A remote dental prosthetic diagnosis and treatment system and method based on a 5G network

Abstract

The application relates to the technical field of communication protocols, in particular to a remote oral prosthetic diagnosis and treatment system and method based on a 5G network, which comprises an image acquisition module, a feature analysis module, a strategy generation module, a trajectory prediction module and a scheme feedback module.In the application, a three-dimensional scanning data stream of the oral cavity is transmitted in real time, a Gaussian filtering algorithm is used to filter out noise, the signal-to-noise ratio of original data is improved, environmental interference and equipment errors are reduced, a contour boundary is extracted based on edge detection, edge sharpness and curvature parameters are calculated, geometric features of lesions are quantified, irregular defect recognition accuracy is improved, a convolutional neural network is used to fuse defect area, depth value and damage index, a repair precision coefficient is generated, multi-dimensional indexes and repair strategies are dynamically mapped, scheme adaptability is improved, an LSTM time sequence is used to predict a defect change trend, a lesion development trajectory is predicted, repair parameters are dynamically adjusted, repair body failure is avoided, and a closed-loop logic chain is formed in each link, so that data processing and decision-making efficiency are optimized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of communication protocols, and in particular to a remote oral prosthetic diagnosis and treatment system and method based on a 5G network. BACKGROUND

[0002] The technical field of communication protocols includes a series of rules, standards and control methods set in a communication network to achieve data transmission, resource sharing and terminal interoperability. The core content of the communication protocol includes data format, signal exchange sequence, error detection and correction mechanism, session control, synchronization mechanism and security protection measures. Overall, the technical field of communication protocols mainly covers protocol standards at various levels of physical layer, data link layer, network layer, transport layer and application layer, and realizes end-to-end data transmission and information exchange function through the combination of different protocol stacks, involving various communication environments such as wireless communication, wired communication, satellite communication and Internet protocol, and is widely used in various information systems and intelligent terminal devices to support seamless communication and collaborative operation between networks.

[0003] Among them, the remote oral prosthetic diagnosis and treatment system refers to a system based on 5G communication protocol, connecting the oral prosthetic expert end and the patient end through the remote medical platform, realizing the interaction of diagnosis and treatment information, real-time transmission of image data and operation guidance. The technical matters targeted by the patent subject cover the use of 5G ultra-low latency and high-reliability bandwidth characteristics to complete the remote real-time uploading of patient's three-dimensional imaging data, the expert end based on the received data to make oral prosthetic diagnosis and develop a scheme, and send control instructions to the local prosthetic device through the remote control terminal to guide the local device to perform accurate prosthetic operation; The system uses standardized communication protocols to achieve uniform diagnosis and treatment data formats, and the transmission process is secured by network encryption protocols to ensure accurate and timely information transmission during remote diagnosis and treatment.

[0004] The prior art relies on 5G transmission of raw scan data, but does not integrate a preprocessing mechanism, and noise or artifacts are not filtered out, resulting in a risk of distortion of the three-dimensional image received remotely, affecting the accuracy of diagnosis. The existing system uses a fixed threshold to segment the lesion area, and the recognition accuracy of complex morphological edges is insufficient, which may cause feature extraction deviation, resulting in errors between the prosthetic scheme and the real lesion geometric parameters. The traditional method relies on expert experience to develop static prosthetic strategies, lacks data modeling and prediction ability for dynamic development of lesions, and the prosthetic parameters cannot be adjusted with the progression of the lesion, which may cause long-term adaptability decline of the prosthetic body and the dental tissue. The existing data encryption mechanism only targets the transmission layer, and does not design an encryption algorithm in combination with the characteristics of three-dimensional image data, which may cause the risk of data being partially restored. The existing protocol stack lacks cross-layer collaboration, and the synchronization of data flow and control instructions in network fluctuation scenarios may cause device response delay or operation interruption. For example, unoptimized protocol compatibility causes data packet loss during high-concurrent transmission, affecting the continuity of prosthetic operation. SUMMARY

[0005] The present application aims to solve the problems existing in the prior art and proposes a remote dental prosthetic diagnosis and treatment system and method based on a 5G network.

[0006] In order to achieve the above-mentioned purpose, the present application adopts the following technical scheme: a remote dental prosthetic diagnosis and treatment system based on a 5G network comprises:

[0007] An image acquisition module is configured to receive three-dimensional scanning data stream of the oral cavity through 5G real-time transmission bandwidth, call a Gaussian filtering algorithm to filter out noise, extract a lesion area coordinate set, generate three-dimensional reconstruction data, and deliver the three-dimensional reconstruction data to a feature analysis module;

[0008] A feature analysis module is configured to receive the three-dimensional reconstruction data, locate a contour boundary based on an edge detection algorithm, calculate an edge sharpness value and a curvature parameter, generate a lesion feature parameter, and deliver the lesion feature parameter and the edge sharpness value to a strategy generation module;

[0009] A strategy generation module is configured to receive the lesion feature parameter and the edge sharpness value, input the lesion feature parameter and the edge sharpness value into a convolutional neural network model for lesion classification, output a defect area, a depth value, and a structure damage index, generate a repair precision coefficient based on a product, map the repair precision coefficient to a repair technology number, generate a dynamic repair strategy table, and deliver the dynamic repair strategy table and the depth value to a trajectory prediction module;

[0010] A trajectory prediction module is configured to receive the dynamic repair strategy table and the depth value, call an LSTM time series model to model a depth change rate, combine a defect area growth rate to generate an evolution trajectory prediction vector, output an updated strategy instruction, and deliver the updated strategy instruction to a scheme feedback module.

[0011] As a further scheme of the present application, the three-dimensional reconstruction data comprises a lesion area spatial coordinate set, surface point cloud data, and denoised texture feature data, the lesion feature parameter comprises an edge sharpness distribution, a curvature change matrix, and a morphological complexity index, the dynamic repair strategy table comprises a repair technology number, a repair priority, and an estimated repair period, and the evolution trajectory prediction vector comprises a depth change rate sequence, an area expansion rate curve, and an evolution time node marker.

[0012] As a further scheme of the present application, the image acquisition module comprises:

[0013] A 5G transmission submodule is configured to collect real-time three-dimensional scanning data stream of the oral cavity, establish a communication link through a 5G transmission protocol, divide the data stream into independent data packets and add a time stamp, complete data reorganization according to a transmission serial number, and generate a transmission data stream;

[0014] The filter denoising sub-module calls a Gaussian filtering algorithm, constructs a three-dimensional spatial coordinate system based on the transmission data stream, performs neighborhood weighted average calculation on each coordinate point, sets a standard deviation parameter as 0.8, eliminates the coordinate offset of outliers, and generates filtered data;

[0015] The lesion coordinate generation sub-module performs gradient calculation based on the filtered data, performs difference operation on the gray values of adjacent coordinate points, selects a region with a gray difference value exceeding 0.12 as a boundary judgment reference, extracts a vertex coordinate set of a continuous closed region, and generates three-dimensional reconstruction data.

[0016] As a further scheme of the application, the feature analysis module comprises:

[0017] The edge detection sub-module calls an edge detection algorithm based on the three-dimensional reconstruction data, performs gradient amplitude and direction calculation on each coordinate point, sets high and low threshold values as 0.3 and 0.1 respectively, selects a point with a gradient amplitude higher than the high threshold value as a boundary point, connects adjacent points to form a closed contour, and generates a boundary coordinate set;

[0018] The sharpness curvature calculation sub-module calls the boundary coordinate set, performs second derivative calculation on the coordinates in the neighborhood of each boundary point, extracts the maximum curvature absolute value as a curvature parameter, simultaneously performs normalization processing on the gradient amplitude, calculates the average value of the gradient direction change rate of adjacent points, and generates an edge sharpness value and a curvature parameter;

[0019] The feature parameter generation sub-module performs weighted fusion calculation on a region with a sharpness value exceeding 0.25 based on the edge sharpness value and the curvature parameter, sets a curvature weight coefficient as 0.6 and a sharpness weight coefficient as 0.4, integrates all point parameters in the region to generate an average value and a standard deviation, and obtains a lesion feature parameter.

[0020] As a further scheme of the application, the strategy generation module comprises:

[0021] The classification calculation sub-module inputs a parameter matrix into a fully connected layer of a convolutional neural network based on the lesion feature parameter and the edge sharpness value, extracts a multi-dimensional feature vector, performs linear superposition operation of the feature vector and a weight matrix, outputs a probability distribution through an activation function, and obtains a defect area, a depth value and a structure damage index;

[0022] The coefficient generation sub-module performs unit area standardization on the defect area, converts the depth value into a relative depth ratio, performs product operation of the three in combination with the structure damage index, introduces a weight correction factor to adjust the product result, and generates a repair precision coefficient;

[0023] The policy mapping submodule matches corresponding technical numbers in a preset mapping table according to a numerical interval of the repair precision coefficient, divides levels of the depth value at a precision of 0.1 mm, integrates the numbers and the depth levels, and establishes a dynamic repair strategy table.

[0024] As a further scheme of the present application, the trajectory prediction module comprises:

[0025] The depth modeling submodule calls a depth value sequence in the dynamic repair strategy table, inputs the depth value sequence into an LSTM time sequence model, calculates time weights through a forgetting gate and an input gate, performs sliding window segmentation on the depth value, outputs a hidden layer state vector, and generates a depth change rate;

[0026] The trajectory generation submodule performs adjacent time point difference calculation on the missing area based on the depth change rate, fits an area growth curve through linear regression, extracts a slope parameter as a growth rate, performs matrix splicing on the change rate and the growth rate, and generates an evolution trajectory prediction vector;

[0027] The policy updating submodule performs matching operation of vectors and preset policy rules according to a dimension feature of the evolution trajectory prediction vector, screens candidate strategies that meet an error threshold, marks an optimal strategy as a priority, and outputs an updated policy instruction.

[0028] As a further scheme of the present application, the system further comprises:

[0029] The scheme feedback module is configured to receive the updated policy instruction, match a repair material database, extract a compressive strength and an adhesion coefficient, calculate an adaptation degree in combination with real-time occlusal force monitoring data, generate a repair scheme, and synchronize the repair scheme to a patient-side wearable sensor.

[0030] The repair scheme comprises a material adaptation degree parameter, a compressive strength value, and an adhesion performance index.

[0031] As a further scheme of the present application, the scheme feedback module comprises:

[0032] The material matching submodule calls a repair technical number in the updated policy instruction, traverses a primary key of the repair material database based on a number hash value, extracts a compressive strength and an adhesion coefficient field after verifying entry validity, and generates a material parameter set;

[0033] The adaptation degree calculation submodule performs division operation by taking the compressive strength of the material parameter set as a numerator and real-time occlusal force monitoring data as a denominator, extracts a product term of the adhesion coefficient and an occlusal force change rate, superimposes a weight coefficient after normalizing two results, and generates an adaptation degree coefficient;

[0034] The scheme generation sub-module divides the adaptation level interval according to the numerical distribution of the adaptation coefficient, screens the material combination that satisfies the interval requirement of the compressive strength and the adhesion coefficient, aligns the occlusal force data and the material parameters in time sequence to generate an index matrix, and establishes a repair scheme.

[0035] A remote dental restoration diagnosis and treatment method based on a 5G network, which is executed based on the remote dental restoration diagnosis and treatment system based on the 5G network, and includes the following steps:

[0036] S1: receiving a real-time transmitted three-dimensional scanning data stream of the oral cavity through a 5G communication protocol, calling a Gaussian filtering algorithm to filter noise of the data stream, extracting a lesion area coordinate set, and generating three-dimensional reconstruction data;

[0037] S2: inputting the three-dimensional reconstruction data into an edge detection algorithm to locate a contour boundary, calculating an edge sharpness value by calculating a boundary pixel gradient amplitude, and generating a lesion feature parameter by combining a difference in curvature of adjacent pixels;

[0038] S3: inputting the lesion feature parameter and the edge sharpness value into a convolutional neural network model for classification training, outputting a defect area, a depth value and a structure damage index, performing a product operation on the depth value and the structure damage index to generate a repair precision coefficient, matching the repair precision coefficient with a preset repair technology mapping table, and generating a dynamic repair strategy table;

[0039] S4: calling an LSTM time series model based on the dynamic repair strategy table and the depth value to predict a change rate of the depth value, combining a difference in adjacent time stamps of the defect area to generate an evolution trajectory prediction vector, and outputting an update strategy instruction containing a correction parameter;

[0040] S5: traversing a repair material database according to the update strategy instruction, matching a candidate material set that satisfies the compressive strength and the adhesion coefficient, calculating a cosine similarity of the candidate material and the evolution trajectory prediction vector by combining real-time collected occlusal force monitoring data, and generating a repair scheme.

[0041] Compared with the prior art, the advantages and positive effects of the present application are that:

[0042] In the present application, the noise is filtered out by real-time transmission of the three-dimensional scanning data stream of the oral cavity and then using the Gaussian filtering algorithm, the environmental interference and equipment error are reduced in the data preprocessing stage, the signal-to-noise ratio of the original data is improved, and the misjudgment caused by noise in subsequent analysis is avoided. Based on the edge detection algorithm, the contour boundary is extracted and the edge sharpness and curvature parameters are calculated, the geometric features of the lesion area are quantified, and the recognition accuracy of irregular defect morphology is enhanced. The convolution neural network model is used to fuse the defect area, depth value and structure damage index to generate a repair accuracy coefficient, establish a dynamic mapping relationship between multi-dimensional indicators and clinical repair strategy, and improve the scheme adaptability. Through the LSTM time series model, the evolution trend of the defect depth change rate and the area growth rate is predicted, the lesion development trajectory is predicted, the repair parameters are dynamically adjusted, and the failure of the traditional static scheme caused by uncontrollable development is avoided. The technical means of each link forms a closed-loop logic chain, optimizes the data processing efficiency and the scientific nature of decision-making. BRIEF DESCRIPTION OF DRAWINGS

[0043] Figure 1 The system flowchart of the present application is shown in the figure;

[0044] Figure 2 The image acquisition module flowchart of the present application is shown in the figure;

[0045] Figure 3 The feature analysis module flowchart of the present application is shown in the figure;

[0046] Figure 4 The strategy generation module flowchart of the present application is shown in the figure;

[0047] Figure 5 The trajectory prediction module flowchart of the present application is shown in the figure;

[0048] Figure 6 The scheme feedback module flowchart of the present application is shown in the figure. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical scheme and advantages of the present application clearer and more understandable, the present application will be further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and do not limit the present application.

[0050] In the description of the present application, it should be understood that the terms "length", "width", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. In addition, in the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise explicitly and specifically limited.

[0051] Please refer to Figure 1 The present application provides a technical solution: a remote oral prosthetic diagnosis and treatment system based on a 5G network, comprising:

[0052] An image acquisition module is configured to receive oral three-dimensional scanning data stream through 5G real-time transmission bandwidth, call Gaussian filtering algorithm to filter out noise, extract lesion area coordinate set, generate three-dimensional reconstruction data, and deliver the three-dimensional reconstruction data to a feature analysis module;

[0053] The feature analysis module is configured to receive the three-dimensional reconstruction data, locate the contour boundary based on an edge detection algorithm, calculate the edge sharpness value and the curvature parameter, generate the lesion feature parameter, and deliver the lesion feature parameter and the edge sharpness value to a strategy generation module.

[0054] Edge sharpness value: refers to the clarity of the edge part of the image, which is usually calculated by image gradient, and measures the distinctness of the edge and the steepness of the turning point.

[0055] Curvature parameter: describes the bending degree of a curve at a certain point, and is usually used to judge the change of surface shape.

[0056] The strategy generation module is configured to receive the lesion feature parameter and the edge sharpness value, input them into a convolutional neural network model for lesion classification, output the defect area, depth value and structure damage index, generate a repair precision coefficient based on the product, map it to a repair technology number, generate a dynamic repair strategy table, and deliver the dynamic repair strategy table and the depth value to a trajectory prediction module.

[0057] Defect area: represents the surface area of the lesion area in three-dimensional space, which is usually obtained by geometric calculation of the point cloud data of the lesion area.

[0058] Depth value: the depth of the lesion area from the surface to the base, which is usually measured by the spatial coordinates in the three-dimensional reconstruction data.

[0059] Structure damage index: a numerical value that evaluates the structural integrity of the lesion site, which is calculated based on the comparison of the feature parameters and the adjacent regions.

[0060] The trajectory prediction module is configured to receive the dynamic repair strategy table and the depth value, call an LSTM time series model to model the depth change rate, combine the defect area growth rate to generate an evolution trajectory prediction vector, output an updated strategy instruction, and deliver the updated strategy instruction to the scheme feedback module.

[0061] The scheme feedback module is configured to receive the updated strategy instruction, match the repair material database, extract the compressive strength and adhesion coefficient, calculate the adaptability in combination with real-time occlusion force monitoring data, generate a repair scheme, and synchronize the repair scheme to the patient-side wearable sensor.

[0062] Compressive strength: the ability of a material to resist external pressure, commonly measured by compression tests, and a higher value indicates that the material is more durable.

[0063] Adhesion coefficient: describes the adhesion of a material on a bonding surface, usually obtained by inherent values from shear tests or tensile tests.

[0064] The three-dimensional reconstruction data includes a set of spatial coordinates of the lesion area, surface point cloud data, and denoised texture feature data. The lesion feature parameters include edge sharpness distribution, curvature change matrix, and morphological complexity index. The dynamic repair strategy table includes repair technology number, repair priority, and estimated repair period. The evolution trajectory prediction vector includes depth change rate sequence, area expansion rate curve, and evolution time node marker. The repair scheme includes material adaptability parameters, compressive strength value, and adhesion performance index.

[0065] Edge sharpness distribution: represents the distribution of edge sharpness in different regions of the image, providing detailed information about the lesion boundary.

[0066] Curvature change matrix: obtained by analyzing the point cloud of the lesion area surface, reflecting the complexity of the surface.

[0067] Morphological complexity index: used to quantitatively evaluate the complexity of the lesion shape, usually combined with boundary length, area, and other parameters.

[0068] Please refer to Figure 2 , the image acquisition module includes:

[0069] The 5G transmission submodule collects real-time oral three-dimensional scanning data streams, establishes a communication link through a 5G transmission protocol, divides the data streams into independent data packets and adds time stamps, completes data reorganization according to the transmission sequence number, and generates a transmission data stream.

[0070] 5G transmission sub-module collects real-time oral three-dimensional scanning data stream. First, a handheld intraoral scanner is used, such as using structured light or confocal laser technology, to scan the inside of the patient's mouth (e.g., tooth number 26 with distal proximal interproximal caries). The probe moves on the tooth surface, and a series of point cloud data containing spatial coordinates (x, y, z) and possible color or grayscale information (c) are acquired in real time, forming raw data points, for example, tens of thousands of points P1(10.1, 20.3, 5.5, 0.7), P2(10.2, 20.4, 5.6, 0.75), P3(10.1, 20.5, 5.5, 0.72) are collected in the first second, forming an initial data stream. Subsequently, a 5G communication chip integrated in the scanning device or connected to the computing unit is used to start the 5G communication protocol stack with the remote server or edge computing node, and a communication link based on TCP / IP or UDP protocol more suitable for real-time streaming is established. The data stream is dynamically divided into data packets of appropriate size, for example, set each data packet to carry 1000 data points, then the first 1000 points constitute data packet #1, the next 1000 points constitute data packet #2, and so on. When packaging, a time stamp accurate to milliseconds is attached to each data packet, such as data packet #1 timestamp 2025-04-28 10:30:01.123, data packet #2 timestamp 2025-04-28 10:30:01.256, and each data packet is assigned a strictly increasing transmission sequence number, such as data packet #1 sequence number SN=1, data packet #2 sequence number SN=2. At the receiving end, due to network fluctuations, data packets may arrive out of order (e.g., SN=2 is received first, then SN=1). At this time, the received data packets are sorted and buffered according to the sequence number in the data packet, and the data is recombined in the order of SN=1, SN=2, SN=3… to form a continuous data point sequence, for example, data points P1 to P1000 of SN=1 and data points P1001 to P2000 of SN=2 are connected in order to generate a transmission data stream.

[0071] The filter denoising sub-module calls the Gaussian filter algorithm, constructs a three-dimensional coordinate system based on the transmission data stream, and performs neighborhood weighted average calculation on each coordinate point. The standard deviation parameter is set to 0.8 to eliminate the coordinate offset of outliers, and the filtered data is generated.

[0072] The filtering and noise reduction submodule receives the transmitted data stream, which is a sequence of three-dimensional coordinate points, such as P1(10.1,20.3,5.5), P2(10.2,20.4,5.6), ..., Pn(x_n,y_n,z_n). First, a three-dimensional spatial coordinate system is established based on the coordinate range of these points. Then, each coordinate point Pi(xi,yi,zi) in the data stream is processed, and a neighborhood is defined around it, for example, a spherical region with a radius of 1.5mm centered on Pi. All neighboring points Pj, Pk, ... within this region are found, and the Euclidean distance between Pi and each neighboring point Pk is calculated. Based on the set standard deviation parameter mm, calculate Gaussian weights ,this The values ​​are empirical values ​​derived from the analysis of noise characteristics of a large amount of oral scan data, aiming to balance denoising effect and detail preservation. For example, if point Pj is 1.0mm away from Pi, then its weight is... If the distance between point Pk and Pi is 0.5mm, then its weight... For Pi itself, the distance is 0, and the weight is... Next, calculate the new coordinates of Pi. Its calculation method is the weighted average of the coordinates of all points in the neighborhood (including Pi itself). , , For example, if the neighborhood of Pi(10.0,20.0,5.0) contains only Pj(11.0,20.2,5.1) and Pk(9.8,19.9,4.9), and the calculated distance and weights are as in the example above, then... Calculate the same By comparing the offsets of the original coordinates Pi and the new coordinates P'i The system identifies and processes points whose offsets are much greater than the average offset of the neighborhood. These points are considered outliers, and their coordinates are smoothly adjusted to eliminate drastic deviations in the coordinates of isolated points caused by scanning noise or artifacts, thereby generating filtered data.

[0073] The lesion coordinate generation submodule performs gradient calculation based on the filtered data, performs differential operation on the gray values ​​of adjacent coordinate points, selects regions with gray value differences exceeding 0.12 as boundary judgment benchmarks, extracts the vertex coordinate set of continuous closed regions, and generates three-dimensional reconstruction data.

[0074] The lesion coordinate generation submodule receives the filtered data, which is a smoother three-dimensional point cloud. For each point in the point cloud, if the scanning data contains grayscale information (e.g., reflecting surface reflectivity or material density), the grayscale value is directly used; if not, a pseudo-grayscale value can be estimated based on local point density or surface normal variation. Then, the difference operation of the grayscale value is performed on adjacent coordinate points in space, for example, considering the point Pi (xi, yi, zi) and its immediately adjacent points in the x, y, and z directions Px+(xi+1, yi, zi), Py+(xi, yi+1, zi), and Pz+(xi, yi, zi+1), the grayscale difference , , is calculated, and the gradient vector is obtained, whose amplitude approximately represents the grayscale change intensity. The gradient amplitude is compared with the preset boundary judgment reference value 0.12, which is set based on the statistical analysis of the grayscale / density difference between a large number of known healthy dental tissues and carious tissues (such as enamel demineralization or dentin caries) under a specific scanning mode, to distinguish between normal structural changes and pathological boundaries. For example, if the grayscale value of point Pa is 0.85 and the grayscale value of its adjacent point Pb is 0.70, the difference is 0.15, which is greater than 0.12, then the boundary between Pa and Pb is marked as part of the potential lesion boundary. If the difference between another pair of adjacent points Pc (0.80) and Pd (0.75) is 0.05, which is less than 0.12, then it is not marked as a boundary. All boundary segments composed of adjacent point pairs with a grayscale difference greater than 0.12 are screened out. Then, using region growing or contour tracking, these continuous boundary segments are connected to find boundary point sets that can form closed loops. For example, starting from a boundary point that meets the conditions, search for adjacent boundary points that also meet the conditions, and continue to connect until returning to the starting point or unable to continue connecting, forming a closed region. Extract the three-dimensional coordinate set of all vertices (i.e., boundary points) that constitute these continuous closed regions, such as the set {(x1, y1, z1), (x2, y2, z2), …, (xm, ym, zm)}. These coordinate points collectively outline the contour of the potential lesion area, generating three-dimensional reconstruction data.

[0075] Please refer to Figure 3 , the feature analysis module includes:

[0076] The edge detection submodule calls an edge detection algorithm based on the three-dimensional reconstruction data, calculates the gradient amplitude and direction for each coordinate point, sets high and low threshold values of 0.3 and 0.1 respectively, selects points with a gradient amplitude higher than the high threshold value as boundary points, connects adjacent points to form a closed contour, and generates a boundary coordinate set;

[0077] The edge detection submodule receives the three-dimensional reconstruction data, which is a set of vertex coordinates {(x1, y1, z1),..., (xm, ym, zm)} defining the potential lesion boundary. For each coordinate point Pi in the set, the gradient information in its local neighborhood is calculated again, this time focusing more on the change in spatial coordinates or the associated gradient with the original filtered data gray scale / density. The gradient magnitude and the gradient direction are calculated, for example, using a three-dimensional Sobel operator acting on the coordinates or gray scale values around Pi to obtain the gradient vector The direction of this vector is denoted as and a high threshold and a low threshold These two thresholds are determined by histogram analysis of the gradient magnitude distribution of known boundary points and non-boundary points, for example, by selecting a value such that most strong edges (such as cavity edges) have a gradient magnitude higher than and most weak edges and noise are lower than All points with are marked as strong boundary points, for example, if the point Pk has then it is a weak boundary point (between and ), and if Pq has it is ignored. Starting from a strong boundary point, it is connected to its neighbors that also satisfy the boundary condition (i.e., ), where "neighbor" not only means spatial proximity but also requires similar gradient direction, for example, an angle difference less than 30 degrees. The connection is continued until a closed contour is formed by strong boundary points and weak boundary points connected by strong boundary points. Those contours that are entirely composed of weak boundary points or are not closed are discarded, and the boundary coordinate set is generated.

[0078] The sharpness curvature calculation submodule calls the boundary coordinate set and calculates the second derivative for the coordinates in the neighborhood of each boundary point to extract the maximum curvature absolute value as the curvature parameter. At the same time, the gradient magnitude is normalized, and the average of the gradient direction change rate of adjacent points is calculated to generate the edge sharpness value and the curvature parameter.

[0079] The sharpness curvature calculation submodule receives the boundary coordinate set, which is an ordered set of points P1, P2,..., Pk forming a closed contour. For each boundary point Pi, its previous and next neighboring points Pi-1 and Pi+1 on the contour are examined (for the first and last points P1 and Pk, ring connection processing is performed, i.e., the previous point of P1 is Pk, and the next point of Pk is P1). The second derivative is calculated using the coordinates of these three points to approximate the local curvature. One way to calculate is​​ Where ||…|| represents the vector norm (length), the curvature value of point Pi is calculated. Then, within a small neighborhood of Pi (e.g., including Pi-2, Pi-1, Pi, Pi+1, Pi+2), the maximum absolute value of these curvature values ​​is found, and this maximum value is used as the curvature parameter of Pi. For example, if the curvatures of points near Pi are calculated to be -1.2, -1.8, 1.5, and 0.9 mm, then the curvature parameter of Pi is... Simultaneously, retrieve the gradient magnitude of each boundary point Pi calculated in the previous step. For all boundary points The values ​​are normalized, for example, linearly normalized to the [0,1] interval: ,in and The minimum and maximum values ​​of the gradient magnitudes at all boundary points are given. Next, the rate of change of the gradient direction between adjacent points is calculated. For point Pi, its gradient direction vector is calculated. Compared to the previous point And the next point The angle difference between and The average of these two angle differences is taken as the edge sharpness value of point Pi. For example, if , , The calculated angle differences are respectively and Then the sharpness value (Or use radians) to generate edge sharpness values ​​and curvature parameters for each boundary point.

[0080] The feature parameter generation submodule performs weighted fusion calculations on regions with sharpness values ​​exceeding 0.25 based on edge sharpness values ​​and curvature parameters. The curvature weight coefficient is set to 0.6 and the sharpness weight coefficient is set to 0.4. The module integrates the point parameters within the region to generate the average value and standard deviation, thus obtaining the lesion feature parameters.

[0081] The feature parameter generation submodule receives the edge sharpness value. With the curvature parameter (For all boundary points i=1tok), first filter out regions whose sharpness values ​​exceed a specific threshold, and set the sharpness threshold to 1. (The unit here must be consistent with the sharpness calculation. If sharpness is expressed in angles, it is, for example, 15 degrees; if a normalized value is used, it is 0.25.) This threshold is based on the distribution analysis of sharpness values ​​for different dental lesions (such as sharp cavity edges vs. relatively gentle wear), selecting values ​​that can better distinguish sharp features and identifying all those that meet the criteria. of the continuous boundary point segments, constituting high sharpness regions, for each point Pi located within these high sharpness regions, perform a weighted fusion calculation, set the curvature weight coefficient and the sharpness weight coefficient , the two weight coefficients reflect the relative importance of curvature (morphological complexity) and sharpness (edge clarity / saliency) in assessing lesion severity or characteristics, according to clinical experience or data analysis, it is believed that the complexity of the morphology (reflected by the curvature) may be more important than the sharpness of the edge in some diagnostic scenarios, so a higher weight 0.6 is given to the curvature, the weight setting can be determined by optimizing on the training set to maximize the correlation between the generated feature parameters and the clinical diagnosis results, calculate the fusion value for each point Pi in the region , for example, a high sharpness region contains points P1 and P2, with parameters P1 and P2 , then , , then integrate the fusion parameters of all points in the region , calculate the average value and the standard deviation of these values, where N is the number of points in the region, continuing the above example, , , to obtain the lesion feature parameters.

[0082] Please refer to Figure 4 , the strategy generation module includes:

[0083] The classification calculation submodule inputs the parameter matrix into the fully connected layer of the convolutional neural network based on the lesion feature parameters and the edge sharpness value, extracts a multi-dimensional feature vector, performs a linear superposition operation of the feature vector and the weight matrix, outputs a probability distribution through an activation function, and obtains the defect area, depth value and structure damage index;

[0084] The classification calculation submodule receives the lesion feature parameters (average fusion value and standard deviation ) and may also include global edge sharpness statistics (such as the average sharpness of the entire boundary ), combines these parameters into a feature vector , which is input into the input layer of a pre-trained convolutional neural network (here referring to the fully connected layer part of its structure), through the forward propagation process of the network, the feature vector first passes through the fully connected layer, performs a linear superposition operation with the weight matrix of the layer, and adds a bias term , that is For example, if the fully connected layer has 3 output nodes corresponding to the preliminary estimates of the lesion area, depth value and structural damage index, the weight matrix is a matrix, the bias is a vector, and the calculation process is , , where and are the parameters obtained by network training, and the calculated vector is then passed through an activation function. If the output is a continuous value (such as area or depth), a ReLU or linear activation function or may be used; if it is a classification probability (such as damage index classification), a Softmax function is used. Assuming that corresponds to the area and depth, a linear activation is used to obtain the preliminary estimate values, for example, the area estimate is 14.5 (mm²) and the depth estimate is 2.8 (mm). Assuming that and its related output are used for the structural damage index (classified into three levels: mild, moderate, and severe), the probability distribution of each level is output by Softmax, for example, P(mild) = 0.1, P(moderate) = 0.6, and P(severe) = 0.3. Then, the moderate level with the highest probability is selected as the structural damage index, and the lesion area (14.5 mm²), depth value (2.8 mm), and structural damage index (moderate) are obtained.

[0085] The coefficient generation submodule performs unit area standardization on the lesion area, converts the depth value into a relative depth ratio, and performs a product operation on the three values, and introduces a weight correction factor to adjust the product result to generate a repair accuracy coefficient.

[0086] The coefficient generation submodule receives the lesion area mm², the depth value mm, and the structural damage index $Index = moderate obtained in the previous step. First, the lesion area is standardized by unit area, for example, if 1 mm² is taken as the standard unit area mm², the standardized area (dimensionless) is obtained. Then, the depth value is converted into a relative depth ratio, which requires a reference maximum depth , which can be set based on the tooth type and anatomical structure, for example, for premolars, the maximum depth that caries can reach (before reaching the pulp cavity) may be estimated as 4.5 mm, so the relative depth is obtained. Then, the structural damage index is quantified, with values set as mild = 1, moderate = 2, and severe = 3, so The three quantitative indicators are multiplied: A weight correction factor is introduced to adjust the product result, the setting of the correction factor may be based on patient age, tooth position (such as posterior teeth bearing force is large, may be slightly higher), whether there are complications and other factors, for example, for a middle-aged patient's posterior teeth (such as tooth 26), may be set to 1.15, this value is determined by regression analysis on the historical case database, find out the adjustment value that can make the final coefficient better predict the complexity of repair and prognosis, means that the basic risk assessment value needs to be magnified by 15% for this kind of situation, calculate the final repair precision coefficient , generate the repair precision coefficient.

[0087] The strategy mapping submodule matches the corresponding technical number in the preset mapping table according to the numerical interval of the repair precision coefficient, divides the depth value into grades with 0.1mm precision, integrates the number and depth grade, and establishes a dynamic repair strategy table.

[0088] The strategy mapping submodule receives the repair precision coefficient , and according to the value, finds the corresponding interval in a preset technical mapping table, which establishes the corresponding relationship between the repair precision coefficient value domain and the specific repair technology number.

[0089] Table 1 Repair strategy mapping table

[0090] ;

[0091] As shown in Table 1, the repair precision coefficient falls within the interval of 10.01-25, so the matching technical number is T02, and the corresponding technical type is direct composite resin filling. At the same time, the previously calculated depth value mm is divided into grades with 0.1mm precision, the depth grade is calculated, and the technical number T02 and the depth grade 28 are integrated, and the pair of information {technical number: T02, depth grade: 28} is added to or updated to the dynamic repair strategy table of the case. The table records the disease assessment and recommended strategy over time, for example, the table entry may include {timestamp: 2025-04-28 11:00:00, case ID: P123, technical number: T02, depth grade: 28}, and a dynamic repair strategy table is established.

[0092] Please refer to Figure 5 , the trajectory prediction module includes:

[0093] The deep modeling submodule calls the depth value sequence in the dynamic repair strategy table, inputs it into the LSTM time series model, calculates the time weight through the forget gate and the input gate, performs sliding window segmentation on the depth value, outputs the hidden layer state vector, and generates the depth change rate.

[0094] The depth modeling submodule calls a series of depth values ​​recorded in the dynamic repair strategy table for a specific lesion (e.g., tooth 26 in case P123). If the lesion has been monitored multiple times, a time series is formed, for example, the depth values ​​(in mm) of the past three examinations are respectively... The current value is 2.8mm. Sequence this depth value... The input is processed into a Long Short-Term Memory (LSTM) unit structure. Inside the LSTM unit, for each time point in the sequence... depth value It will be in the same hidden state as the previous moment. and unit state Together they participate in the calculation, first passing through the forget gate. The calculation determines the state of the unit. How much information is discarded, and what is the weight? and bias These are learning parameters. It's the sigmoid function, then calculated through the input gate. Decide which new information to update, and the status of candidate units. Prepare the information to add, then update the cell status. ( (representing element-wise product), finally passed through the output gate. Calculate and determine the hidden state of the output This process is performed stepwise on the sequence [2.1, 2.4, 2.8], ultimately yielding the hidden layer state vector. (T being the end of the sequence) contains dynamic information about the sequence, and Through a linear layer (or directly using) (Partial dimensions) are used to predict the depth change at the next time step, i.e., the depth change rate. For example, based on the growth trend of the sequence [2.1, 2.4, 2.8] (intervals of 0.3 mm and 0.4 mm respectively), the model may predict the rate of change in depth for the next time unit (e.g., the next 3 months). mm / unit time, generating depth change rate.

[0095] The trajectory generation submodule calculates the difference between adjacent time points of the defect area based on the depth change rate, fits the area growth curve through linear regression, extracts the slope parameter as the growth rate, and splices the change rate and the growth rate to generate an evolution trajectory prediction vector;

[0096] The trajectory generation submodule receives the depth change rate mm / unit time, and obtains a defect area sequence corresponding to the depth sequence, for example mm2, calculates the difference between adjacent time points of the defect area sequence to obtain the area change mm2, mm2, in order to fit the area growth trend, the area value is linearly regressed with the corresponding time point (or sequence index 1, 2, 3) to establish a model , where is the time point or index, and the slope is calculated by the least square method, for example, the slope calculated by linear regression of points (1, 12.5), (2, 13.8), and (3, 14.5) is , i.e., the area growth rate may be mm2 / unit time, and the calculated depth change rate and the area growth rate are spliced to form a two-dimensional evolution trajectory prediction vector , which represents the predicted depth and area change of the lesion in the future unit of time, and generates an evolution trajectory prediction vector.

[0097] The strategy updating submodule performs matching operations of the vector and a preset strategy rule according to the dimension characteristics of the evolution trajectory prediction vector, filters candidate strategies that meet the error threshold, marks the optimal strategy as a priority, and outputs an updated strategy instruction.

[0098] The strategy updating submodule receives the evolution trajectory prediction vector , performs matching operations of the vector and a series of preset strategy adjustment rules according to the dimension characteristics (i.e., the depth change rate and the area growth rate ) of the vector, and these rule libraries define strategy adjustment suggestions corresponding to different prediction trajectory intervals, for example, the rule library may include: rule A: if and , then the candidate strategy is {shorten the recheck period to 3 months, consider upgrading the repair material grade}; rule B: if If the candidate strategy is {immediate re-visit, evaluate if more invasive treatment like root canal preparation is needed, upgrade the restoration plan to onlay / high onlay}, rule C: if and then the candidate strategy is {maintain current observation / restoration plan, re-visit period can be extended appropriately}, the current prediction vector is matched with these rules, it is found that it satisfies the condition of rule A ( and ), therefore, the filtered candidate strategy set is {shorten the re-visit period to 3 months, consider upgrading the restoration material grade}, next, the candidate strategies need to be evaluated to choose the optimal strategy, which may involve evaluating the expected treatment effect, risk, cost, and other factors under different strategies, set an error threshold or risk score standard, for example, if "upgrade the material grade" can reduce the predicted long-term failure rate by more than 10% (below the error threshold), it is considered better than "shorten the re-visit period", assuming that the evaluation result shows that "upgrade the restoration material grade" combined with "shorten the re-visit period" is the optimal choice, then mark the combined strategy containing the two actions as the highest priority, generate an updated strategy instruction containing specific instructions, such as {instruction type: strategy update, case ID: P123, recommended operation: [shorten the re-visit period to 3 months, upgrade the restoration material by one grade]}, output the updated strategy instruction.

[0099] Please refer to Figure 6 , the scheme feedback module includes:

[0100] The material matching sub-module calls the restoration technology number in the updated strategy instruction, traverses the restoration material database primary key based on the number hash value, extracts the compressive strength and adhesion coefficient fields after verifying the entry validity, and generates a material parameter set;

[0101] The material matching sub-module receives the update policy instruction, such as the instruction requiring {repair material to be upgraded by one level}, and combines the current recommended repair technology number (assuming it is still T02, but a higher performance composite resin needs to be selected). The instruction triggers a query of the repair material database. First, according to the technology number T02 (direct composite resin filling) and the new requirement (for example, the original A2 level material needs to be upgraded to A1 or higher strength level), the query condition is determined, and the technology number T02 or its hash value and the material level or performance requirement (such as compressive strength range) are used as the primary key or index to search for matching repair material entries in the database. For example, there are material records M1 {ID:RCM005,Name:BrandXFlowableA2,Type:T02,Strength:150MPa,Adhesion:0.7}, M2 {ID:RCM010,Name:BrandYUniversalA1,Type:T02,Strength:300MPa,Adhesion:0.85}, M3 {ID:RCM015,Name:BrandZHighStrengthBody,Type:T02,Strength:380MPa,Adhesion:0.90} in the database. The program verifies the validity of these entries (such as whether they are within the valid period and whether they meet regional regulations), and then extracts the key performance parameter fields, mainly the compressive strength (Compressive Strength) and adhesion coefficient (Adhesion Coefficient, which may represent the bonding strength or an index related to the bonding ability of the tooth tissue), assuming that M2 and M3 are valid and meet the upgrade requirements (strength higher than M1), the extracted parameter set is [{Material:M2,Strength:300MPa,Adhesion:0.85},{Material:M3,Strength:380MPa,Adhesion:0.90}], and the material parameter set is generated.

[0102] The adaptation degree calculation sub-module takes the compressive strength of the material parameter set as the numerator and the real-time bite force monitoring data as the denominator to perform division operation, extracts the product of the adhesion coefficient and the bite force change rate, and adds the weight coefficient after standardizing the two results to generate the adaptation degree coefficient;

[0103] The adaptation degree calculation sub-module receives the material parameter set [{M2,300MPa,0.85},{M3,380MPa,0.90}] generated in the previous step, and obtains the real-time or recent monitoring data of the patient's bite force, for example, the maximum masticatory force of the patient on the repair side is measured by a bite force sensor N, the average change rate of bite force during mastication N / s, first, the occlusal force needs to be converted to stress, which requires estimating the stress area of the restoration , assuming that for T02 restorations, the average contact area is about mm², then the maximum occlusal stress MPa, for each candidate material, calculate two indicators: 1. Strength ratio = material compressive strength / maximum occlusal stress - for M2: - for M3: 2. Adhesion-dynamic term = material adhesion coefficient occlusal force change rate (assuming the adhesion coefficient is in the range of 0-1, the change rate unit is N / s) - for M2: - for M3: Next, normalize these two indicators, assuming that according to a large amount of data statistics, the typical range of strength ratio is [1.5, 5.0], and the typical range of adhesion-dynamic term is [30, 100], linearly normalized to the [0, 1] interval:- - M2: - M3: - - M2: - M3: Set weight coefficients, for example, for posterior teeth restorations that need to bear larger occlusal forces, compressive strength may be more important, set the compressive strength weight , the adhesion-related factor weight These two weights are determined by clinical experts according to the importance of restoration type and location, and , calculate the final adaptation coefficient : - M2: - M3: Generate the adaptation coefficient.

[0104] The scheme generation submodule divides the adaptation level interval according to the numerical distribution of the adaptation coefficient, filters the material combination that meets the interval requirements of compressive strength and adhesion coefficient, aligns the occlusal force data and material parameters according to the time sequence to generate the index matrix, and establishes the restoration scheme.

[0105] The scheme generation submodule receives the adaptation coefficient of each candidate material, M2: , M3: , according to the numerical distribution of these coefficients, the adaptation level interval is demarcated, for example: [0-0.3): not recommended; [0.3-0.5): acceptable; [0.5-0.7): good; [0.7-1.0]: excellent, accordingly, material M2(0.426) belongs to the "acceptable" level, material M3(0.592) belongs to the "good" level, screen out those materials whose adaptation level reaches the preset minimum requirement (for example "acceptable" or higher), and their original compressive strength and adhesion coefficient must also meet the basic requirements of the repair technology (T02, direct composite resin filling), assuming that the basic requirements of T02 are Strength>250MPa and Adhesion>0.8, M2(300MPa,0.85) and M3(380MPa,0.90) both meet, then both are candidate materials, but M3 has a higher adaptation degree, when generating the final scheme, the combination of materials with the highest adaptation degree is usually recommended first (if multiple materials are needed), in this example, material M3(Brand Z High Strength Body) is recommended first, then the selected material information (such as the ID of M3, RCM015), the timestamp or characteristics of the occlusal force data used when calculating the adaptation degree (such as N, N / s@2025-04-2811:30), the key parameters of the material (Strength=380MPa, Adhesion=0.90) and the adaptation degree score (0.592) are aligned and recorded, and an index matrix or structured record containing these information is created to finally determine and record the detailed repair scheme, for example, the scheme record is {Case ID: P123, Tooth Position: 26, Recommended Technology: T02, Recommended Material: M3(RCM015), Adaptation Degree: 0.592(Good), Key Parameters: {Strength: 380MPa, Adhesion: 0.90}, Reference Occlusal Force: {Max: 600N, Rate: 70N / s}}, and the repair scheme is established.

[0106] A remote dental restoration diagnosis and treatment method based on a 5G network, the remote dental restoration diagnosis and treatment method based on a 5G network is executed based on the above-mentioned remote dental restoration diagnosis and treatment system based on a 5G network, comprising the following steps:

[0107] S1: receiving real-time transmission of three-dimensional scanning data stream of oral cavity through 5G communication protocol, calling Gaussian filtering algorithm to filter out noise of data stream, extracting lesion area coordinate set, and generating three-dimensional reconstruction data;

[0108] S2: inputting the three-dimensional reconstruction data into an edge detection algorithm to locate the contour boundary, calculating the edge sharpness value generated by the boundary pixel gradient amplitude, and combining the curvature difference of adjacent pixels to generate lesion feature parameters;

[0109] S3: input the lesion feature parameters and the edge sharpness value into the convolutional neural network model for classification training, output the defect area, the depth value and the structure damage index, multiply the depth value and the structure damage index to generate a repair accuracy coefficient, match the repair accuracy coefficient with a preset repair technology mapping table to generate a dynamic repair strategy table;

[0110] S4: call an LSTM time series model to predict the depth value change rate based on the dynamic repair strategy table and the depth value, generate an evolution trajectory prediction vector combined with the adjacent time stamp difference of the defect area, and output an update strategy instruction containing a correction parameter;

[0111] S5: traverse the repair material database according to the update strategy instruction, match a candidate material set that meets the compressive strength and the adhesion coefficient, calculate the cosine similarity of the candidate material and the evolution trajectory prediction vector combined with the real-time collected occlusion force monitoring data, and generate a repair scheme.

[0112] The above is only a preferred embodiment of the present application, and does not limit the present application in other forms. Any person skilled in the art can use the disclosed technical content to make changes or modifications as equivalent embodiments applied to other fields, but any simple modification, equivalent change and modification made on the basis of the technical essence of the present application to the above embodiments without departing from the technical solution content of the present application shall still fall within the protection scope of the present application.

Claims

1. A remote dental prosthetic diagnosis and treatment system based on a 5G network, characterized in that, The system comprises: An image acquisition module is configured to receive a three-dimensional scanning data stream of an oral cavity through a 5G real-time transmission bandwidth interface, call a Gaussian filtering algorithm to filter out noise, extract a set of coordinates of a lesion area, generate three-dimensional reconstruction data, and deliver the three-dimensional reconstruction data to a feature analysis module; The feature analysis module is configured to receive the three-dimensional reconstruction data, locate a contour boundary based on an edge detection algorithm, calculate an edge sharpness value and a curvature parameter, generate lesion feature parameters, and deliver the lesion feature parameters and the edge sharpness value to a strategy generation module; The strategy generation module is configured to receive the lesion feature parameters and the edge sharpness value, input the lesion feature parameters and the edge sharpness value into a convolutional neural network model for lesion classification, output a defect area, a depth value, and a structure damage index, generate a repair accuracy coefficient based on a product, map the repair accuracy coefficient to a repair technology number, generate a dynamic repair strategy table, and deliver the dynamic repair strategy table and the depth value to a trajectory prediction module; The trajectory prediction module is configured to receive the dynamic repair strategy table and the depth value, call an LSTM time series model to model a depth change rate, combine a defect area growth rate to generate an evolution trajectory prediction vector, output an updated strategy instruction, and deliver the updated strategy instruction to a scheme feedback module; The feature analysis module comprises: An edge detection submodule is configured to call an edge detection algorithm based on the three-dimensional reconstruction data, calculate a gradient amplitude and a direction for each coordinate point, set a high threshold and a low threshold to be 0.3 and 0.1 respectively, filter out points with a gradient amplitude higher than the high threshold as boundary points, connect adjacent points to form a closed contour, and generate a boundary coordinate set; A sharpness curvature calculation submodule is configured to call the boundary coordinate set, calculate a second derivative for coordinates in a neighborhood of each boundary point, extract a maximum curvature absolute value as a curvature parameter, normalize a gradient amplitude, calculate an average of gradient direction change rates of adjacent points, and generate an edge sharpness value and a curvature parameter; A feature parameter generation submodule is configured to call the edge sharpness value and the curvature parameter, perform weighted fusion calculation on a region with a sharpness value higher than 0.25, set a curvature weight coefficient to be 0.6 and a sharpness weight coefficient to be 0.4, integrate all point parameters in the region to generate an average value and a standard deviation, and obtain lesion feature parameters; The system further comprises: A scheme feedback module is configured to receive the updated strategy instruction, match a repair material database, extract a compressive strength and an adhesion coefficient, calculate an adaptation degree based on real-time bite force monitoring data, generate a repair scheme, and synchronize the repair scheme to a wearable sensor of a patient end; The repair scheme comprises a material adaptation degree parameter, a compressive strength value, and an adhesion performance index; The scheme feedback module comprises: A material matching submodule is configured to call a repair technology number in the updated strategy instruction, traverse a primary key of a repair material database based on a number hash value, extract a compressive strength and an adhesion coefficient field after verifying validity of an entry, and generate a material parameter set; An adaptation degree calculation submodule is configured to take the compressive strength of the material parameter set as a numerator, perform a division operation with real-time bite force monitoring data as a denominator, extract a product of an adhesion coefficient and a bite force change rate, and superimpose a weight coefficient on results of the two items to generate an adaptation degree coefficient. The scheme generation sub-module divides an adaptation level interval according to a numerical distribution of the adaptation coefficients, screens a material combination that satisfies interval requirements of both the compressive strength and the adhesion coefficient, aligns occlusal force data and material parameters in a time sequence to generate an index matrix, and establishes a repair scheme. 2.The 5G network-based remote dental prosthetic diagnosis and treatment system of claim 1, wherein, The three-dimensional reconstruction data include a lesion region spatial coordinate set, surface point cloud data, and denoised texture feature data, the lesion feature parameters include edge sharpness distribution, curvature change matrix, and morphological complexity index, the dynamic repair strategy table includes repair technology number, repair priority, and estimated repair period, and the evolution trajectory prediction vector includes depth change rate sequence, area expansion rate curve, and evolution time node marker. 3.The 5G network-based remote dental prosthetic diagnosis and treatment system of claim 2, wherein, The image acquisition module includes: The 5G transmission sub-module collects real-time oral cavity three-dimensional scanning data stream, establishes a communication link through a 5G transmission protocol, divides the data stream into independent data packets and adds time stamps, completes data reorganization according to a transmission serial number, and generates a transmission data stream; The filtering and denoising sub-module calls a Gaussian filtering algorithm, constructs a three-dimensional spatial coordinate system based on the transmission data stream, performs neighborhood weighted average calculation on each coordinate point, sets a standard deviation parameter to 0.8, eliminates outlier coordinate offset, and generates filtered data; The lesion coordinate generation sub-module performs gradient calculation based on the filtered data, performs difference operation on gray values of adjacent coordinate points, selects a region with a gray difference value greater than 0.12 as a boundary judgment reference, extracts a vertex coordinate set of a continuous closed region, and generates three-dimensional reconstruction data. 4.The 5G network-based remote dental prosthetic diagnosis and treatment system of claim 1, wherein, The strategy generation module includes: The classification calculation sub-module inputs a parameter matrix into a fully connected layer of a convolutional neural network based on the lesion feature parameters and the edge sharpness value, extracts a multi-dimensional feature vector, performs linear superposition operation of the feature vector and a weight matrix, outputs a probability distribution through an activation function, and obtains a defect area, a depth value, and a structure damage index; The coefficient generation sub-module performs unit area standardization on the defect area, converts the depth value into a relative depth ratio, performs product operation of the three in combination with the structure damage index, introduces a weight correction factor to adjust the product result, and generates a repair precision coefficient; The strategy mapping sub-module matches a corresponding technology number in a preset mapping table according to a numerical interval of the repair precision coefficient, divides the depth value into levels at a precision of 0.1 mm, integrates the number and the depth level, and establishes a dynamic repair strategy table. 5.The 5G network-based remote dental prosthetic diagnosis and treatment system of claim 1, wherein, The trajectory prediction module includes: The depth modeling sub-module calls a depth value sequence in the dynamic repair strategy table, inputs the depth value sequence into an LSTM time series model, calculates a time weight through a forgetting gate and an input gate, divides the depth value into sliding windows, outputs a hidden layer state vector, and generates a depth change rate; The trajectory generation sub-module performs adjacent time point difference calculation on the defect area based on the depth change rate, fits an area growth curve through linear regression, extracts a slope parameter as a growth rate, performs matrix splicing on the change rate and the growth rate, and generates an evolution trajectory prediction vector; The policy updating submodule performs matching operation of vector and preset policy rule according to dimension feature of the evolution trajectory prediction vector, screens candidate policy meeting error threshold, marks optimal policy as priority, and outputs updated policy instruction.

6. A remote dental prosthetic diagnosis and treatment method based on a 5G network, characterized in that, The method is used for implementing the 5G network-based remote dental prosthetic diagnosis and treatment system according to any one of claims 1-5, and comprises the following steps: S1: receiving real-time transmitted oral three-dimensional scanning data stream through a 5G communication protocol, calling a Gaussian filtering algorithm to filter noise of the data stream, extracting a lesion area coordinate set, and generating three-dimensional reconstruction data; S2: inputting the three-dimensional reconstruction data into an edge detection algorithm to locate a contour boundary, calculating an edge sharpness value by calculating a boundary pixel gradient amplitude, and generating a lesion feature parameter by combining a difference in curvature of adjacent pixels; S3: inputting the lesion feature parameter and the edge sharpness value into a convolutional neural network model for classification training, outputting a defect area, a depth value, and a structure damage index, performing product operation on the depth value and the structure damage index to generate a repair precision coefficient, matching the repair precision coefficient with a preset repair technology mapping table to generate a dynamic repair strategy table; S4: calling an LSTM time series model to predict a depth value change rate based on the dynamic repair strategy table and the depth value, generating an evolution trajectory prediction vector by combining a difference in adjacent time stamps of the defect area, and outputting an updated policy instruction containing a correction parameter; S5: traversing a repair material database according to the updated policy instruction, matching a candidate material set meeting compression strength and adhesion coefficient, calculating a cosine similarity of the candidate material and the evolution trajectory prediction vector by combining real-time collected occlusion force monitoring data, and generating a repair scheme.

Images