the feature extraction capability and the training depth of the network:

[0055]

[0056] Among them, ReLU is the linear rectification function, BN means batch normalization, SE means the compression-excitation function, which recalibrates the feature response of the channel, and α is the mixing parameter used to balance the contribution of the two feature extraction techniques;

[0057] S34. During the decoding process, the encoded features are used to reconstruct the information of the defective area to estimate the type and degree of the defect. The decoding process uses deconvolution and dynamic routing mechanisms:

[0058]

[0059] Among them, * represents the convolution operation, Up represents the upsampling operation, and z k is the feature map from the previous layer, is the kth decoding convolution kernel, β k is the coefficient calculated by the dynamic routing algorithm, indicating the importance of the kth feature map in the decoding process;

[0060] S35. Perform multi-dimensional evaluation on the decoded output to determine the type and degree of the defect, use predefined classification standards and quantitative indicators to perform the evaluation, and output the evaluated defect type and degree.

[0061] Optionally, S4 includes the following steps:

[0062] S41. Receive assessed defect type and degree information;

[0063] S42, comparing the received defect type and degree information with the historical data and treatment plans stored in the clinical decision support system, and using a decision tree algorithm to determine the most suitable repair plan for the current defect situation;

[0064]

[0065] Among them, T(x) is the final repair solution decision output, x is the input defect information, and h i represents the output of the i-th decision tree model, α i is the weight of the i-th decision tree model, Θ i is the model parameter of the i-th decision tree, y is the ideal treatment outcome, σ 2 is the variance of the model output, which is used to adjust the contribution weight of each model;

[0066] S43, automatically generating a specific implant position and bone augmentation morphology design according to the selected restoration plan, and using a genetic algorithm to adjust and optimize the design parameters so that the designed implant position and bone augmentation morphology best adapt to the specific defect of the patient;

[0067]

[0068] Among them, P new and P old are the design parameters before and after optimization, η is the learning rate, controlling the update step, β j is the selection probability of the jth individual, usually determined by the fitness function f, mutate(P old ,δ j is the mutation operation applied to the parameter P, δ jis the intensity of the mutation operation, J is the number of individuals in the population;

[0069] S44. Use computational fluid dynamics simulation technology to simulate the designed implant position and bone augmentation morphology to evaluate its biomechanical adaptability;

[0070]

[0071] Among them, F sim represents the result of fluid dynamics simulation, ρ is the density of the fluid, is the fluid velocity vector, t is the time variable, p is the pressure of the fluid, μ is the dynamic viscosity of the fluid, and f is the volume force acting on the fluid;

[0072] S45. Based on the simulation and evaluation results, the implant position and bone augmentation morphology are further adjusted, and the detailed information of the final confirmed restoration plan is output, including the specific position of the implant, the detailed description of the bone augmentation morphology, and the expected treatment effect.

[0073] Optionally, S5 includes the following steps:

[0074] S51. Receive the final confirmed restoration plan details, including the specific location of the implant and the shape of the bone augmentation;

[0075] S52, converting the received restoration plan information into an augmented reality compatible format, including generation and optimization of a three-dimensional model:

[0076]

[0077] Among them, M 3D It is the original three-dimensional model data obtained by CBCT data through volume rendering technology. Indicates the model rotation, which depends on the Euler angle θ, ψ,T(s x ,s y ,s z ) represents the scaling transformation of the model, which is used to adjust the scale in different directions. L(λ,v) represents the illumination transformation of the model, where λ is the illumination intensity and v is the illumination direction vector.

[0078] S53. Display a 3D view of the defect area and its expected effect after repair to doctors and patients through a head-mounted display or smartphone application:

[0079]

[0080] Among them, V display represents the final augmented reality view, M AR (p) is the 3D model data at position p, is from the user's position P user Through the rendering function The calculated camera perspective transformation,Ω represents the set of the entire display area;

[0081] S54. Doctors and patients adjust the view angle and zoom level through the touch screen or voice commands to understand the treatment plan and expected results:

[0082]

[0083] Among them, U interaction represents the updated view after user interaction, ΔV is based on the user input C input and the view change function produced by adjusting the parameters ξ, the zoom level, and κ, the rotation angle;

[0084] S55. Adjusting the display of the augmented reality view based on user feedback and device feedback:

[0085]

[0086] Among them, V final is the final adjusted augmented reality view, F user and F device denote the adjustment functions from user and device feedback, respectively.

[0087] The beneficial effects of the present invention are:

[0088] (1) The present invention integrates deep learning technology with advanced image processing algorithms to automatically and accurately identify the type and extent of alveolar bone defects from CBCT medical images, and uses a structure-sensitive codec network to perform multi-dimensional morphological analysis of the defect, providing more accurate and detailed diagnostic information than traditional methods, reducing human errors and diagnostic time.

[0089] (2) The present invention intelligently generates a restoration plan that best suits the patient's specific defect condition based on the automatic diagnosis results and the recommendations of the clinical decision support system. It also uses a decision tree algorithm and a genetic algorithm to adjust and optimize the implant position and bone augmentation morphology, thereby ensuring the personalization and adaptability of the design plan.

[0090] (3) The present invention utilizes augmented reality technology, allowing doctors and patients to intuitively view the defect area and the expected effect after repair in a three-dimensional view in a real environment, helping patients better understand the treatment plan and making doctor-patient communication more efficient and satisfactory. BRIEF DESCRIPTION OF THE DRAWINGS

[0091] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0092] Figure 1 This is a flow chart of an augmented reality-based intelligent diagnosis and repair scheme design method for alveolar bone defects proposed by the present invention. DETAILED DESCRIPTION

[0093] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0094] refer to Figure 1 , an augmented reality-based intelligent diagnosis and repair scheme design method for alveolar bone defects, comprising the following steps:

[0095] S1, collecting cone beam computed tomography image data of the patient's alveolar bone;

[0096] In this implementation, S1 includes the following steps:

[0097] S11. Configure cone beam computed tomography equipment to collect alveolar bone images, identify and adjust the corresponding patient's scanning parameters through the automated patient recognition system. The scanning parameters include the voltage and current of the radiation source, exposure time and scanning angle, and adjust the radiation dose received by the patient:

[0098]

[0099] Among them, V opt and I opt are the optimized voltage and current, V base and I base is the basic voltage and current, H patient is the height of the patient's head, H standard is the standard head height, D defect is the defect depth, D standard is the standard defect depth;

[0100] S12. Use the patient positioning system to automatically adjust the patient's head position to the optimal imaging area. The patient positioning system automatically corrects any slight movement of the patient through a real-time image feedback mechanism to eliminate image blur caused by incorrect or unstable posture of the patient:

[0101] θ opt =θ current +Δθ;

[0102]

[0103] Among them, θ opt is the optimized posture angle, θ current is the currently detected attitude angle, e(t) is the real-time attitude error, is the error change rate, K p and K d are the proportional and derivative control parameters;

[0104] S13. Implement enhanced image acquisition protocol. Add auxiliary customized scanning path based on standard cone beam computed tomography scanning. According to the specific defect area of ​​the patient's alveolar bone, optimize the scanning sequence through software algorithm, increase the coverage depth and detail resolution of the scanning for specific areas, and obtain more detailed defect area images:

[0105]

[0106] Among them, S opt is the optimized scanning sequence, S base is the basic scanning sequence, A defect is the defect area, A total is the total scan area;

[0107] S14, applying automatic image quality enhancement technology to evaluate the quality of the obtained image after image acquisition is completed, and automatically perform signal enhancement and noise reduction processing;

[0108] S15. After completing image acquisition, all image data are standardized, the images are converted into a format suitable for deep learning model processing, and the data is compressed and stored.

[0109] S2, inputting the image data into a pre-trained convolutional neural network model, using the convolutional neural network model to analyze the image data, and automatically identifying and locating the defect area in the alveolar bone;

[0110] In this implementation, S2 includes the following steps:

[0111] S21, receiving the standardized processed alveolar bone cone beam computed tomography image data;

[0112] S22, inputting the received image data into a pre-trained improved convolutional neural network model;

[0113] S23, using an improved convolutional neural network model to perform feature extraction and analysis, wherein the improved convolutional neural network model includes an adaptive feature learning layer, and controls the depth and width of feature extraction by dynamically adjusting learning parameters;

[0114] S24, performing multiple iterative learning through adaptive forward propagation and back propagation algorithms until the improved convolutional neural network model reaches a predetermined recognition accuracy or an upper limit of the number of iterations, so that each input cone beam computed tomography image data can be located in the defect area;

[0115] S25, outputting the defect area information in the identified and located alveolar bone:

[0116] O=λ·D out (N(I))+(1-λ)·I;

[0117] Where λ represents the balance factor, I represents the input image, the original or preprocessed image data, N(I) represents the noise reduction function, and D out Represents the output of a depthwise separable convolutional layer.

[0118] In this implementation, the improved convolutional neural network model specifically includes:

[0119] The spatial pyramid pooling layer based on the attention mechanism is integrated into the improved convolutional neural network model. The size of the feature map is dynamically adjusted by adjusting the pooling ratio γ to capture different scale features from local to global:

[0120]

[0121] Among them, γ represents the pooling ratio, which is dynamically adjusted according to the size of the alveolar bone area, L is the loss function, which usually uses the cross entropy loss and is suitable for classification tasks, f(x i ; θ,γ) represents the input x i The network output under the model parameters θ and pooling ratio γ, η is the learning rate, which controls the update step size;

[0122] Each convolutional layer in the improved convolutional neural network model is followed by an adaptively adjusted batch normalization layer. The normalization layer automatically adjusts the feature distribution after convolution using a trainable scaling factor δ and shift factor ξ:

[0123]

[0124] Among them, g(z j ; δ,ξ) represents the normalized and transformed output, M is the number of samples in the batch;

[0125] Adaptive gating units are introduced to adjust the information flow of each layer of the improved convolutional neural network model, and the output of the previous layer is used to dynamically determine the amount of feature information required by the current layer;

[0126] G new =σ(W g ·H old +b g );

[0127] Among them, G new represents the output of the gated unit, which controls the degree of information transmission. σ is the Sigmoid activation function, which ensures that the output is between 0 and 1 and controls the amount of information transmission. Wg and b g are the weights and biases of the gating unit;

[0128] The depthwise separable convolutional layer in the improved convolutional neural network model decomposes the standard convolution operation into a spatial convolution followed by a depthwise convolution.

[0129]

[0130] in, represents the result of applying the lth spatial convolution kernel to input I, represents the operation of applying the lth depthwise convolution kernel, where L is the number of depthwise convolution kernels.

[0131] S3, perform morphological analysis on the identified defect area and use a structure-sensitive codec network to perform a multi-dimensional assessment of the degree and type of the defect;

[0132] In this implementation, S3 includes the following steps:

[0133] S31, receiving output defect area information in the alveolar bone;

[0134] S32, inputting the received defect area information into a structure-sensitive encoding and decoding network for analyzing the morphology and degree of alveolar bone defects, including a multi-layer encoding and decoding structure to extract the features of the defect area and perform in-depth multi-dimensional analysis:

[0135]

[0136] Among them, x represents the original defect area input data, μ and σ 2 Represent the mean and variance of the data respectively;

[0137] S33. Through the encoding process, the network extracts the features of the defective area layer by layer, and uses the deep residual network structure to enhance the feature extraction capability and the training depth of the network:

[0138] E(x)=σ(W e ·(α·ReLU(BN(x))+(1-α)·SE(x))+b e );

[0139] Among them, ReLU is the linear rectification function, BN means batch normalization, SE means the compression-excitation function, which recalibrates the feature response of the channel, and α is the mixing parameter used to balance the contribution of the two feature extraction techniques;

[0140] S34. During the decoding process, the encoded features are used to reconstruct the information of the defective area to estimate the type and degree of the defect. The decoding process uses deconvolution and dynamic routing mechanisms:

[0141]

[0142] Among them, * represents the convolution operation, Up represents the upsampling operation, and z k is the feature map from the previous layer, is the kth decoding convolution kernel, β k is the coefficient calculated by the dynamic routing algorithm, indicating the importance of the kth feature map in the decoding process;

[0143] S35. Perform multi-dimensional evaluation on the decoded output to determine the type and degree of the defect, use predefined classification standards and quantitative indicators to perform the evaluation, and output the evaluated defect type and degree.

[0144] S4. Automatically generate restoration plans for specific defect types based on the output of the convolutional neural network model and the recommendations of the clinical decision support system, including the design of implant position and bone augmentation morphology;

[0145] In this implementation, S4 includes the following steps:

[0146] S41. Receive assessed defect type and degree information;

[0147] S42, comparing the received defect type and degree information with the historical data and treatment plans stored in the clinical decision support system, and using a decision tree algorithm to determine the most suitable repair plan for the current defect situation;

[0148]

[0149] Among them, T(x) is the final repair solution decision output, x is the input defect information, and h i represents the output of the i-th decision tree model, α i is the weight of the i-th decision tree model, Θ i is the model parameter of the i-th decision tree, y is the ideal treatment outcome, σ 2 is the variance of the model output, which is used to adjust the contribution weight of each model;

[0150] S43, automatically generating a specific implant position and bone augmentation morphology design according to the selected restoration plan, and using a genetic algorithm to adjust and optimize the design parameters so that the designed implant position and bone augmentation morphology best adapt to the specific defect of the patient;

[0151]

[0152] Among them, P new and P old are the design parameters before and after optimization, η is the learning rate, controlling the update step, β jis the selection probability of the jth individual, usually determined by the fitness function f, mutate(P old ,δ j is the mutation operation applied to the parameter P, δ j is the intensity of the mutation operation, J is the number of individuals in the population;

[0153] S44. Use computational fluid dynamics simulation technology to simulate the designed implant position and bone augmentation morphology to evaluate its biomechanical adaptability;

[0154]

[0155] Among them, F sim represents the result of fluid dynamics simulation, ρ is the density of the fluid, is the fluid velocity vector, t is the time variable, p is the pressure of the fluid, μ is the dynamic viscosity of the fluid, and f is the volume force acting on the fluid;

[0156] S45. Based on the simulation and evaluation results, the implant position and bone augmentation morphology are further adjusted, and the detailed information of the final confirmed restoration plan is output, including the specific position of the implant, the detailed description of the bone augmentation morphology, and the expected treatment effect.

[0157] S5. Use augmented reality technology to intuitively display the generated restoration plan in a real environment, so that doctors and patients can view the defect area and the expected effect after restoration in a three-dimensional view;

[0158] S6. Provides an interactive design experience in an augmented reality environment, allowing doctors to adjust implant positions and bone augmentation designs in real time and view the impact of the adjustments instantly.

[0159] In this implementation, S5 includes the following steps:

[0160] S51. Receive the final confirmed restoration plan details, including the specific location of the implant and the shape of the bone augmentation;

[0161] S52, converting the received restoration plan information into an augmented reality compatible format, including generation and optimization of a three-dimensional model:

[0162]

[0163] Among them, M 3D It is the original three-dimensional model data obtained by CBCT data through volume rendering technology. Indicates the model rotation, which depends on the Euler angle θ, ψ,T(s x ,s y ,s z) represents the scaling transformation of the model, which is used to adjust the scale in different directions. L(λ,v) represents the illumination transformation of the model, where λ is the illumination intensity and v is the illumination direction vector.

[0164] S53. Display a 3D view of the defect area and its expected effect after repair to doctors and patients through a head-mounted display or smartphone application:

[0165]

[0166] Among them, V display represents the final augmented reality view, M AR (p) is the 3D model data at position p, is from the user's position P user Through the rendering function The calculated camera perspective transformation,Ω represents the set of the entire display area;

[0167] S54. Doctors and patients adjust the view angle and zoom level through the touch screen or voice commands to understand the treatment plan and expected results:

[0168]

[0169] Among them, U interaction represents the updated view after user interaction, ΔV is based on the user input C input and the view change function produced by adjusting the parameters ξ, the zoom level, and κ, the rotation angle;

[0170] S55. Adjusting the display of the augmented reality view based on user feedback and device feedback:

[0171]

[0172] Among them, V final is the final adjusted augmented reality view, F user and F device denote the adjustment functions from user and device feedback, respectively.

[0173] S7. After confirming the final repair plan, detailed execution instructions are automatically generated for clinical use.

[0174] By integrating deep learning technology with advanced image processing algorithms, the present invention can automatically and accurately identify the type and extent of alveolar bone defects from CBCT medical images, and use a structure-sensitive codec network to perform multi-dimensional morphological analysis of the defect, providing more accurate and detailed diagnostic information than traditional methods, reducing human errors and diagnostic time.

[0175] The present invention intelligently generates a restoration plan that best suits the patient's specific defect condition based on the automatic diagnosis results and the recommendations of the clinical decision support system. It adjusts and optimizes the implant position and bone augmentation morphology by using a decision tree algorithm and a genetic algorithm to ensure the personalization and adaptability of the design plan.

[0176] The present invention utilizes augmented reality technology, allowing doctors and patients to intuitively view the defect area and the expected effect after repair in a real environment in a three-dimensional view, helping patients to better understand the treatment plan and making doctor-patient communication more efficient and satisfactory.

[0177] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. An augmented reality-based intelligent diagnosis and repair scheme design method for alveolar bone defects, characterized in that: The following steps are involved: S1, collecting cone beam computed tomography image data of the patient's alveolar bone; S2, inputting the image data into a pre-trained convolutional neural network model, using the convolutional neural network model to analyze the image data, and automatically identifying and locating the defect area in the alveolar bone; S3, perform morphological analysis on the identified defect area and use a structure-sensitive codec network to perform a multi-dimensional assessment of the degree and type of the defect; S4. Automatically generate restoration plans for specific defect types based on the output of the convolutional neural network model and the recommendations of the clinical decision support system, including the design of implant position and bone augmentation morphology; S5. Use augmented reality technology to intuitively display the generated restoration plan in a real environment, so that doctors and patients can view the defect area and the expected effect after restoration in a three-dimensional view; S6. Provides an interactive design experience in an augmented reality environment, allowing doctors to adjust implant positions and bone augmentation designs in real time and view the impact of the adjustments instantly. S7, after confirming the final repair plan, automatically generate detailed execution instructions for clinical use; The S1 comprises the following steps: S11. Configure cone beam computed tomography equipment to collect alveolar bone images, identify and adjust the corresponding patient's scanning parameters through the automated patient recognition system. The scanning parameters include the voltage and current of the radiation source, exposure time and scanning angle, and adjust the radiation dose received by the patient: Among them, V opt and I opt are the optimized voltage and current, V base and I base is the basic voltage and current, H patient is the height of the patient's head, H standard is the standard head height, D defect is the defect depth, D standard is the standard defect depth; S12. Use the patient positioning system to automatically adjust the patient's head position to the optimal imaging area. The patient positioning system automatically corrects any slight movement of the patient through a real-time image feedback mechanism to eliminate image blur caused by incorrect or unstable posture of the patient: i opt =θ current +Δθ; Among them, θ opt is the optimized posture angle, θ current is the currently detected attitude angle, e(t) is the real-time attitude error, is the error change rate, K p and K d are the proportional and derivative control parameters; S13. Implement enhanced image acquisition protocol. Add auxiliary customized scanning path based on standard cone beam computed tomography scanning. According to the specific defect area of ​​the patient's alveolar bone, optimize the scanning sequence through software algorithm, increase the coverage depth and detail resolution of the scanning for specific areas, and obtain more detailed defect area images: Among them, S opt is the optimized scanning sequence, S base is the basic scanning sequence, A defect is the defect area, A total is the total scan area; S14, applying automatic image quality enhancement technology to evaluate the quality of the obtained image after image acquisition is completed, and automatically perform signal enhancement and noise reduction processing; S15. After completing image acquisition, all image data are standardized, the images are converted into a format suitable for deep learning model processing, and the data is compressed and stored.

Images