A three-dimensional scanning-based virtual try-on system for dental prostheses

Abstract

A kind of virtual try-on system of dental prosthesis based on three-dimensional scanning, it relates to the field of dental prosthetics;Extract a plurality of facial points and its displacement trajectory information, output the classification result representing the type of facial point, obtain the stability score of facial point in different preset time window, extract the stable interval of jaw position and obtain the jaw position information, and generate the try-on model in combination with the maxillary model, the mandibular model and the prosthesis model, obtain the contact change rate of the try-on model and judge whether it is less than the change threshold, if not, adjust the prosthesis model until it is less than the change threshold;It has the advantages of effectively reducing the interference of soft tissue relaxation on jaw position judgment, improving the accuracy of jaw position information extraction and improving the reliability of prosthesis design.

Description

Technical Field

[0001] This application relates to the field of oral prosthetics technology, specifically a virtual fitting system for oral prostheses based on three-dimensional scanning. Background Technology

[0002] In dental restoration applications, accurate positioning of the physiological jaw position is the core link to ensure the functional fit and aesthetic effect of the restoration. Its accuracy will directly affect the occlusal coordination, chewing efficiency and facial appearance harmony of the restoration. However, the existing technology has significant limitations when dealing with patients with relatively loose facial tissues.

[0003] Elderly patients or individuals with decreased soft tissue elasticity exhibit high variability in their facial skin, muscles, and subcutaneous tissues during dynamic movements. This makes it difficult for traditional static three-dimensional scanning or single occlusion recording to capture the true and stable jaw position. Doctors need to repeatedly try on the prosthesis intraorally and manually adjust it, which not only prolongs the treatment cycle and increases patient discomfort, but also leads to a decrease in strength and material consumption due to multiple corrections.

[0004] Existing methods lack an intelligent recognition mechanism for stable time windows during dynamic facial movements, which cannot effectively reduce the interference of soft tissue laxity on jaw position judgment, severely restricting the reliability and personalization of prosthesis design. To address these issues, existing technologies urgently need improvement. Summary of the Invention

[0005] The purpose of this application is to provide a virtual fitting system for dental prostheses based on three-dimensional scanning, which has the advantages of effectively reducing the interference of soft tissue laxity on jaw position judgment, improving the accuracy of jaw position information extraction, and thus improving the reliability of prosthesis design.

[0006] The objective of this application can be achieved through the following technical solution: Firstly, a virtual fitting system for dental prostheses based on three-dimensional scanning, comprising the following modules:

[0007] The data acquisition module is used to synchronously acquire facial point cloud sequence data, mandibular positioning data, and electromyographic signal data within a preset detection period;

[0008] The data processing module is used to extract several facial points and their displacement trajectory information based on the facial point cloud sequence data and the mandibular positioning data.

[0009] The facial point classification module is used to input the displacement trajectory information of each facial point into a preset feature segmentation model to output a classification result that represents the corresponding facial point type;

[0010] The facial point assessment module is used to obtain the stability score of each facial point in different preset time windows based on the classification results, displacement trajectory information, and electromyographic signal data, and to extract the jaw position stability interval in the time axis of the facial point cloud sequence data based on the stability score.

[0011] The model combination module is used to obtain jaw position information representing the spatial relative relationship between the maxilla and mandible based on the jaw position stability range, and to generate a trial model based on the jaw position information combined with the preset maxillary model, mandibular model and prosthesis model.

[0012] The model feedback module is used to perform motion simulation on the trial model to obtain the contact change rate characterizing its stability, and to determine whether the contact change rate is less than a preset change threshold. If yes, the trial model is output; otherwise, the prosthesis model is adjusted until the contact change rate is less than the preset change threshold.

[0013] Secondly, a method for virtual fitting of dental prostheses based on three-dimensional scanning includes the following steps:

[0014] Simultaneously acquire facial point cloud sequence data, jaw positioning data, and electromyographic signal data within a preset detection period;

[0015] Based on the facial point cloud sequence data and mandibular positioning data, extract several facial points and their displacement trajectory information;

[0016] The displacement trajectory information of each facial point is input into a preset feature segmentation model to output a classification result representing the corresponding facial point type;

[0017] Based on the classification results, displacement trajectory information, and electromyographic signal data, the stability score of each facial point is obtained within different preset time windows. Based on the stability score, the jaw position stability interval is extracted from the time axis of the facial point cloud sequence data.

[0018] Based on the jaw position stability range, jaw position information representing the spatial relative relationship between the maxilla and mandible is obtained. Based on the jaw position information, a trial fitting model is generated by combining the preset maxillary model, mandibular model, and prosthesis model.

[0019] The trial model is subjected to motion simulation to obtain the contact change rate characterizing its stability. It is determined whether the contact change rate is less than a preset change threshold. If so, the trial model is output. If not, the prosthesis model is adjusted until the contact change rate is less than the preset change threshold.

[0020] Thirdly, a computer storage medium storing computer-executable instructions, which, when executed, implement the virtual fitting system for dental prostheses based on three-dimensional scanning described in the first aspect.

[0021] Compared with the prior art, the beneficial effects of this application are:

[0022] This application acquires facial point cloud sequence data, mandibular positioning data, and electromyographic signal data within a preset detection period simultaneously, intelligently classifies facial point types, assesses stability scores, and extracts mandibular stability intervals, thereby accurately determining physiological mandibular position and effectively reducing the interference of soft tissue laxity. It has the advantages of improving the reliability and personalization of prosthesis design. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of a module of a virtual fitting system for dental prostheses based on three-dimensional scanning according to this application;

[0024] Figure 2 This is a schematic diagram illustrating the steps of a virtual fitting method for dental prostheses based on three-dimensional scanning, as described in this application. Detailed Implementation

[0025] The technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components described and shown in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but only to illustrate selected embodiments of this application.

[0026] Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this application. It should be noted that similar reference numerals and letters in the following figures indicate similar items. Therefore, once an item has been defined in one figure, it does not need to be further defined and explained in subsequent figures. The terms "first", "second", etc. are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance.

[0027] In traditional virtual fitting techniques for dental prostheses, when dealing with complex situations such as facial tissue laxity in elderly patients, the non-rigid deformation of facial soft tissues significantly interferes with jaw alignment assessment, leading to a lack of accuracy and stability in determining physiological jaw position. Furthermore, existing technologies fail to effectively distinguish between skeletal, muscular, and cutaneous movement trajectories, introducing soft tissue movement noise into the jaw position modeling process. This results in the functional fit and aesthetic effect of the prosthesis design deviating from physiological needs, thus increasing the frequency of clinical adjustments.

[0028] For example, during the treatment of an elderly patient with significantly loose facial skin, when the system simultaneously acquired dynamic three-dimensional facial sequence data and mandibular movement data, the displacement trajectories of key points in the facial point cloud sequence exhibited irregular fluctuations due to the influence of subcutaneous tissue slippage. Because electromyographic signal data was not included in the analysis framework, the movement patterns of bony and cutaneous feature points could not be effectively distinguished, leading to deviations in the identification of stable jaw positions. Consequently, the jaw position information used in the registration of the static maxillary and mandibular dentition models was inaccurate, and the spatial relative relationship between the maxilla and mandible in the generated virtual patient model could not accurately reflect the physiological state. After trial fitting of the prosthesis, unstable occlusal contact occurred, requiring repeated clinical adjustments to correct the morphology.

[0029] If these issues are not addressed, the functionality of the prosthesis will fail to meet the physiological requirements of chewing and speech, potentially causing persistent discomfort for patients. Furthermore, the continuity of the treatment process will be disrupted, increasing the consumption of medical resources. In particular, persistent misalignment of the jawline directly increases the risk of prosthesis design failure, forcing a prolonged treatment period and impacting the overall quality of care.

[0030] Therefore, this application provides a virtual fitting system for dental prostheses based on three-dimensional scanning, such as... Figure 1 As shown, it includes the following modules:

[0031] The data acquisition module is used to synchronously acquire facial point cloud sequence data, mandibular positioning data, and electromyographic signal data within a preset detection period;

[0032] The data processing module is used to extract several facial points and their displacement trajectory information based on the facial point cloud sequence data and the mandibular positioning data.

[0033] The facial point classification module is used to input the displacement trajectory information of each facial point into a preset feature segmentation model to output a classification result that represents the corresponding facial point type;

[0034] The facial point assessment module is used to obtain the stability score of each facial point in different preset time windows based on the classification results, displacement trajectory information, and electromyographic signal data, and to extract the jaw position stability interval in the time axis of the facial point cloud sequence data based on the stability score.

[0035] The model combination module is used to obtain jaw position information representing the spatial relative relationship between the maxilla and mandible based on the jaw position stability range, and to generate a trial model based on the jaw position information combined with the preset maxillary model, mandibular model and prosthesis model.

[0036] The model feedback module is used to perform motion simulation on the trial model to obtain the contact change rate characterizing its stability, and to determine whether the contact change rate is less than a preset change threshold. If yes, the trial model is output; otherwise, the prosthesis model is adjusted until the contact change rate is less than the preset change threshold.

[0037] The system first includes a data acquisition module. This module is responsible for acquiring various physiological data from the patient. For example, a 3D scanner can continuously record the 3D morphological changes of the patient's face over a period of time, generating facial point cloud sequence data. Simultaneously, a mandibular motion tracker can record the movement trajectory of the patient's mandible, generating mandibular positioning data. Furthermore, surface electromyography (EMG) sensors can acquire electromyographic signal data of the patient's masticatory muscles. This data can be acquired independently using different devices and then time-stamped, or it can be acquired synchronously using an integrated multimodal acquisition device.

[0038] Next, the data processing module processes the collected data. Based on the facial point cloud sequence data and mandibular positioning data, this module extracts multiple facial points from the patient's face and calculates the displacement trajectory information of these facial points over time. For example, specific points on the face can be manually selected, and their three-dimensional coordinates can be tracked at each time point to form a displacement trajectory. Alternatively, a feature recognition-based algorithm can be used to automatically identify key points on the face and record their spatial positions at different time points, thereby generating a displacement trajectory. During processing, the facial point cloud data and mandibular positioning data can be rigidly registered to ensure that the facial point displacement trajectories are described in a unified coordinate system.

[0039] Subsequently, the facial point classification module classifies the extracted facial points. This module inputs the displacement trajectory information of each facial point into a pre-defined feature partitioning model to output a classification result representing the corresponding facial point type. For example, a rule-based classifier can be constructed to classify facial points into different categories based on simple statistical features of displacement trajectories. Alternatively, a simple support vector machine model can be used to learn the classification boundaries of different facial point types by training on historical data. The classification result can indicate whether facial points are mainly affected by skeletal movement, muscle activity, or skin elasticity.

[0040] Furthermore, the facial point assessment module evaluates the stability of the classified facial points. Based on the classification results, displacement trajectory information, and electromyographic (EMG) signal data, this module calculates a stability score for each facial point within different preset time windows. For example, the displacement variance of a facial point within a time window can be simply used as its stability score; the smaller the displacement variance, the higher the score. For EMG signal data, its average intensity can be used as an auxiliary indicator. Then, based on these stability scores, jaw position stability intervals are identified and extracted from the time axis of the facial point cloud sequence data. These intervals can be selected by setting a fixed stability score threshold and choosing all consecutive time periods with scores higher than that threshold as candidate intervals.

[0041] Building upon this, the model assembly module is responsible for constructing a virtual fitting model. This module obtains jaw position information representing the spatial relationship between the maxilla and mandible based on the jaw position stability interval. For example, any time point within the jaw position stability interval can be selected, and the relative position of the maxilla and mandible at that time point can be used as jaw position information. Then, the pre-set maxillary model, mandibular model, and prosthesis model are combined in a simple three-dimensional space based on this jaw position information, for example, through manual adjustment or rigid transformation based on a small number of corresponding points, to generate an initial fitting model.

[0042] Finally, the model feedback module evaluates and optimizes the generated trial model. This module performs motion simulation on the trial model to obtain the contact change rate, which characterizes its stability. For example, it can simulate simple opening and closing movements of the mandibular model and calculate the change in contact area between the prosthesis model and the maxillary and mandibular models before and after the movement. Then, it determines whether the contact change rate is less than a preset change threshold. If the contact change rate meets the requirements, the trial model is output as the final design scheme. If the contact change rate does not meet the requirements, the prosthesis model needs to be adjusted, for example, by manually modifying the shape of the prosthesis, and then the motion simulation and evaluation are repeated until the contact change rate is less than the preset change threshold.

[0043] It should be further explained that, in the specific implementation process, the process of simultaneously acquiring facial point cloud sequence data, jaw positioning data, and electromyographic signal data within the preset detection period includes:

[0044] The facial point cloud sequence data refers to an ordered set of data that reflects the three-dimensional shape of the patient's face and its changes over time, acquired at continuous time points. The data at each time point is usually represented as a facial point cloud or mesh model composed of a large number of three-dimensional coordinate points. The entire sequence is arranged at fixed time intervals (e.g., 1 / 45 seconds) to fully record the dynamic changes in facial morphology within a detection cycle (e.g., relaxation-clenching-relaxation).

[0045] The data is acquired using a non-contact 3D scanning device. The patient assumes a natural head position and remains still. The 3D scanning device performs continuous scanning when the patient executes a preset mandibular movement command. The data output is a continuous 3D point cloud or mesh sequence file containing timestamps.

[0046] For example, a structured light 3D scanner projects a specially coded grating pattern onto the patient's face, captures the pattern deformed by the curvature of the face through one or more cameras, and calculates the 3D coordinates of each point in real time using the principle of triangulation. This system has a high frame rate (usually 30-60 frames / second) and is suitable for dynamic capture.

[0047] The mandibular positioning data refers to the data that characterizes the position and orientation of the patient's mandible in three-dimensional space. Its core function is to provide reference information that reflects the movement trajectory of the mandible itself, independent of the movement of facial soft tissues.

[0048] The acquisition method is achieved through an additional tracking method synchronized with the above facial scan, using optical tracking markers: small reflective balls or special pattern markers are pasted on the surface of accessible and relatively stable anatomical landmarks of the mandible (such as the corresponding points on the skin of the lateral condyles on both sides);

[0049] By synchronously tracking the changes in the three-dimensional coordinates of these marker points in space using the aforementioned three-dimensional scanning equipment or a dedicated optical motion capture camera, the motion trajectory coordinates of the mandible can be directly obtained. These coordinates are strictly synchronized with the facial point cloud sequence data in time and ultimately unified into the same three-dimensional coordinate system.

[0050] The electromyographic signal data refers to the time-varying sequence data of bioelectrical signals collected from the skin surface of the masticatory muscle group through surface electrodes. It reflects the electrophysiological activities (such as activation period and intensity changes) of the corresponding muscles during chewing, clenching and other actions.

[0051] Using a multi-channel surface electromyography (sEMG) signal acquisition instrument, surface electrode pairs are attached to the skin surface of the muscle belly of the target masticatory muscle according to anatomical standards. When the patient performs the same jaw movement command as the facial scan, the raw electromyographic signals of each channel are acquired simultaneously. The sampling frequency is usually above 1000 Hz to capture signal details.

[0052] The raw electromyographic (EMG) signals need to undergo a series of preprocessing steps, including bandpass filtering, full-wave rectification, and smoothing. The processed EMG signal envelope data is then aligned with the time axis of the facial point cloud sequence data. This allows the activation period and intensity changes of the muscles to be correlated with the motion events of facial points (such as displacement initiation and peak value) for analysis. This is used to verify the rationality of muscular point movements and to help determine the timing of functional occlusal contact.

[0053] In some of the embodiments described above in this application, it is necessary to extract facial points and their displacement trajectory information. However, accurately defining these facial points and precisely acquiring their displacement trajectory information during the patient's mandibular movement is a key challenge to ensure the subsequent extraction of the jaw stability zone and the generation of the fitting model. If the facial points are not clearly defined or the displacement trajectory information is not accurately acquired, it may lead to deviations in jaw position information, thereby affecting the accuracy of the prosthesis fitting.

[0054] In response, this application further proposes that the facial points in the above system refer to points selected from different locations on the patient's face. The point cloud coordinates of a single facial point at continuous time points are obtained based on the facial point cloud sequence data. The coordinate system of the preset marker point on the patient's mandible at the same time point is obtained based on the mandibular positioning data. The ICP algorithm is used to align the point cloud coordinates of each facial point at the same time point to the coordinate system at the corresponding time point. The aligned point cloud coordinates of a single facial point at continuous time points are normalized to the displacement trajectory information of the corresponding facial point.

[0055] In the system of this application, to accurately acquire facial points and their displacement trajectory information, the data processing module first selects discrete facial points from different locations on the patient's face. These facial points are tracked in a continuous sequence of facial point cloud data to obtain their original three-dimensional coordinates at different time points. Simultaneously, the system synchronously acquires the coordinate system of a preset marker point on the patient's mandible at the same time point; this coordinate system represents the instantaneous spatial position and posture of the mandible. To eliminate the influence of the patient's overall head movement on the displacement trajectory and focus on facial deformation related to mandibular movement, the system uses the ICP algorithm to precisely align the point cloud coordinates of each facial point at the same time point to the corresponding mandibular coordinate system at that time point.

[0056] Through this alignment operation, the coordinates of facial points are transformed into a local coordinate system with the patient's mandible as a reference, thus accurately reflecting the actual movement of facial tissues relative to the mandible. Subsequently, the point cloud coordinates of a single facial point at consecutive time points after alignment are normalized to eliminate the influence of individual differences and measurement units, ultimately forming displacement trajectory information characterizing the displacement changes of that facial point during mandibular movement. This precise displacement trajectory information provides a reliable data foundation for the subsequent facial point assessment module to accurately extract the stable jaw position interval, thereby ensuring the accuracy of the generated trial model.

[0057] Through the above technical solutions, this system can accurately define facial points and align the displacement coordinates of these points to the mandibular coordinate system using the ICP algorithm. This effectively eliminates the interference of the patient's overall head movement on the displacement trajectory, ensuring that the acquired displacement trajectory information truly reflects the deformation of facial tissues during mandibular movement. Furthermore, the aligned point cloud coordinates are normalized, further improving the comparability of displacement trajectories between different patients or different facial points. This precise and standardized displacement trajectory information significantly improves the accuracy of the subsequent facial point assessment module in extracting the stable jaw position region, thus laying a solid foundation for the model combination module to generate a trial model that highly conforms to the patient's physiological state. Ultimately, this improves the reliability and clinical applicability of virtual fitting of dental prostheses.

[0058] In some of the embodiments described above in this application, a facial point classification module is proposed to classify the displacement trajectory information of facial points. However, the motion characteristics of facial points differ significantly in different anatomical regions. If an effective and accurate classification method is lacking, it will be difficult to ensure the accuracy of subsequent stability assessment, thereby affecting the extraction accuracy of the jaw position stability region.

[0059] In response, this application further proposes that the facial point types include bony feature points, muscular feature points, and skin feature points. The preset process of the feature segmentation model includes: taking the displacement trajectory information of each facial point with known facial point types as historical displacement trajectory information, and obtaining feature parameters that characterize its data features, including trajectory curvature, instantaneous displacement vector, and variance of distance change between facial points; constructing a convolutional neural network; taking each historical displacement trajectory information and its feature parameters as input, taking the corresponding facial point type as output; and training and validating the convolutional neural network to obtain the feature segmentation model.

[0060] This application addresses the challenge of classifying facial points due to their complex and diverse motion characteristics by introducing facial point type classification and a feature segmentation model based on convolutional neural networks. Specifically, the system first classifies facial points into bony feature points, muscular feature points, and skin feature points, based on their different anatomical foundations and kinematic properties. To accurately identify these types, the system pre-defines a feature segmentation model. This model collects historical displacement trajectory information of a large amount of known facial point types and extracts key feature parameters such as trajectory curvature, instantaneous displacement vector, and variance of distance changes between facial points.

[0061] Subsequently, the constructed convolutional neural network takes these historical displacement trajectory information and their feature parameters as input and the corresponding facial point type as output. Through repeated training and model validation, it learns and establishes a complex mapping relationship between input features and facial point types. Once this feature segmentation model is trained, it can accurately identify the type of new, unclassified facial point displacement trajectory information. This refined classification provides crucial prior information for the subsequent facial point evaluation module, enabling the calculation of stability scores to employ more targeted strategies for different types of facial points. This significantly improves the accuracy and reliability of jaw stability interval extraction, thus laying the foundation for generating more accurate trial models.

[0062] As a specific implementation method, facial point types can be defined based on their anatomical location and movement characteristics. For example, bony feature points can be relatively stable bony landmarks such as the root of the nose and the highest point of the cheekbone; muscular feature points can be areas significantly affected by muscle activity, such as the corner of the mouth and the mentalis muscle; and skin feature points can be skin areas such as the middle of the cheek and the center of the forehead. In the pre-setting process of the feature segmentation model, facial point cloud sequence data and mandibular positioning data can be collected from multiple patients during oral functional activities such as chewing and pronunciation, and electromyographic signal data can be recorded simultaneously. These facial points are then classified into bony, muscular, or skin types through manual or semi-automatic annotation. For each type of facial point, feature parameters are extracted from its displacement trajectory information.

[0063] For example, trajectory curvature can be approximated by calculating the reciprocal of the radius of the arc at every three consecutive points on the trajectory; the instantaneous displacement vector can be obtained by calculating the difference in coordinates between adjacent time points; the variance of the distance change between facial points can be obtained by selecting specific pairs of facial points and calculating the variance of their distance within a preset time window. These feature parameters, along with the original displacement trajectory information, serve as input to a convolutional neural network. This convolutional neural network can use multiple one-dimensional convolutional layers to process time-series data, combined with pooling layers for feature dimensionality reduction, and finally output the classification result through a fully connected layer. During training, a cross-entropy loss function and the Adam optimizer can be used, and the model can be validated using K-fold cross-validation to ensure its generalization ability.

[0064] Through the above technical solution, this application enables intelligent classification of facial points. By dividing facial points into bony, muscular, and skin features, and utilizing convolutional neural networks to learn and recognize displacement trajectory information and its feature parameters, the problem of insufficient classification accuracy in handling complex facial movements by traditional methods is overcome. This accurate classification provides a solid foundation for subsequent facial point stability assessment, enabling the system to apply more appropriate assessment strategies for different types of facial points. This significantly improves the accuracy of jaw stability interval extraction, ultimately helping to generate a trial model that better reflects the patient's actual physiological state, thereby improving the success rate of oral restoration and patient comfort.

[0065] In other embodiments, this application uses a facial point evaluation module to obtain stability scores for each facial point within different preset time windows based on facial point classification results, displacement trajectory information, and electromyographic signal data, and extracts jaw position stability intervals accordingly. However, how to accurately and comprehensively quantify the stability of different types of facial points to ensure the accuracy of jaw position stability intervals is a technical problem that needs to be solved. Relying solely on displacement trajectory information may not fully reflect the true stability of facial points during complex chewing movements, especially for muscular feature points that are significantly affected by muscle activity; their stability assessment requires deeper consideration.

[0066] To address this, this application further proposes obtaining displacement direction indices of corresponding facial points within a preset time window based on the displacement trajectory information of each facial point. Speed ​​change index For each facial point classified as a muscular feature point, the electromyographic synchronization index of the corresponding facial point within a preset time window is obtained by combining its electromyographic signal data. ;

[0067] The stability score , The weight values ​​for displacement direction index, velocity change index, and electromyographic synchronization index are given for each facial point classified as a bony feature point or a skin feature point. , It is equal to the reciprocal of the standard deviation of the angle between the displacement direction vectors and the average direction vector of the i-th facial point within the preset time window. It is equal to the variance of the velocity change values ​​between adjacent time points of the i-th facial point within the preset time window.

[0068] The proposed solution processes the patient's facial point cloud sequence data and mandibular positioning data in the system, extracts multiple facial points and their displacement trajectory information, classifies these facial points using a feature segmentation model, and then performs a refined stability assessment of these facial points using a facial point evaluation module.

[0069] Specifically, for each facial point, the system first calculates its displacement direction index and velocity change index based on its displacement trajectory information within a preset time window. The displacement direction index reflects the stability of its motion direction by quantifying the reciprocal of the standard deviation of the angle between the various displacement direction vectors and the average direction vector of the facial point within the time window; that is, the more consistent the directions, the larger the displacement direction index. The velocity change index reflects the smoothness of its motion velocity by quantifying the variance of the velocity change values ​​between adjacent time points within the time window; that is, the smaller the velocity change, the smaller the velocity change index, and the larger its reciprocal.

[0070] Based on this, the system employs differentiated stability assessment strategies for different types of facial points according to the classification results output by the facial point classification module. For facial points classified as muscular feature points, since their movement is closely related to muscle activity, the system further combines synchronously acquired electromyographic (EMG) signal data to calculate an EMG synchronization index. This index quantifies the degree of synchronization between the movement of muscular feature points and muscle activity, providing additional and crucial biomechanical information for the stability assessment of muscular feature points. However, for facial points classified as bony or skin feature points, since their movement is less affected or indirectly influenced by muscle activity, the system does not consider the EMG synchronization index, setting its corresponding weight value to 0.

[0071] Finally, the stability score of the facial point within a preset time window is obtained by weighting and summing the inverses of the displacement direction index, the velocity change index, and the electromyographic synchronization index (if applicable). These are preset weight values ​​used to adjust the contribution of each indicator to the overall stability score. This comprehensive and differentiated stability score calculation method can more fully and accurately reflect the true stability of each facial point during chewing movements, thus providing a more reliable basis for the subsequent extraction of jaw position stability intervals.

[0072] Through the above technical solution, this application provides a more comprehensive method for assessing facial point stability. By introducing displacement direction and velocity change indices, the stability of facial points can be quantified from two dimensions: the stability and smoothness of the motion trajectory. Differential weighting is applied to different types of facial points, allowing the stability assessment to fully consider the biomechanical characteristics of the facial points themselves. This multi-dimensional, categorized stability score calculation significantly improves the accuracy of stability assessment for each facial point, thus providing a solid foundation for the precise extraction of subsequent jaw stability intervals. Ultimately, this helps generate a fitting model that better reflects the patient's actual physiological state, improving the realism and effectiveness of virtual fitting of dental prostheses.

[0073] In some embodiments described above, a stability score for each facial point within different preset time windows is obtained based on classification results, displacement trajectory information, and electromyographic (EMG) signal data. For each muscular feature point, an EMG synchronization index for the corresponding facial point within the preset time window is obtained by combining its EMG signal data. However, how to accurately and reliably quantify the synchronicity between facial point movement and muscle activity to ensure that the EMG synchronization index accurately reflects the stability of the facial point remains a technical problem that needs to be solved.

[0074] To this end, this application further proposes a process for obtaining electromyographic synchronization indices, including: obtaining the velocity values ​​of the i-th facial point at different time points within a preset time window based on the displacement trajectory information of the i-th facial point; extracting the envelope representing the muscle activity intensity of the i-th facial point at different time points within the corresponding time window based on the synchronously acquired electromyographic signal data; and obtaining the Pearson correlation coefficient r between the velocity values ​​and muscle activity intensity at different time points within the corresponding time window. .

[0075] Obtaining the velocity value of the i-th facial point at different time points within a preset time window based on its displacement trajectory information refers to determining its motion rate by analyzing the positional changes of the facial point at consecutive time points. For example, the Euclidean distance between the facial point positions at adjacent time points can be calculated and divided by the time interval to obtain the instantaneous velocity. Alternatively, the displacement trajectory of the facial point can be curve-fitted, and the velocity value can be obtained by differentiating the fitted curve.

[0076] Extracting the envelope representing the muscle activity intensity of the i-th facial point at different time points within a corresponding time window based on synchronously acquired electromyographic (EMG) signal data refers to extracting trend information reflecting the force of muscle contraction from the raw EMG signal. EMG signals are typically high-frequency and random, and their envelope can better represent the intensity changes of muscle activity. For example, a smooth envelope can be obtained by full-wave rectification of the raw EMG signal followed by low-pass filtering; alternatively, the root mean square (RMS) method can be used to calculate the effective value of the EMG signal within a sliding window to obtain its activity intensity envelope.

[0077] The Pearson correlation coefficient, calculated by measuring the velocity values ​​and muscle activity intensity at different time points within a corresponding time window, is a statistical method for quantifying the degree of linear correlation between changes in facial point velocity and changes in muscle activity intensity. The Pearson correlation coefficient ranges from -1 to 1, where 1 indicates a perfect positive correlation, -1 indicates a perfect negative correlation, and 0 indicates no linear correlation. By calculating this coefficient, the synchronicity between facial point movements and muscle activity can be objectively assessed.

[0078] The electromyographic synchronization index refers to a specific mathematical transformation of the Pearson correlation coefficient to obtain the final electromyographic synchronization index. This transformation maps the correlation coefficient to a non-negative index value that is easier to interpret and use. This approach allows the electromyographic synchronization index to better reflect the positive synchronization between facial point movements and muscle activity intensity, while avoiding the negative impact of negative correlation on stability scores.

[0079] The above technical solutions enable more precise quantification of the synchronicity between muscular feature point movement and muscle activity, allowing electromyographic synchronization indicators to more accurately reflect facial point stability. This helps improve the calculation accuracy of stability scores, making the extraction of jaw position stability zones more reliable. It provides a more accurate basis for subsequent trial model generation and prosthesis model adjustment, thereby improving the overall performance and accuracy of the virtual trial system for dental prostheses.

[0080] In some embodiments described above in this application, a method is proposed to extract jaw stability intervals from the time axis of facial point cloud sequence data based on the stability scores of each facial point within different preset time windows. However, simply determining jaw stability intervals based on the stability scores of individual facial points may have limitations. For example, some facial points may experience temporary instability due to local muscle activity or scanning noise, but this does not necessarily indicate that the entire jaw position is unstable. Therefore, more refined considerations are needed.

[0081] In response, this application further proposes to use the stability score of each facial point in different preset time windows as its stability score at each time point in the corresponding preset time window; to obtain the average stability score of each facial point belonging to the same facial point type at a single time point, and use it as the average score of the corresponding facial point type at the corresponding time point, including the average score J of bone type, the average score C of muscle type, and the average score M of skin type.

[0082] Type weight values ​​are set for the average scores of different facial point types at each time point. The type weight values ​​are positively correlated with the average scores of the corresponding facial point types at the corresponding time points. The average scores of different facial point types at a single time point are multiplied by the corresponding type weight values ​​at the corresponding time points and then summed to obtain the comprehensive score at that time point.

[0083] , , ,in, These are the type weight values ​​for the average score J of bones, the average score C of muscles, and the average score M of skin, respectively. The higher the average score, the higher the type weight value.

[0084] Extract continuous time periods with a comprehensive score consistently higher than a preset score threshold from the time axis of the facial point cloud sequence data, and select the continuous time period with the longest duration as the jaw position stability interval. If there are multiple continuous time periods with the same duration, select the continuous time period with the highest average comprehensive score.

[0085] This application's solution first assigns a stability score within a preset time window to each time point within that window, ensuring that each time point has a clear stability indicator. Based on this, at each time point, the stability scores of all facial points belonging to the same facial point type are averaged to obtain the average score for bone, muscle, and skin. This categorized averaging process effectively reduces the impact of random fluctuations in individual facial points on the overall assessment, making the stability assessment of different physiological functional areas more accurate.

[0086] Furthermore, this application sets dynamic type weight values ​​for the average scores of different facial point types, and these weight values ​​are positively correlated with the average scores of the corresponding facial point types. This means that at a certain point in time, if bony feature points exhibit higher stability, their contribution to the final comprehensive score will be amplified, thus better reflecting the overall stability of the jawbone, which is crucial for assessing jaw position stability. Subsequently, by summing these weighted average scores, a comprehensive score is obtained, which can comprehensively and dynamically reflect the patient's jaw position stability at each time point.

[0087] By identifying consecutive time periods on the timeline where the overall score consistently exceeds a preset threshold, and prioritizing the intervals with the longest duration or the highest average overall score as the jaw stability intervals, this application can more accurately and robustly locate the critical moments for jaw stability. This multi-level, dynamically weighted evaluation method effectively overcomes the limitations of single-indicator or fixed-weight evaluations, making the extracted jaw stability intervals more consistent with physiological reality and providing a more reliable foundation for subsequent prosthetic model generation.

[0088] Through the aforementioned technical solution, this application can more accurately identify continuous time periods during which a patient's jaw position is stable. This refined method for extracting stable jaw position intervals effectively avoids interference from local or transient instability factors on the overall assessment, making the obtained jaw position information more reliable. Therefore, the trial model generated based on this stable jaw position interval will more accurately reflect the patient's actual oral condition, thereby significantly improving the accuracy and reliability of virtual fitting of dental prostheses and providing a solid foundation for subsequent prosthesis design and fabrication.

[0089] In some other embodiments, this application generates a trial model based on jaw position information combined with a preset maxillary model, mandibular model, and prosthesis model. However, how to accurately construct jaw position information representing the spatial relationship between the maxilla and mandible based on the obtained jaw position stability range and different types of facial points, especially bony feature points, and generate a highly realistic trial model based on this information is a technical problem that needs to be solved.

[0090] In response, this application further proposes to obtain the point cloud coordinates of a single bony feature point within the stable jaw position range based on facial point cloud sequence data, and to use the average of these coordinates as the average point cloud coordinates of the bony feature point within the stable jaw position range. The average point cloud coordinates of each bony feature point within the stable jaw position range are then obtained and combined to form jaw position information.

[0091] The bony feature points are mapped to the corresponding positions on the surfaces of the patient's pre-set maxillary and mandibular models. The maxillary and mandibular models are then transformed in three-dimensional space until the bony feature points mapped on their surfaces are registered to the corresponding average point cloud coordinates in the jaw position information. The maxillary and mandibular models at this point are used as the patient models. The patient models are then combined with the pre-set prosthesis models to generate a trial model.

[0092] Bony feature points refer to relatively stable facial points that are not easily affected by muscle activity or skin deformation. Within the stable jaw position range, the spatial position of these points changes the least, therefore their average point cloud coordinates are more representative of the position of the bony feature point under stable jaw position. The system can continuously record the three-dimensional coordinates of each bony feature point in the facial point cloud sequence data at all time points within the stable jaw position range. Then, for each bony feature point, the arithmetic mean of all recorded coordinates within the stable range is calculated to obtain its average point cloud coordinates.

[0093] By combining the average point cloud coordinates of multiple bony feature points, a stable and representative three-dimensional point set can be formed, accurately defining the relative positions of the upper and lower jaws in a stable jaw position. The average point cloud coordinates of all bony feature points can be stored and represented as an ordered point set or a coordinate matrix. For example, a local coordinate system can be defined, and the relative positional relationships of these average point cloud coordinates in this coordinate system can be used as jaw position information.

[0094] To align the virtual model with the patient's actual jaw position information, it is necessary to establish a correlation between bony feature points obtained from the patient's face and corresponding anatomical points on pre-defined maxillary and mandibular models generated from the patient's intraoral scans or CT data. On the pre-defined maxillary and mandibular models, professionals can manually mark the anatomical points corresponding to the facial bony feature points; the three-dimensional coordinates of these marked points on the models will be recorded. Alternatively, image processing and pattern recognition technologies can be used to automatically or semi-automatically identify and map the anatomical regions or points corresponding to the facial bony feature points on the pre-defined models.

[0095] By performing rigid body transformations (translation and rotation) on the maxillary and mandibular models, the mapped points on them are precisely aligned with the jaw position information (average point cloud coordinates) obtained from the patient's face, thus placing the virtual model in the patient's actual jaw position. The Iterative Closest Point (ICP) algorithm or its variants can be used, using the average point cloud coordinates from the jaw position information as the target point set and the mapped bony feature points on the maxillary and mandibular models as the source point set. The optimal 3D transformation matrix is ​​found through iterative optimization, minimizing the distance between the source and target point sets. Alternatively, a feature-point-based registration algorithm can be used. For example, by calculating the geometric relationship between the two point sets (such as centroid, principal axis direction, etc.), the transformation matrix can be directly calculated and then applied to the maxillary and mandibular models.

[0096] After precise registration, the maxillary and mandibular models accurately reflect the patient's true jaw position within the stable jaw range, thus serving as the base models for virtual fitting. Once registration is complete, the transformed maxillary and mandibular models can be saved as a new model file containing their relative positions in a stable jaw position. Alternatively, the transformed model data can be maintained in memory and labeled as the patient model for direct use in subsequent steps.

[0097] The patient model (including the registered maxillary and mandibular models) is geometrically merged with the pre-designed prosthesis model in the same three-dimensional coordinate system. The prosthesis model is a three-dimensional model that is pre-made based on the patient's intraoral scan data and design requirements and can reflect the morphological characteristics of the prosthesis. Alternatively, the patient model and the prosthesis model can be treated as independent geometric objects, but their relative positions in three-dimensional space are fixed, together forming a logical trial model.

[0098] This application's solution focuses on bony feature points within the extracted stable jaw position region, calculating their average point cloud coordinates to form precise jaw position information. Subsequently, the corresponding bony feature points on the patient's pre-set maxillary and mandibular models are registered with this jaw position information through a three-dimensional spatial transformation, thereby obtaining a patient model that accurately reflects the patient's true jaw position relationship. Finally, this patient model is combined with the prosthesis model to generate a highly realistic trial model. This series of steps ensures that the trial model is based on the patient's physiologically stable jaw position data, effectively solving the challenge of accurately extracting stable jaw position information from dynamic facial scan data and constructing a high-precision virtual trial model.

[0099] In some other embodiments, this application obtains the contact change rate, which characterizes the stability of the trial model, by performing motion simulation on the trial model, and determines whether the contact change rate is less than a preset change threshold. However, how to perform motion simulation to accurately assess the stability of the trial model and obtain a reliable contact change rate are technical details that need to be further clarified.

[0100] In this regard, this application further proposes that the motion simulation refers to performing multi-directional offset operations on the mandibular model in the trial model, and obtaining the contact area between the prosthesis model and the maxillary and mandibular models at each offset position, which is recorded as the offset area; the contact area between the prosthesis model and the maxillary and mandibular models before the motion simulation is taken as the offset area before the offset, the ratio obtained by dividing the difference between the single offset area and the offset area by the offset area before the offset is taken as the offset change rate of the corresponding offset operation, and the average of the offset change rates of each offset operation is taken as the contact change rate of the trial model.

[0101] Determine whether the contact change rate is less than a preset change threshold. If yes, output the trial model. If no, adjust the prosthesis model until the contact change rate is less than the preset change threshold. The adjustment of the prosthesis model includes two methods: first, directly replace the prosthesis model in the current trial model with another preset prosthesis model; second, continuously change the morphological characteristics of the prosthesis model in the current trial model.

[0102] The motion simulation refers to the use of computer simulation technology to simulate the dynamic behavior of the prosthesis during oral movement. Its purpose is to pre-assess the fit and stability of the prosthesis with the patient's oral structure in a virtual environment, avoiding discomfort or functional impairment after actual fabrication. By defining multiple discrete offset vectors in three-dimensional space, the mandibular model is made to move slightly along these vector directions to simulate forces or contact situations in different directions. The area between the prosthesis model and the maxillary and mandibular models after each offset is obtained as a key indicator of the fit between the prosthesis and the oral structure.

[0103] Obtaining the post-offset area after mandibular model offset allows for the quantification of force distribution and stability of the prosthesis under different jaw positions. The contact area between the prosthesis model and the maxillary and mandibular models before motion simulation is used as the pre-offset area, serving as a baseline to measure changes in contact area after motion simulation. This represents the fit of the prosthesis in the initial, ideal jaw position. The ratio of the difference between a single post-offset area and the pre-offset area to the pre-offset area is taken as the offset change rate for the corresponding offset operation. This offset change rate quantifies the relative change in contact area of ​​the prosthesis in a specific offset direction, reflecting the stability or adaptability of the prosthesis in that direction. The average offset change rate of each offset operation is used as the contact change rate of the trial model. This average contact change rate comprehensively reflects the overall stability of the prosthesis under offset in multiple directions.

[0104] In another embodiment, this application also provides a method for virtual fitting of dental prostheses based on three-dimensional scanning, such as... Figure 2 As shown, it includes the following steps:

[0105] Simultaneously acquire facial point cloud sequence data, jaw positioning data, and electromyographic signal data within a preset detection period;

[0106] Based on the facial point cloud sequence data and mandibular positioning data, extract several facial points and their displacement trajectory information;

[0107] The displacement trajectory information of each facial point is input into a preset feature segmentation model to output a classification result representing the corresponding facial point type;

[0108] Based on the classification results, displacement trajectory information, and electromyographic signal data, the stability score of each facial point is obtained within different preset time windows. Based on the stability score, the jaw position stability interval is extracted from the time axis of the facial point cloud sequence data.

[0109] Based on the jaw position stability range, jaw position information representing the spatial relative relationship between the maxilla and mandible is obtained. Based on the jaw position information, a trial fitting model is generated by combining the preset maxillary model, mandibular model, and prosthesis model.

[0110] The trial model is subjected to motion simulation to obtain the contact change rate characterizing its stability. It is determined whether the contact change rate is less than a preset change threshold. If so, the trial model is output. If not, the prosthesis model is adjusted until the contact change rate is less than the preset change threshold.

[0111] This method effectively overcomes the interference of facial soft tissue movement on jaw position determination by fusing multi-source data and conducting dynamic stability assessment. When extracting jaw position stability intervals, it considers not only the displacement trajectory of facial points but also incorporates electromyography (EMG) signal data for simultaneous assessment of muscular feature points, thereby more accurately identifying stable periods reflecting true bony movement. For facial points classified as bony feature points, their stability scores are primarily based on displacement direction and velocity change indices; while for muscular feature points, an additional EMG synchronization index is introduced to ensure comprehensive assessment. In the model combination stage, the system uses the average position of bony feature points within the jaw position stability interval to determine jaw position information, avoiding the influence of transient instability.

[0112] Through the aforementioned technical solutions, this method can significantly improve the physiological accuracy of virtual jaw position relationships and reduce prosthesis design defects caused by inaccurate jaw position determination. This method can effectively reduce the frequency of clinical adjustments to prostheses for elderly patients and other individuals with loose facial tissues, thereby improving treatment efficiency and patient satisfaction. By employing pre-emptive functional simulation and closed-loop optimization mechanisms, the dynamic stability of the prosthesis is pre-verified in a virtual environment, significantly reducing repeated adjustments during clinical fitting and achieving precision and efficiency in the oral prosthodontic treatment process.

[0113] In another embodiment, this application also provides a computer storage medium storing computer-executable instructions, which, when executed, implement the aforementioned virtual fitting system for dental prostheses based on three-dimensional scanning.

[0114] The core innovation of this embodiment lies in constructing a complete closed-loop optimization mechanism by storing the computer-executable instructions for implementing the above system in a computer storage medium. This mechanism effectively eliminates the interference of facial soft tissue laxity on jaw position judgment by synchronously collecting multimodal physiological data and intelligently distinguishing skeletal, muscular, and skin movement trajectories; then, it automatically identifies stable jaw position intervals based on stability scores, ensuring the physiological accuracy of jaw position information; and finally, it achieves dynamic optimization of prosthesis design through contact change rate assessment in motion simulation. Because the above technical solution can accurately capture the stable jaw position state in complex situations such as elderly patients, it significantly improves the functional adaptability and aesthetic effect of prosthesis design, and reduces the number and time of clinical trials and subsequent adjustments. Overall, this embodiment solves the problem of prosthesis design defects caused by inaccurate jaw position determination in the prior art, and achieves a simultaneous improvement in the efficiency and personalized precision of oral prosthodontic treatment.

[0115] The above embodiments are only used to illustrate the technical methods of this application and are not intended to limit it. Although this application has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of this application without departing from the spirit and scope of the technical methods of this application.

Claims

1. A virtual fitting system for dental prostheses based on three-dimensional scanning, characterized in that, Includes the following modules: The data acquisition module is used to synchronously acquire facial point cloud sequence data, mandibular positioning data, and electromyographic signal data within a preset detection period; The data processing module is used to extract several facial points and their displacement trajectory information based on the facial point cloud sequence data and the mandibular positioning data. The facial point classification module is used to input the displacement trajectory information of each facial point into a preset feature partitioning model to output a classification result representing the corresponding facial point type. The facial point types include bony feature points, muscular feature points, and skin feature points. The facial point assessment module is used to obtain the stability score of each facial point in different preset time windows based on the classification results, displacement trajectory information, and electromyographic signal data, and to extract the jaw position stability interval in the time axis of the facial point cloud sequence data based on the stability score. The model combination module is used to obtain jaw position information representing the spatial relative relationship between the maxilla and mandible based on the jaw position stability range, and to generate a trial model based on the jaw position information combined with the preset maxillary model, mandibular model and prosthesis model. The model feedback module is used to perform motion simulation on the trial model to obtain the contact change rate that characterizes its stability, and to determine whether the contact change rate is less than a preset change threshold. If yes, the trial model is output; otherwise, the prosthesis model is adjusted until the contact change rate is less than the preset change threshold. The process of obtaining the stability score includes: obtaining the displacement direction index of the corresponding facial point within a preset time window based on the displacement trajectory information of each facial point. Speed ​​change index For each facial point classified as a muscular feature point, the electromyographic synchronization index of the corresponding facial point within a preset time window is obtained by combining its electromyographic signal data. ; The stability score , The weight values ​​for displacement direction index, velocity change index, and electromyographic synchronization index are given for each facial point classified as a bony feature point or a skin feature point. , It is equal to the reciprocal of the standard deviation of the angle between the displacement direction vectors and the average direction vector of the i-th facial point within the preset time window. It is equal to the variance of the velocity change values ​​of the i-th facial point between adjacent time points within a preset time window; The process of acquiring the electromyographic synchronization index includes: obtaining the velocity values ​​of the i-th facial point at different time points within a preset time window based on the displacement trajectory information of the i-th facial point; extracting the envelope representing the muscle activity intensity of the i-th facial point at different time points within the corresponding time window based on the synchronously acquired electromyographic signal data; and obtaining the Pearson correlation coefficient r between the velocity values ​​and muscle activity intensity at different time points within the corresponding time window. ; The motion simulation refers to performing multi-directional offset operations on the mandibular model in the trial fitting model, and obtaining the contact area between the prosthesis model and the maxillary and mandibular models at each offset position, which is denoted as the offset area. The contact area between the prosthesis model and the maxillary and mandibular models before motion simulation is taken as the area before offset. The ratio obtained by dividing the difference between the area after offset and the area before offset by the area before offset is taken as the offset change rate of the corresponding offset operation. The average of the offset change rates of each offset operation is taken as the contact change rate of the trial model.

2. The virtual fitting system for dental prostheses based on three-dimensional scanning according to claim 1, characterized in that, The facial points refer to points selected from different locations on the patient's face. The point cloud coordinates of a single facial point at continuous time points are obtained based on the facial point cloud sequence data. The coordinate system of the preset marker point on the patient's jaw at the same time point is obtained based on the jaw positioning data. The ICP algorithm is used to align the point cloud coordinates of each facial point at the same time point to the coordinate system at the corresponding time point, and normalize the aligned point cloud coordinates of a single facial point at consecutive time points to the displacement trajectory information of the corresponding facial point.

3. The virtual fitting system for dental prostheses based on three-dimensional scanning according to claim 1, characterized in that, The preset process of the feature segmentation model includes: The displacement trajectory information of each facial point with known facial point type is used as historical displacement trajectory information, and feature parameters that characterize its data features are obtained, including trajectory curvature, instantaneous displacement vector, and variance of distance change between facial points. A convolutional neural network is constructed, and the historical displacement trajectory information and its feature parameters are used as inputs, and the corresponding facial point type is used as output. The convolutional neural network is trained and the model is validated to obtain a feature segmentation model.

4. The virtual fitting system for dental prostheses based on three-dimensional scanning according to claim 1, characterized in that, The stability score of each facial point within different preset time windows is taken as its stability score at each time point within the corresponding preset time window. The average stability score of each facial point belonging to the same facial point type at a single time point is obtained and used as the average score of the corresponding facial point type at the corresponding time point, including the average score of bone, muscle, and skin. Type weight values ​​are set for the average scores of different facial point types at each time point. The type weight values ​​are positively correlated with the average scores of the corresponding facial point types at the corresponding time points. The average scores of different facial point types at a single time point are multiplied by the corresponding type weight values ​​at the corresponding time points and then summed to obtain the comprehensive score at that time point. Extract continuous time periods with a comprehensive score consistently higher than a preset score threshold from the time axis of the facial point cloud sequence data, and select the continuous time period with the longest duration as the jaw position stability interval. If there are multiple continuous time periods with the same duration, select the continuous time period with the highest average comprehensive score.

5. The virtual fitting system for dental prostheses based on three-dimensional scanning according to claim 1, characterized in that, Based on the facial point cloud sequence data, the point cloud coordinates of a single bony feature point within the jaw position stability interval are obtained, and the average value of the coordinates is used as the average point cloud coordinates of the bony feature point within the jaw position stability interval. The average point cloud coordinates of each bony feature point within the jaw position stability interval are obtained respectively, and they are combined into jaw position information. The bony feature points are mapped to the corresponding positions on the surfaces of the patient's preset maxillary and mandibular models. The maxillary and mandibular models are then transformed in three-dimensional space until each of the bony feature points mapped on their surfaces is registered to the corresponding average point cloud coordinates in the jaw position information. The maxillary and mandibular models at this point are used as the patient model. The patient model is then combined with the preset prosthesis model to generate a trial model.

6. A method for virtual fitting of dental prostheses based on three-dimensional scanning, characterized in that, Includes the following steps: Simultaneously acquire facial point cloud sequence data, jaw positioning data, and electromyographic signal data within a preset detection period; Based on the facial point cloud sequence data and mandibular positioning data, extract several facial points and their displacement trajectory information; The displacement trajectory information of each facial point is input into a preset feature segmentation model to output a classification result representing the corresponding facial point type. The facial point type includes bony feature points, muscular feature points, and skin feature points. Based on the classification results, displacement trajectory information, and electromyographic signal data, the stability score of each facial point is obtained within different preset time windows. Based on the stability score, the jaw position stability interval is extracted from the time axis of the facial point cloud sequence data. Based on the jaw position stability range, jaw position information representing the spatial relative relationship between the maxilla and mandible is obtained. Based on the jaw position information, a trial fitting model is generated by combining the preset maxillary model, mandibular model, and prosthesis model. The trial fitting model is subjected to motion simulation to obtain the contact change rate characterizing its stability. It is determined whether the contact change rate is less than a preset change threshold. If yes, the trial fitting model is output. If no, the prosthesis model is adjusted until the contact change rate is less than the preset change threshold. The process of obtaining the stability score includes: obtaining the displacement direction index of the corresponding facial point within a preset time window based on the displacement trajectory information of each facial point. Speed ​​change index For each facial point classified as a muscular feature point, the electromyographic synchronization index of the corresponding facial point within a preset time window is obtained by combining its electromyographic signal data. ; The stability score , The weight values ​​for displacement direction index, velocity change index, and electromyographic synchronization index are given for each facial point classified as a bony feature point or a skin feature point. , It is equal to the reciprocal of the standard deviation of the angle between the displacement direction vectors and the average direction vector of the i-th facial point within the preset time window. It is equal to the variance of the velocity change values ​​of the i-th facial point between adjacent time points within a preset time window; The process of acquiring the electromyographic synchronization index includes: obtaining the velocity values ​​of the i-th facial point at different time points within a preset time window based on the displacement trajectory information of the i-th facial point; extracting the envelope representing the muscle activity intensity of the i-th facial point at different time points within the corresponding time window based on the synchronously acquired electromyographic signal data; and obtaining the Pearson correlation coefficient r between the velocity values ​​and muscle activity intensity at different time points within the corresponding time window. ; The motion simulation refers to performing multi-directional offset operations on the mandibular model in the trial fitting model, and obtaining the contact area between the prosthesis model and the maxillary and mandibular models at each offset position, which is denoted as the offset area. The contact area between the prosthesis model and the maxillary and mandibular models before motion simulation is taken as the area before offset. The ratio obtained by dividing the difference between the area after offset and the area before offset by the area before offset is taken as the offset change rate of the corresponding offset operation. The average of the offset change rates of each offset operation is taken as the contact change rate of the trial model.

7. A computer storage medium storing computer-executable instructions, characterized in that, When the computer-executable instructions are executed, they implement the virtual fitting system for dental prostheses based on three-dimensional scanning as described in any one of claims 1-5.

Images