A method, apparatus, system and storage medium for fixed denture abutment evaluation

Abstract

This invention relates to the field of denture interproximal contact technology, and discloses a method, device, system, and storage medium for evaluating the interproximal contact of fixed dentures. The evaluation method uses a MEMS piezoresistive sensor to collect dynamic data of interproximal forces in real time, and simultaneously employs laser scanning to acquire the three-dimensional morphology of the contact surface. The mechanical data is input into a regression model to analyze tightness and stability. The three-dimensional data is processed by a 3D convolutional neural network to extract geometric features. Finally, the results of multimodal analysis are fused to generate an evaluation report containing multi-dimensional indicators of contact quality. The device includes a data acquisition module, an intelligent analysis module, and a data output module. By employing multimodal data fusion and intelligent algorithms, simultaneous analysis and dynamic monitoring of the morphology and mechanics of the fixed denture contact surface are achieved. Quantitative evaluation indicators are generated through an automated process, overcoming the limitations of subjective judgment and static detection in traditional methods, and significantly improving the accuracy and clinical efficiency of interproximal contact quality evaluation.

Description

Technical Field

[0001] This invention relates to the field of prosthesis proximal contact technology, specifically to a method, apparatus, system, and storage medium for evaluating proximal contact of fixed dentures. Background Technology

[0002] Fixed prosthesis is a core treatment method in the field of prosthodontics, and the contact quality between its proximal surfaces and natural teeth directly affects the long-term stability of the prosthesis and the patient's oral health. Traditional clinical assessment mainly relies on visual examination by dentists, occlusal paper staining, and manual probe palpation, supplemented by two-dimensional imaging examinations. While these methods can preliminarily determine the tightness of the contact and morphological fit, they are limited by subjective experience and low-dimensional data acquisition capabilities, making it difficult to accurately quantify the mechanical state and geometric characteristics of the contact surfaces. With the popularization of digital dental technology, equipment such as 3D scanning and pressure sensing are gradually being applied in clinical practice, but they mostly remain at the level of single-point data acquisition or static analysis, and a systematic dynamic assessment system has not yet been formed.

[0003] Existing technologies based on single-modal detection methods have significant limitations: contact surface morphology detection devices (such as intraoral scanners) can acquire high-precision three-dimensional models, but cannot simultaneously reflect the mechanical distribution during the actual occlusal process; pressure sensing devices can measure the magnitude of adjacent forces, but lack correlation analysis with the geometric features of the contact surface. Furthermore, traditional dynamic monitoring often employs low-frequency sampling techniques, making it difficult to capture crucial details such as stress release during the initial placement of the prosthesis and instantaneous overload during chewing movements. These technological gaps often lead to contradictory clinical phenomena such as "morphologically acceptable but patient discomfort" or "mechanically satisfactory but rapid debonding," exposing inherent deficiencies in the current assessment system regarding temporal continuity and multi-parameter collaborative analysis.

[0004] The aforementioned problems severely restrict the standardization process and long-term success rate of fixed denture restoration. To address the shortcomings of existing technologies, this invention proposes a method, apparatus, system, and storage medium for assessing the proximal surfaces of fixed dentures. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method, device, system, and storage medium for assessing the proximal contact of fixed dentures, which solves the problem of inaccurate clinical judgment caused by the fragmentation of assessment of contact surface morphology and mechanical state in fixed denture restoration.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] The first aspect of this invention provides a method for evaluating the proximal contact of fixed dentures, comprising the following steps:

[0008] S1. Real-time contact force data between the fixed denture and adjacent teeth is collected by MEMS piezoresistive array sensor, and three-dimensional data of the contact surface is obtained by 635nm laser scanner.

[0009] S2. Input the adjacency force data into the regression analysis model, and simultaneously input the three-dimensional data of the contact surface into a 3D convolutional neural network to extract features and generate quantitative evaluation results, wherein the quantitative evaluation results include: tightness evaluation, position evaluation, shape and area evaluation.

[0010] S3. Generate an evaluation report from the quantitative evaluation results and upload it to the cloud platform for storage via an encrypted transmission protocol.

[0011] Preferably, step S1 includes:

[0012] S1.1 The smart sensor dental floss with built-in ±0.01N sensitivity pressure sensor monitors the adjacency force in real time, measures the pressure change in the contact area, and generates a pressure distribution map and force change curve.

[0013] S1.2 The miniature end-effector scanning device acquires three-dimensional data of the contact surface through laser scanning or optical imaging with a resolution of 0.05mm, and calculates the position, shape and area of ​​the contact surface.

[0014] Preferably, step S2 includes:

[0015] S2.1 Train a convolutional neural network model to convert the three-dimensional data of the contact surface into a 64×64×32 voxel matrix, and extract geometric features through a channel attention mechanism. The geometric features include contact surface accuracy, shape alignment and surface curvature distribution.

[0016] S2.2 Based on the geometric features and adjacency force time series data, establish an adjacency force dynamic regression model, and use an optimization algorithm to adjust the regression coefficient β to minimize the prediction error;

[0017] S2.3. The geometric features are weighted and fused with the tightness index and stability coefficient output by the regression model to generate a comprehensive score, and it is determined whether the contact surface position meets the preset conditions.

[0018] Preferably, the optimization algorithm in step S2.2 is the Levenberg-Marquardt algorithm, and the error in step S2.2 is defined by the following formula:

[0019]

[0020] Where β is the regression coefficient vector. F is the model's predicted value. i Let N be the measured adjacent force value at the i-th time point, and N be the total number of data sampling points.

[0021] Preferably, the preset condition in step S2.3 is:

[0022] The contact point location satisfies:

[0023] Where x and y are the coordinate offsets of the contact point on the cross-section of the tooth, z is the position along the long axis of the tooth, and H is the clinical crown height;

[0024] The contact area should be 1.5-2.0 mm in the anterior tooth region. 2 2.0-3.0mm in the posterior region 2 The adjacent force range is 1.0-2.5N.

[0025] A second aspect of the present invention provides a fixed denture proximal contact assessment device, applied to the aforementioned fixed denture proximal contact assessment method, comprising:

[0026] The data acquisition module is equipped with a MEMS piezoresistive array sensor and a 635nm laser scanner for real-time acquisition of adjacent force data and three-dimensional data of the contact surface.

[0027] The intelligent analysis module, whose input end is connected to the output end of the data acquisition module, includes a 3D-CNN processor and a Levberg-Marquardt optimizer, and is used to perform feature extraction and mechanical analysis on the adjacency force data and the three-dimensional contact surface data to generate quantitative evaluation results.

[0028] The data output module, with its input end connected to the output end of the intelligent analysis module, is used to generate an evaluation report from the quantitative evaluation results and transmit it to the cloud platform for storage.

[0029] Preferably, the intelligent analysis module includes:

[0030] Model training unit: Constructs a convolutional neural network architecture based on voxelized 3D data to extract geometric features of the contact surface;

[0031] Feature extraction unit: Identifies irregular areas and alignment deviations on the contact surface through multi-scale curvature analysis;

[0032] Mechanical analysis unit: Optimizes dynamic adjacency force model parameters and evaluates contact force tightness and stability;

[0033] Integrated decision-making unit: Determines the compliance of adjacent quality based on preset location thresholds and area standards.

[0034] Preferably, the data acquisition module includes:

[0035] The pressure sensing unit monitors the dynamic pressure distribution between adjacent teeth in real time, and achieves ±0.01N sensitivity detection through a MEMS piezoresistive array sensor.

[0036] The three-dimensional scanning unit reconstructs the three-dimensional morphology of the contact surface with high precision. It uses a λ=635nm laser to scan at a 45° incident angle to generate point cloud data.

[0037] A third aspect of the present invention provides a fixed denture proximal interpro ...

[0038] A fourth aspect of the present invention provides a storage medium storing a computer program that, when running, executes the fixed denture proximal interproximal assessment method as described above.

[0039] This invention provides a method, apparatus, system, and storage medium for evaluating the proximal interproximal contact of fixed dentures. It offers the following advantages:

[0040] 1. This invention integrates high-precision sensing, 3D scanning and deep learning algorithms to simultaneously analyze the morphological features and mechanical state of the contact surface, solving the problem of contradictory conclusions caused by the separation of morphological detection and mechanical evaluation in traditional methods. Existing technologies can only measure geometric parameters or adjacent force magnitudes separately and cannot correlate the two. This solution achieves multi-dimensional joint diagnosis of contact quality for the first time through cross-modal data fusion technology.

[0041] 2. Based on feature extraction and quantitative calculation of a convolutional neural network and regression model, this invention automatically generates objective scores for indicators such as contact surface accuracy and tightness. Compared with manual measurement that relies on doctors' experience, this technical solution avoids human visual error and differences in operation techniques, making the evaluation results of different medical institutions and different operators repeatable and comparable.

[0042] 3. This invention employs high-frequency data acquisition and time series analysis algorithms to capture the trend of adjacent force changes during the insertion and occlusion of fixed dentures in real time. Traditional static detection methods can only obtain instantaneous state data, while this solution effectively identifies dynamic problems such as abnormal stress release and occlusal force fluctuations through continuous monitoring technology, providing complete process data support for clinical adjustments.

[0043] 4. This invention uses an end-to-end automated processing architecture to integrate 3D scanning, mechanical testing, and data analysis into a single workflow, directly outputting a visual evaluation report. Compared to the complex process in existing technologies that requires step-by-step operation of multiple devices and manual data integration, this solution significantly reduces the operational threshold and improves diagnostic and treatment efficiency, making it particularly suitable for standardized applications in primary healthcare institutions. Attached Figure Description

[0044] Figure 1 This is a flowchart of the evaluation method of the present invention;

[0045] Figure 2 This is a flowchart of the device of the present invention;

[0046] Figure 3 This is a schematic diagram of the system structure of the present invention.

[0047] Among them, 40 is a fixed denture proximal assessment system; 41 is a processor; 42 is a memory; and 43 is a storage medium. Detailed Implementation

[0048] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0049] Please see the appendix Figure 1 This invention provides a method for evaluating the proximal contact of fixed dentures, comprising the following steps:

[0050] S1. Real-time contact force data between the fixed denture and adjacent teeth is collected by MEMS piezoresistive array sensor, and three-dimensional data of the contact surface is obtained by 635nm laser scanner.

[0051] Step S1.1 involves real-time monitoring of the adjacency force using a smart sensor dental floss with a built-in ±0.01N sensitivity pressure sensor, and includes the following implementation process:

[0052] Physical Structure: The smart sensor dental floss comprises a pressure sensing unit, a signal conditioning unit, and a wireless transmission unit. The pressure sensing unit consists of a 5×5 array of MEMS piezoresistive sensors, with adjacent sensors spaced 0.5 mm apart, and is connected to the operational amplifier in the signal conditioning unit via a flexible circuit. The wireless transmission unit integrates a Bluetooth 5.0 module for communication with the main control board.

[0053] Signal conversion and processing:

[0054] Mechanical signal conversion: The sensor converts the adjacent force F(t) into a voltage signal.

[0055] V(t) = 200·F(t) + ε(t) (unit: mV);

[0056] in It is Gaussian white noise.

[0057] Noise Reduction: The signal conditioning department uses a Butterworth low-pass filter to eliminate high-frequency noise. Its transfer function is:

[0058] (s is a complex frequency variable);

[0059] The filtered signal is discretized at a sampling rate of 100Hz to generate a digital sequence {F}. i |i=1,2,...,N}.

[0060] Pressure distribution map generation: The contact area is divided into 0.1mm × 0.1mm grids, and the pressure at each grid point is calculated using bilinear interpolation.

[0061]

[0062] Among them, F k w represents the measurements from four adjacent sensors. k This is the distance weighting coefficient.

[0063] A pressure distribution map can accurately show the force distribution in the contact area between the denture and adjacent teeth, reflecting the magnitude and uniformity of the contact force. This is crucial for assessing the tightness and uniformity of the contact area. By analyzing the pressure distribution map, dentists can determine whether the fixed denture has good contact with adjacent teeth and whether there are areas of excessive or insufficient pressure.

[0064] Construction of force change curves: For time series data {F i Perform cubic spline interpolation to generate a continuous function F(t) and its first derivative:

[0065] (Unit: N / s);

[0066] in, F(t+0.01) is the instantaneous rate of change of adjacency force with time (i.e., the first derivative); F(t+0.01) is the measured value of adjacency force 0.01 seconds after time t; F(t-0.01) is the measured value of adjacency force 0.01 seconds before time t; 0.01 is the time step (i.e., the time interval), corresponding to the sampling period of a 100Hz sampling rate; 0.02 is the total span of the time window (i.e., the total duration of the 0.01 seconds before and after).

[0067] Force variation curves record the trend of contact force over time, reflecting the dynamic change process of interproximal forces. This is crucial for assessing the stability of the contact area and the fixation of fixed prostheses. Force variation curves help dentists understand the performance of the interproximal area under different forces, and whether there is uneven contact or other potential problems.

[0068] Step S1.2 Acquires three-dimensional data through 0.05mm resolution laser scanning, including the following implementation process:

[0069] Physical structure: The miniature end-effector scanner includes a laser emitter, an optical receiver, and a motion control unit. The laser emitter uses a 635nm semiconductor laser with a pulse frequency of 1MHz; the optical receiver integrates a CMOS sensor with a pixel size of 1.4μm; and the motion control unit drives the scanning head to rotate (0-180°) via a stepper motor.

[0070] Laser scanning technology generates detailed 3D point cloud data on the contact surface, providing its geometry and precise location. The high penetration of lasers ensures high-quality data acquisition even in complex oral environments, allowing for flexible operation of miniature end-effector scanning devices, even in confined spaces. Optical imaging technology captures images of the contact surface using a high-definition probe and performs pixel-level image processing to obtain detailed information such as the contact surface's position, shape, and area. This image data is then transformed into quantifiable 3D data through image processing algorithms, ensuring high resolution and accuracy of the contact surface data.

[0071] 3D point cloud generation: Calculation of laser ranging values ​​using the time-of-flight method.

[0072]

[0073] Where Δt is the time difference between the transmitted and received pulses, and the scanning accuracy is ±0.02mm.

[0074] Coordinate Transformation and Reconstruction: The point cloud (x′, y′, z′) in the local coordinate system is transformed into the global coordinate system using a rotation matrix R and a translation vector T.

[0075]

[0076] Where R is a 3×3 rotation matrix, T=[t x ,t y ,t z ] T It is a translation vector, pre-calibrated using a calibration plate.

[0077] Contact surface feature calculation:

[0078] Contact area: Surface area calculated based on Delaunay triangulation algorithm:

[0079]

[0080] Where A is the total area of ​​the contact surface (unit: mm) 2 M represents the total number of triangular patches generated by the Delaunay triangulation algorithm; v1, v2, and v3 are the coordinates of the three vertices of a single triangular patch (unit: mm), which belong to points in the 3D point cloud data; (v2-v1)×(v3-v1) is the vector cross product operation, and the result is a vector. This is the normalization coefficient.

[0081] The standard width for the anterior teeth region is 1.5-2.0 mm. 2 The posterior region is 2.0-3.0 mm. 2 .

[0082] Positional deviation: Center point of contact surface (x) c ,y c ,z c The Euclidean distance between the theoretical position (x0, y0, z0) and the theoretical position (x0, y0, z0) must satisfy:

[0083]

[0084] Where H is the clinical crown height, and the contact point must be located at the proximal 1 / 3 to the middle 1 / 3 of the gingiva (0.3H≤z). c ≤0.5H).

[0085] S2. Input the adjacency force data into the regression analysis model, and simultaneously input the three-dimensional data of the contact surface into a 3D convolutional neural network to extract features and generate quantitative evaluation results, wherein the quantitative evaluation results include: tightness evaluation, position evaluation, shape and area evaluation.

[0086] Data preprocessing and model training:

[0087] 3D data voxelization:

[0088] Three-dimensional point cloud data of the contact surfaces between the fixed prosthesis and adjacent teeth were acquired using a miniature end-effector scanning device. The point cloud was then converted into a 64×64×32 voxel matrix using a voxelization algorithm. The voxel resolution was 0.05mm×0.05mm×0.05mm, and each voxel value was defined as follows:

[0089]

[0090] This step converts the raw scan data into a structured 3D matrix, providing a standardized format for the input of the subsequent convolutional neural network (CNN). The voxel resolution (0.05 mm) is set based on clinical needs to ensure that micrometer-level geometric deviations at the contact surface can be captured.

[0091] Data arrangement optimization:

[0092] Genetic algorithms are used to optimize data permutations, reducing the similarity between data with different labels and enhancing the correlation between data of the same type. The objective function is:

[0093]

[0094] Among them, S inter S represents the cosine similarity of different label matrices. intraThe similarity is defined as the similarity between identical label matrices, where N is the total number of samples in the dataset, and M is the number of samples in the dataset. i Let be the matrix representation of the i-th sample. After optimization, the discrimination between different categories of data is improved by 15%-20%.

[0095] Optimizing the data arrangement enhances the internal consistency of similar data, thereby improving CNN training efficiency. This step is directly related to the accuracy of the CNN in subsequent steps, as the optimized data distribution is more conducive to feature extraction.

[0096] Logistic regression model construction:

[0097] Based on the stress distribution map in the historical dataset of fixed denture adjacency relationships, target influencing factors were screened:

[0098] Factor selection criteria: Frequency of occurrence of stress regions f k ≥20% and overlap with clinical assessment results r k ≥85%.

[0099] Model Construction: Establish the association between adjacency strength and clinical pass rate using a logistic regression model.

[0100] (Regularization coefficient λ = 0.01);

[0101] Where p is the probability of a successful connection between adjacent prostheses. β is the logarithmic odds, β0 is the intercept term of the regression model, and β k For the k-th independent variable F k The regression coefficients, ∈, are the random error terms.

[0102] This step filters out target influencing factors (such as the value F for a specific pressure region). k This will be used as an input variable for the dynamic regression model in the following steps to ensure the clinical relevance of the mechanical analysis.

[0103] Contact surface feature extraction:

[0104] 3D Convolutional Neural Network (CNN) Architecture:

[0105] The network employs a 3DResNet-18 structure with four convolutional layers. The input is a 64×64×32 voxel matrix, and the output is a 128-dimensional geometric feature vector G = (G1, G2, G3), specifically including:

[0106] Contact surface accuracy G1: Calculates the average absolute error (in mm) between the contact surface position and the reference position.

[0107]

[0108] Where, xi Let x be the actual three-dimensional coordinates of the i-th sampling point. ref The theoretical reference position for the contact surface, |x i -x ref | represents the Euclidean distance deviation between the i-th sampling point and the reference position.

[0109] Alignment G2: Calculates the angular deviation of the contact surface (in °) using the angle between the normal vectors.

[0110]

[0111] Where n1 is the actual normal vector of the contact surface of the fixed denture, n2 is the theoretical normal vector of the contact surface of the adjacent tooth, n1·n2 is the dot product of the two normal vectors, and ||n1|| and ||n2|| are the magnitudes (norms) of the normal vectors.

[0112] Irregularity G3: Statistical percentage of non-planar triangular facets (%):

[0113]

[0114] CNNs extract geometric features directly through voxel matrices, and their network structures (such as ResNet-18) are designed to address the vanishing gradient problem, ensuring the effective transfer of deep features.

[0115] Channel attention module:

[0116] In a CNN, a channel attention mechanism is embedded, and the weights are calculated as follows:

[0117]

[0118] Where σ is the Sigmoid function, and H = 64, W = 64, and D = 32 are the feature map dimensions.

[0119] The channel attention module dynamically adjusts channel weights based on the spatial distribution of the feature map, such as enhancing the response of the contact surface edge region, thereby improving the sensitivity of position assessment.

[0120] Adjacency force dynamic regression modeling:

[0121] Dynamic regression model construction:

[0122] Adjacency force prediction model based on cubic B-spline basis function expansion:

[0123]

[0124] Among them, B k (t) is a cubic B-spline basis function (number of nodes K = 15), β k is the regression coefficient.

[0125] The choice of cubic B-spline basis functions (rather than polynomial or Fourier basis functions) is more suitable for describing the dynamic changes of adjacency forces due to their local support and smoothness, thus avoiding overfitting.

[0126] Levenberg-Marquardt optimization algorithm:

[0127] Optimize the regression coefficients using the error function defined in the claims:

[0128]

[0129] Iterative update formula:

[0130] β (n+1) =β (n) -(J T J+λI) -1 J T r;

[0131] Where J is the Jacobian matrix, and the elements J ik =B k (t i λ is the damping factor (initial value 0.001, dynamically adjusted); r is the residual vector. β is the regression coefficient vector, and its dimension is determined by the number of nodes k in the cubic B-spline basis function, i.e., β = (β1, β2, ..., β2). K );F i β is the measured adjacency force value at the i-th time point; (n) β (n+1) J is the regression coefficient vector for the nth and (n+1)th iterations; T I is the transpose of the Jacobian matrix, with dimensions K×N; I is the identity matrix, with dimensions K×K.

[0132] The Levenberg-Marquardt algorithm combines the advantages of gradient descent and Gauss-Newton's method, and converges quickly to the optimal solution based on the selected factors, ensuring the reliability of the tightness index S.

[0133] Output index calculation:

[0134] Tightness Index S: The average value of predicted adjacent forces (in N / mm) 2 ):

[0135]

[0136] Stability coefficient C: Standard deviation of the rate of change of adjacent force (in N / s):

[0137]

[0138] in, For time point t i The rate of change of adjacency force at a given location; μ is the average rate of change of adjacency force. This represents the deviation of the rate of change from the average value at a single point in time.

[0139] The calculation of S and C directly depends on the optimized regression coefficient β. * Its physical meaning is the mean and fluctuation of the contact force, which provides a mechanical quantitative indicator for the comprehensive score.

[0140] Comprehensive assessment and judgment:

[0141] Weighted fusion and score generation:

[0142] Geometric features and mechanical properties are combined according to the weights of the claims to generate a comprehensive score:

[0143] Q=0.4G1+0.3S+0.2C+0.1(1-G3);

[0144] The weighting (40% for geometric accuracy and 50% for mechanical properties) is based on clinical research findings to ensure that the score reflects both morphological and functional fit.

[0145] Preset standard judgment:

[0146] Adjacent force range:

[0147] Anterior tooth region: 1.0N≤F avg ≤2.0N;

[0148] Posterior region: 1.5N≤F avg ≤2.5N;

[0149] Among them, F avg This represents the average value of the adjacency force.

[0150] Contact location:

[0151]

[0152] Where H is the clinical crown height (8mm for anterior teeth and 10mm for posterior teeth), x and y are the coordinate offsets of the contact point on the cross-section of the tooth, and z is the position along the long axis of the tooth.

[0153] Contact area:

[0154] Anterior teeth region: 1.5-2.0mm 2 ;

[0155] Posterior region: 2.0-3.0mm 2 .

[0156] The criteria were derived from clinical studies published in the Journal of Dental Research, forming a closed-loop validation with geometric features (G1, G2, G3) and mechanical parameters (S, C).

[0157] S3. Generate an evaluation report from the quantitative evaluation results and upload it to the cloud platform for storage via an encrypted transmission protocol.

[0158] Assessment report generation and structuring:

[0159] First, the quantitative evaluation results (tightness index S, positional deviation Δ, contact area A, and comprehensive score Q) output from step S2 are integrated according to a preset template. The template is in JSON format and includes the following fields:

[0160] Tightness assessment: Numerical field "tightness":{"value":1.8,"unit":"N / mm"} 2 "};

[0161] Position assessment: Coordinate field "position":{"x":0.1,"y":-0.05,"z":3.0,"H":8};

[0162] Shape and area assessment: "area":{"value":1.9,"unit":"mm 2 "}

[0163] For example, data serialization can be achieved using Python's json library to generate standardized report files (with the extension .repjson).

[0164] Encrypted transmission protocol implementation:

[0165] The report file is encrypted using the AES-256 encryption algorithm, with the key transmitted via RSA-2048 asymmetric encryption. The encryption process includes:

[0166] Key generation: Randomly generate a 32-byte symmetric key KsymKsym.

[0167] Data encryption: The encryption function is:

[0168] C = AES.encrypt(M,K) sym ,mode=GCM);

[0169] Where M represents the plaintext report and C represents the ciphertext, GCM mode provides integrity verification.

[0170] Key Encapsulation: Using the cloud platform's public key PK cloud Encryption K sym :

[0171] E key =RSA.encrypt(K sym PK cloud );

[0172] Secure transmission and cloud storage:

[0173] Upload encrypted data packets {C,E} via HTTPS protocol (TLS 1.3) key The transmission process satisfies:

[0174] Data fragmentation: When the file size exceeds 1MB, it is transmitted in 512KB fragments, with the fragment index field "chunk_index": 2 / 5.

[0175] Retransmission mechanism: If the ACK is not returned within 500ms, the interrupted transmission will be resumed.

[0176] The cloud platform storage adopts a distributed architecture (such as HDFS), with a data redundancy strategy of 3 replicas. The storage path is generated based on the patient ID and timestamp hash.

[0177] Path=" / data / "+SHA256(PID∥timestamp)[0:8].

[0178] The fixed denture proximal assessment device described below and the fixed denture proximal assessment method described above can be used interchangeably.

[0179] Please see the appendix Figure 2 A fixed denture proximal assessment device, applied to the aforementioned fixed denture proximal assessment method, includes:

[0180] The data acquisition module is equipped with a MEMS piezoresistive array sensor and a 635nm laser scanner for real-time acquisition of adjacent force data and three-dimensional data of the contact surface.

[0181] The intelligent analysis module, whose input end is connected to the output end of the data acquisition module, includes a 3D-CNN processor and a Levberg-Marquardt optimizer. It is used to perform feature extraction and mechanical analysis on the adjacency force data and the three-dimensional data of the contact surface to generate quantitative evaluation results.

[0182] The data output module, whose input end connects to the output end of the intelligent analysis module, is used to generate an evaluation report from the quantitative evaluation results and transmit it to the cloud platform for storage.

[0183] The intelligent analysis module includes:

[0184] Model training unit: Constructs a convolutional neural network architecture based on voxelized 3D data to extract geometric features of the contact surface;

[0185] Feature extraction unit: Identifies irregular areas and alignment deviations on the contact surface through multi-scale curvature analysis;

[0186] Mechanical analysis unit: Optimizes dynamic adjacency force model parameters and evaluates contact force tightness and stability;

[0187] Integrated decision-making unit: Determines the compliance of adjacent quality based on preset location thresholds and area standards.

[0188] The data acquisition module includes:

[0189] The pressure sensing unit monitors the dynamic pressure distribution between adjacent teeth in real time, and achieves ±0.01N sensitivity detection through a MEMS piezoresistive array sensor.

[0190] The three-dimensional scanning unit reconstructs the three-dimensional morphology of the contact surface with high precision. It uses a λ=635nm laser to scan at a 45° incident angle to generate point cloud data.

[0191] The device in this embodiment can be used to execute the above method embodiments, and its principle and technical effects are similar, so they will not be described again here.

[0192] Please see the appendix Figure 3 The present invention also provides a fixed denture proximal interproximal assessment system 40, comprising: a memory 41 and a processor 42, wherein the memory 42 stores a computer program executed by the processor 41, and the computer program executes the above-described method when executed by the processor 41.

[0193] The present invention also provides a storage medium 43 on which a computer program is stored, and the computer program is executed by a processor 41 to perform the method described above.

[0194] The storage medium 43 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0195] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for evaluating the proximal contact of fixed dentures, characterized in that, Includes the following steps: S1. Real-time contact force data between the fixed denture and adjacent teeth is collected by MEMS piezoresistive array sensor, and three-dimensional data of the contact surface is obtained by 635nm laser scanner. S2. Input the adjacency force data into the regression analysis model, and simultaneously input the three-dimensional data of the contact surface into a 3D convolutional neural network to extract features and generate quantitative evaluation results, wherein the quantitative evaluation results include: tightness evaluation, position evaluation, shape and area evaluation. S3. Generate an evaluation report from the quantitative evaluation results and upload it to the cloud platform for storage via an encrypted transmission protocol; Step S1 includes: S1.1 The smart sensor dental floss with built-in ±0.01N sensitivity pressure sensor monitors the adjacency force in real time, measures the pressure change in the contact area, and generates a pressure distribution map and force change curve. S1.2 The miniature end-effector scanning device acquires three-dimensional data of the contact surface through laser scanning or optical image capture with a resolution of 0.05mm, and calculates the position, shape and area of ​​the contact surface; Step S2 includes: S2.1 Train a convolutional neural network model to convert the three-dimensional data of the contact surface into a 64×64×32 voxel matrix, and extract geometric features through a channel attention mechanism. The geometric features include contact surface accuracy, shape alignment, and surface curvature distribution. S2.

2. Based on the aforementioned geometric features and adjacency force time-series data, a dynamic regression model for adjacency force is established, and an optimization algorithm is used to adjust the regression coefficients. To minimize prediction error; S2.

3. The geometric features are weighted and fused with the tightness index and stability coefficient output by the regression model to generate a comprehensive score, and it is determined whether the contact surface position meets the preset conditions. The optimization algorithm in step S2.2 is the Levenberg-Marquardt algorithm, and the error in step S2.2 is defined by the following formula: ； in, For the regression coefficient vector, These are the model's predicted values. For the first Measured adjacent force values ​​at each time point This represents the total number of data sampling points; The preset conditions in step S2.3 are as follows: The contact point location satisfies: ; in, , This represents the coordinate offset of the contact point on the cross-section of the tooth. The position along the long axis of the tooth. Clinical coronary artery height; The contact area should be 1.5-2.0 mm in the anterior tooth region. 2 2.0-3.0mm in the posterior region 2 The adjacent force range is 1.0-2.5N.

2. A fixed denture proximal assessment device for implementing the fixed denture proximal assessment method of claim 1, comprising: The data acquisition module is equipped with a MEMS piezoresistive array sensor and a 635nm laser scanner for real-time acquisition of adjacent force data and three-dimensional data of the contact surface. The intelligent analysis module, whose input end is connected to the output end of the data acquisition module, includes a 3D-CNN processor and a Levberg-Marquardt optimizer, and is used to perform feature extraction and mechanical analysis on the adjacency force data and the three-dimensional contact surface data to generate quantitative evaluation results. The data output module, with its input end connected to the output end of the intelligent analysis module, is used to generate an evaluation report from the quantitative evaluation results and transmit it to the cloud platform for storage.

3. A fixed denture proximal contact assessment system, characterized in that, include: A memory and a processor, wherein the memory stores a computer program executed by the processor, the computer program performing the fixed denture proximal interproximal assessment method as described in claim 1 when executed by the processor.

4. A storage medium, characterized in that, The storage medium stores a computer program that, when running, executes the fixed denture proximal interproximal assessment method as described in claim 1.

Images