A method and system for denture design based on intelligent facial virtual prediction assistance

Abstract

The application discloses a kind of denture design method and system based on intelligent face virtual prediction auxiliary, comprising the following operations: (1) predict the facial appearance of patient with teeth: based on deep learning technology is combined with counter learning, predict the facial appearance of patient with teeth;(2) obtain high-precision oral soft and hard tissue three-dimensional model: based on intraoral scanning technology, CBCT three-dimensional scanning reconstruction technology, facial scanning technology and multi-source data integration technology, obtain the three-dimensional model including oral soft tissue and facial soft tissue;(3) predict jaw position relationship: based on the facial appearance of patient with teeth predicted in step (1) and the three-dimensional model of oral soft and hard tissue obtained in step (2), the jaw position relationship of patient is predicted by calibration registration technology;(4) digital tooth arrangement: based on the jaw position relationship predicted in step (3), digital tooth arrangement is carried out, and the tooth arrangement result is appropriately adjusted based on the facial appearance predicted in step (1);(5) make complete denture: according to the digital tooth arrangement result obtained in step (4), complete denture is made using cutting, CAM or 3D printing technology.

Description

Technical Field

[0001] This invention belongs to the field of oral treatment, specifically the field of intraoral dental restoration, and particularly relates to a denture design system and method based on intelligent facial virtual prediction assistance. Background Technology

[0002] The oral cavity is an important part of the human body, and its health directly affects overall health. Tooth loss, loose dentures, and reduced chewing efficiency have long been oral health problems plaguing many middle-aged and elderly people. With the aging population, the number of people in China with complete tooth loss, i.e., edentulous jaws, is increasing year by year.

[0003] Edentulous individuals who have lost all their teeth lose the cutting and grinding function of teeth, affecting the initial digestion of food in the mouth and increasing the digestive burden on the gastrointestinal tract. Their speech function is also affected, especially dental and labiodental sounds. Due to the loss of support for the lower third of the face and the soft tissues of the lips and cheeks, they exhibit signs of aging such as a shortened lower third, soft tissue collapse, deepened wrinkles, and drooping corners of the mouth. These problems can also lead to or exacerbate certain systemic diseases, such as digestive system diseases, coronary heart disease, and diabetes. Over time, the alveolar ridge, oral mucosa, temporomandibular joint, and musculoskeletal system of edentulous individuals undergo further degenerative or pathological changes.

[0004] Therefore, these edentulous patients urgently need complete denture restoration.

[0005] Furthermore, currently available handmade complete dentures often fail to achieve a proper functional fit, requiring multiple follow-up visits for adjustments after insertion. Patients endure lengthy waiting periods and painful adjustment periods, and the aesthetic results are often poor. While existing CAD / RP systems for complete dentures can shorten the design and manufacturing cycle and improve quality, they are insufficient to bridge the gap between the number of patients and the number of highly skilled dentists. Additionally, current CAD / RP systems lack the ability to perceive the anatomical and physiological characteristics of edentulous patients. When faced with the highly individualized anatomical and physiological features of different edentulous patients, they rely heavily on expert experience to accurately design complete dentures. Treatment outcomes vary significantly between different experts and design schemes. This presents a challenge to expert design skills; dentists' methods and experience in fabricating complete dentures are accumulated over a long period, involving numerous judgment conditions with a wide range of variations, and some conditions are interconnected, making them difficult to describe formally.

[0006] Therefore, a new denture design system needs to be developed to resolve the mismatch between the number of patients and the number of highly skilled dentists, and to provide intelligent and personalized denture designs for different edentulous patients. Summary of the Invention

[0007] This application integrates deep learning and adversarial learning technologies into the digital design process of complete dentures, solving the problems of lack of facial facial references and reliance on physicians' personal experience in the current clinical practice, and realizing intelligent and precise digital design of complete dentures.

[0008] This application first provides a denture design method based on intelligent facial virtual prediction assistance, including the following steps: (1) Predicting the patient's facial appearance when teeth are present: Based on deep learning technology combined with adversarial learning, predict the patient's facial appearance when teeth are present; (2) Obtaining a high-precision three-dimensional model of intraoral soft and hard tissues: Based on intraoral scanning technology, CBCT three-dimensional scanning reconstruction technology, facial scanning technology and multi-source data integration technology, obtain a three-dimensional model including intraoral soft and hard tissues and facial soft tissues; (3) Predicting jaw position relationship: Based on the patient's facial appearance when teeth are present predicted in step (1) and the three-dimensional model of intraoral soft and hard tissues obtained in step (2), predict the patient's jaw position relationship through calibration and registration technology; (4) Digital tooth arrangement: Based on the jaw position relationship predicted in step (3), perform digital tooth arrangement, and then adjust the tooth arrangement result appropriately based on the facial appearance predicted in step (1); (5) Fabricating complete dentures: Based on the digital tooth arrangement result obtained in step (4), fabricate complete dentures using cutting, CAM or 3D printing technology.

[0009] Step (1) above further includes using deep learning technology to establish a deep model of the relationship between facial scan data of different patients when they are edentulous and facial scan data after they have been fitted with dentures.

[0010] Step (1) above further includes using adversarial learning techniques to further constrain the generated facial appearance when teeth are present using facial data of normal elderly people with teeth, so that the predicted facial data when teeth are present is more consistent with the characteristics of the population.

[0011] Specifically, step (1) includes establishing a sparse point space for facial features and a dense point cloud space for facial contours when teeth are absent.

[0012] In the sparse point space of facial features, the feature point displacement is first obtained from the facial feature points when there are no teeth through the feature point prediction network. The feature point displacement is combined with the facial feature points when there are no teeth to jointly predict the facial feature points when there are teeth. Furthermore, the feature point displacement is used by the feature point displacement learning network to guide the facial appearance prediction when there are teeth in the dense point cloud space of facial contours.

[0013] Within the dense point cloud space of the facial contour, the point cloud features of the facial contour when there are no teeth are first obtained from the 3D contour of the toothless face through the point cloud prediction network. Then, the point cloud features of the facial contour when there are no teeth are concatenated and merged with the feature point displacement features obtained from the feature point displacement learning network. Subsequently, the displacement of the facial contour point cloud is predicted through a fully connected network, and the prediction of the facial contour when there are teeth is achieved by combining the 3D contour of the toothless face.

[0014] Optionally, adversarial learning is performed using normal human group feature points in the sparse point space of facial features, and adversarial learning is performed using normal human group facial contours in the dense point cloud space of facial contours.

[0015] Step (2) involves obtaining a three-dimensional model of the soft and hard tissues in the oral cavity. Combined with the three-dimensional facial data predicted in step (1), the jaw position relationship of the digital model of the upper and lower jaws of the complete denture is determined. Step (4) completes the digital tooth arrangement.

[0016] In step (5), the complete denture is fabricated using cutting, CAM, or 3D printing technologies.

[0017] This application also provides a denture design system based on intelligent facial virtual prediction assistance, which can implement the method described in any of the foregoing claims. The denture design system includes the following modules:

[0018] (1) Facial appearance prediction module when teeth are present: Based on deep learning technology and combined with adversarial learning, predict the facial appearance of patients when they have teeth.

[0019] (2) Three-dimensional model acquisition module of oral soft and hard tissues: Based on intraoral scanning technology, facial scanning technology and multi-source data integration technology, three-dimensional models of oral soft and hard tissues and facial soft tissues are acquired.

[0020] (3) Jaw position relationship prediction module: Based on the above-predicted facial appearance of the patient when teeth are present and the above-obtained three-dimensional model of soft and hard tissues in the oral cavity, the jaw position relationship of the patient is predicted by calibration and registration technology.

[0021] (4) Digital tooth arrangement module: Digital tooth arrangement is performed based on the above-predicted jaw relationship, and the tooth arrangement results are then appropriately adjusted based on the above-predicted facial appearance of the patient when they have teeth.

[0022] (5) Complete denture fabrication module: Based on the digital tooth arrangement results obtained above, complete dentures are fabricated using cutting, CAM or 3D printing technology.

[0023] The technical solution provided by this invention has at least the following beneficial effects:

[0024] (1) By combining the above artificial intelligence technology and digital technology, the present invention can improve the intelligence level of complete denture design and manufacturing, reduce human-computer interaction, improve the manufacturing quality of dentures, reduce denture manufacturing steps and manufacturing time, shorten the treatment course and number of visits for patients, and improve the accuracy of denture design and manufacturing and patient satisfaction.

[0025] (2) This invention establishes a deep learning model and combines it with adversarial learning, thereby making the predicted facial features more consistent with the distribution of the normal population and further improving the accuracy of the prediction.

[0026] (3) This invention obtains an accurate three-dimensional model of the oral soft and hard tissues and facial shape of edentulous patients by using intraoral scanning, CBCT three-dimensional scanning reconstruction technology, facial scanning and multi-source data registration technology.

[0027] (4) The present invention directly obtains the digital jaw position relationship based on the facial virtual prediction model, which is more convenient and faster and saves patients' time.

[0028] (5) The present invention makes appropriate adjustments to the facial appearance when teeth are present during the tooth arrangement process, which can make the teeth row more beautiful and neat. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 This is a schematic diagram of the deep learning and adversarial learning framework for predicting a patient's facial appearance when they have teeth, as described in this invention.

[0031] Figure 2 Three-dimensional models of the upper and lower jaws without teeth, obtained by an intraoral scanner;

[0032] Figure 3 To establish a digital jaw position relationship between the upper and lower jaws;

[0033] Figure 4 To enable digital tooth arrangement based on jaw position relationships;

[0034] Figure 5 Digital fabrication for complete dentures. Detailed Implementation

[0035] To make the technical problems, technical solutions and technical effects to be solved by the present invention clearer, each step and module will be described in detail below with reference to specific embodiments and accompanying drawings.

[0036] (a) Prediction of facial appearance when the patient has teeth

[0037] This step / module primarily uses deep learning technology combined with adversarial learning to predict the facial appearance of edentulous patients when they have teeth, providing a foundation for determining jaw alignment later. To this end, this step first uses deep learning technology to establish a deep model of the relationship between a large number of facial scans from different patients when they are edentulous and after they have had dentures installed. Then, adversarial learning technology is used to further constrain the generated facial appearance when edentulous using facial data from healthy elderly people with teeth, making the predicted facial data more consistent with the characteristics of the population.

[0038] Specifically, Figure 1 This is a schematic diagram of a deep learning framework for facial features when teeth are present. This application first establishes a sparse point space for facial features and a dense point cloud space for facial contours when teeth are absent.

[0039] In the sparse point space of facial features, the feature point displacement is first obtained from the facial feature points when there are no teeth through the feature point prediction network. The feature point displacement is combined with the facial feature points when there are no teeth to jointly predict the facial feature points when there are teeth. Furthermore, the feature point displacement is guided by the feature point displacement learning network to predict the facial appearance when there are teeth in the dense point cloud space of facial contours.

[0040] Within the dense point cloud space of the facial contour, the point cloud features of the facial contour when there are no teeth are first obtained from the 3D contour of the toothless face through the point cloud prediction network. Then, the point cloud features of the facial contour when there are no teeth are concatenated and merged with the feature point displacement features obtained from the feature point displacement learning network. Subsequently, the displacement of the facial contour point cloud is predicted through a fully connected network, and the prediction of the facial contour when there are teeth is achieved by combining the 3D contour of the toothless face.

[0041] Furthermore, adversarial learning is performed using normal population feature points in the sparse point space of facial features and adversarial learning is performed using normal population facial contours in the dense point cloud space of facial contours, thereby making the predicted facial appearance more consistent with the distribution of normal population and further improving the accuracy of prediction.

[0042] (II) Acquisition of three-dimensional models of oral cavity soft and hard tissues and facial contour

[0043] This step / module, based on intraoral scanning technology, CBCT 3D scanning and reconstruction technology, facial scanning technology, and multi-source data integration technology, acquires 3D models including intraoral soft tissue and facial soft tissue (see...). Figure 2 This allows for registration with the facial features predicted in step / module (I) above, providing a foundation for determining the jaw position relationship later.

[0044] (III) Jawline Relationship Prediction

[0045] This step / module, based on the facial appearance of the patient with teeth predicted in step / module (I) and the three-dimensional model of the intraoral soft and hard tissues and facial shape obtained in step / module (II), predicts the patient's jaw relationship using calibration and registration techniques (see...). Figure 3 ).

[0046] Using step / module (II), a precise three-dimensional model of the oral soft and hard tissues and facial shape is obtained. Based on the facial appearance of the patient when teeth are predicted in step / module (I), calibration and registration are performed to determine the jaw position relationship of the digital model of the upper and lower jaws of the complete denture.

[0047] (iv) Digital tooth alignment

[0048] This step / module performs digital tooth arrangement based on the jaw relationship predicted in step / module (III). The tooth arrangement result is then appropriately adjusted based on the facial appearance of the patient when teeth are present, as predicted in step / module (I) (see...). Figure 4 ).

[0049] In individuals with normal teeth, there is a positional relationship between the three-dimensional contour of the lip region and the arrangement of teeth within that region. Deep learning can be used to model this relationship and identify the non-linear connection between the two. In the inference phase, the facial contour when teeth are present can be predicted, and then the position of properly aligned dentures with a good occlusion can be obtained.

[0050] (V) Fabrication of complete dentures

[0051] When the denture designed in step / module (four) meets clinical needs, the design data is exported from the software for digital fabrication of the complete denture. Complete dentures can be fabricated using cutting, CAM, or 3D printing technologies (see [link]). Figure 5 ).

[0052] The above description represents the preferred embodiments of the present invention. It should be noted that, for those skilled in the art, various additions, deletions, improvements, and modifications can be made without departing from the principles of the present invention, and these additions, deletions, improvements, and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for designing dentures based on intelligent facial virtual prediction assistance, comprising the following steps: (1) Predicting the facial appearance of patients with teeth: Based on deep learning technology and combined with adversarial learning, predicting the facial appearance of patients with teeth. (2) Obtain high-precision three-dimensional models of soft and hard tissues in the oral cavity: Based on intraoral scanning technology, CBCT three-dimensional scanning and reconstruction technology, facial scanning technology and multi-source data integration technology, obtain three-dimensional models including soft and hard tissues in the oral cavity and soft tissues of the face; (3) Predicting jaw position relationship: Based on the facial appearance of the patient when teeth are present as predicted in step (1) and the three-dimensional model of soft and hard tissues in the oral cavity obtained in step (2), the jaw position relationship of the patient is predicted by calibration and registration technology; (4) Digital tooth arrangement: Digital tooth arrangement is performed based on the jaw relationship predicted in step (3), and the tooth arrangement results are then adjusted appropriately based on the facial appearance predicted in step (1). (5) Fabrication of complete dentures: Fabricate complete dentures based on the digital tooth arrangement results obtained in step (4); Step (1) further includes using deep learning technology to establish a deep model of the relationship between facial scan data of different patients when they are edentulous and facial scan data after they have been fitted with dentures. Step (1) includes establishing a sparse point space for facial features and a dense point cloud space for facial contours when teeth are absent; In the sparse point space of facial features, the displacement of facial feature points when teeth are absent is first obtained through a feature point prediction network. The displacement of feature points is then combined with the facial feature points when teeth are absent to predict facial feature points when teeth are present. Furthermore, the displacement of feature points is used to guide the prediction of facial appearance when teeth are present in the dense point cloud space of facial contours through a feature point displacement learning network. In the dense point cloud space of facial contours, the point cloud features of the facial contour when teeth are absent are first obtained through a point cloud prediction network from the 3D contour of the face when teeth are absent. Then, the point cloud features of the facial contour when teeth are absent are concatenated and merged with the feature point displacement features obtained by the aforementioned feature point displacement learning network. Subsequently, the displacement of the facial contour point cloud is predicted through a fully connected network, and the prediction of the facial contour when teeth are present is achieved by combining the 3D contour of the face when teeth are absent.

2. The denture design method based on intelligent facial virtual prediction assistance according to claim 1, wherein step (1) further includes using adversarial learning technology to further constrain the generated facial appearance when teeth are present using facial data of normal elderly people with teeth, so that the predicted facial data when teeth are present is more in line with the characteristics of the population.

3. The denture design method based on intelligent facial virtual prediction assistance according to claim 1, wherein adversarial learning is performed using normal human group feature points in the sparse point space of facial features, and adversarial learning is performed using normal human group facial contours in the dense point cloud space of facial contours.

4. The denture design method based on intelligent facial virtual prediction assistance according to claim 1, wherein the three-dimensional model of soft and hard tissues in the oral cavity obtained in step (2) is combined with the three-dimensional data of facial appearance predicted in step (1) to determine the jaw position relationship of the digital model of the upper and lower jaws of the complete denture, and digital tooth arrangement is completed through step (4).

5. The denture design method based on intelligent facial virtual prediction assistance according to any one of claims 1-4, wherein the complete denture fabrication in step (5) is completed using cutting, CAM or 3D printing technology.

6. A denture design system based on intelligent facial virtual prediction assistance, capable of implementing the method of any one of claims 1-5, the denture design system comprising the following modules: (1) Facial appearance prediction module when teeth are present: Based on deep learning technology and combined with adversarial learning, predict the facial appearance of patients when teeth are present. (2) Module for acquiring three-dimensional models of soft and hard tissues in the oral cavity: Based on intraoral scanning technology, CBCT three-dimensional scanning and reconstruction technology, facial scanning technology and multi-source data integration technology, three-dimensional models of soft and hard tissues in the oral cavity and facial soft tissues are acquired; (3) Jaw position relationship prediction module: Based on the above-predicted facial appearance of the patient when teeth are present and the above-obtained three-dimensional model of soft and hard tissues in the oral cavity, the jaw position relationship of the patient is predicted by calibration and registration technology; (4) Digital tooth arrangement module: Digital tooth arrangement is performed based on the above-predicted jaw relationship, and the tooth arrangement result is then appropriately adjusted based on the above-predicted facial appearance of the patient when they have teeth. (5) Complete denture fabrication module: Based on the digital tooth arrangement results obtained above, complete dentures are fabricated using cutting, CAM or 3D printing technology.

Images