EP-4736807-A1 - METHOD FOR GENERATING DIGITAL DATA SET REPRESENTING TARGET TOOTH ARRANGEMENT

EP4736807A1EP 4736807 A1EP4736807 A1EP 4736807A1EP-4736807-A1

Abstract

One aspect of the present application provides a computer implemented method for generating a target position(s) in a selected direction(s) of a selected tooth/teeth, the method comprises: obtaining a first 3D digital model representing a patient's maxillary and mandibular dentitions under an initial tooth arrangement; obtaining a lateral cephalometric radiograph of the patient; registering the first 3D digital model with the lateral cephalometric radiograph; predicting a target position(s) in a selected direction(s) of a selected tooth/teeth of a first dentition in the lateral cephalometric radiograph using a trained first deep neural network, where the first dentition is the maxillary dentition or the mandibular dentition; and generating a target position(s) in the selected direction(s) of the selected tooth/teeth of the first dentition in the first 3D digital model based on the target position(s) in the selected direction(s) of the selected tooth/teeth of the first dentition in the lateral cephalometric radiograph and the registration result.

Inventors

WU, Huikai
YU, Wenke

Assignees

Hangzhou Chohotech Co., Ltd.

Dates

Publication Date: 20260506
Application Date: 20240105

Claims (13)

A computer implemented method for generating a target position(s) in a selected direction(s) of a selected tooth/teeth, where the method comprises: obtaining a first 3D digital model representing a patient's maxillary and mandibular dentitions under an initial tooth arrangement; obtaining a lateral cephalometric radiograph of the patient; registering the first 3D digital model with the lateral cephalometric radiograph; predicting a target position(s) in a selected direction(s) of a selected tooth/teeth of a first dentition in the lateral cephalometric radiograph using a trained first deep neural network, where the first dentition is the maxillary dentition or the mandibular dentition; and generating a target position(s) in the selected direction(s) of the selected tooth/teeth of the first dentition in the first 3D digital model based on the target position(s) in the selected direction(s) of the selected tooth/teeth of the first dentition in the lateral cephalometric radiograph and the registration result.
The method of claim 1, wherein the patient's maxillary and mandibular dentitions in the lateral cephalometric radiograph are under the initial tooth arrangement.
The method of claim 1, wherein a training data set for training the first deep neural network comprises a plurality of groups of training data, where each group of training data comprises one lateral cephalometric radiograph and a target position(s) in the selected direction(s) of the selected tooth/teeth of the first dentition in the lateral cephalometric radiograph.
The method of claim 1, wherein the target position(s) in the selected direction(s) of the selected tooth/teeth comprises one of the following: a target position in sagittal direction of an anterior tooth, a target position in vertical direction of an anterior tooth, a target position in vertical direction of a posterior tooth, a target position in sagittal direction of a posterior tooth, and any combination thereof.
A computer implemented method for generating a digital data set representing a target tooth arrangement based on the target position(s) in the selected direction(s) of the selected tooth/teeth generated by the method of claim 1, where the method comprises: preliminarily aligning the first dentition of the first 3D digital model using a trained second deep neural network; and performing a first optimization on the preliminarily aligned first dentition, where the first optimization comprises: moving the selected tooth/teeth in the preliminarily aligned first dentition to the target position(s) in the selected direction(s).
The method of claim 5, wherein the first optimization further comprises: optimizing a dental arch curve of the first dentition, and performing collision elimination and/or diastema closure on the first dentition.
The method of claim 6, wherein the movement of the selected tooth/teeth of the first dentition to the target position(s) in the selected direction(s), the optimization of the dental arch curve of the first dentition, and the collision elimination and/or diastema closure are performed at the same time.
The method of claim 6, wherein after the selected tooth/teeth is/are moved to the target position(s) in the selected direction(s), the pose(s) of the selected tooth/teeth remains unchanged in the optimization of the dental arch curve.
The method of claim 5 further comprising: preliminarily aligning a second dentition of the first 3D digital model using a trained third deep neural network based on the first dentition model having undergone the first optimization, where the second dentition is opposite to the first dentition; performing a second optimization on the preliminarily aligned second dentition, where the second optimization comprises: performing an occlusal optimization on the preliminarily aligned second dentition based on the first dentition having undergone the first optimization, so that the occlusal relationship between the second dentition having undergone the second optimization and the first dentition having undergone the first optimization meets requirements.
The method of claim 9, wherein while the occlusal optimization is performed, collision elimination and/or diastema closure are/is performed on the second dentition.
The method of claim 9, wherein the occlusal optimization is based on an energy function of occlusal relationship, and occlusal relationship is defined based on positional relationships between selected landmark points of the teeth.
The method of claim 9 further comprising: further aligning the first dentition having undergone the first optimization and the second dentition having undergone the second optimization using a trained fourth deep neural network.
The method of claim 9, wherein the first dentition is the maxillary dentition, and the second dentition is the mandibular dentition.

Description

FIELD OF THE APPLICATION The present application generally relates to a method for generating a digital data set representing a target tooth arrangement using a deep learning neural network based on predetermined constraints. BACKGROUND With rapid development of computer technologies, dental diagnosis and treatment increasingly rely on computer technologies. In the making of an orthodontic treatment plan, a 3D digital model representing a dentition under a target tooth arrangement is usually generated based on a 3D digital model representing the dentition under an initial tooth arrangement. At present, there are mainly two methods for generating a 3D digital model representing a dentition under a target tooth arrangement based on a 3D digital model representing the dentition under an initial tooth arrangement. One method is aligning teeth according to pre-formulated rules which generally derive from dentists' experiences. The Inventors of the present application discover that since shapes and poses (positions and orientations) of teeth of different patients vary greatly, it is very difficult to formulate a set of rules suitable for various situations. Therefore, this method has significant limitations. The other method is based on deep learning, for example, train a deep neural network with a training data set (comprising a plurality of sets of 3D digital models, each set of 3D digital models includes a 3D digital model representing a dentition under an initial tooth arrangement and a 3D digital model representing the dentition under a target tooth arrangement), and align teeth using the trained deep neutral network. The Inventors of the present application discover that results generated by this method might not comply with clinical medical requirements due to lack of constraints. Therefore, it is necessary to provide a new method for generating a digital data set representing a target tooth arrangement. SUMMARY In one aspect, the present application provides a computer implemented method for generating a target position(s) in a selected direction(s) of a selected tooth/teeth, the method comprises: obtaining a first 3D digital model representing a patient's maxillary and mandibular dentitions under an initial tooth arrangement; obtaining a lateral cephalometric radiograph of the patient; registering the first 3D digital model with the lateral cephalometric radiograph; predicting a target position(s) in a selected direction(s) of a selected tooth/teeth of a first dentition in the lateral cephalometric radiograph using a trained first deep neural network, where the first dentition is the maxillary dentition or the mandibular dentition; and generating a target position(s) in the selected direction(s) of the selected tooth/teeth of the first dentition of the first 3D digital model based on the target position(s) in the selected direction(s) of the selected tooth/teeth in the lateral cephalometric radiograph and the registration result. In some embodiments, the patient's maxillary and mandibular dentitions in the lateral cephalometric radiograph may be under the initial tooth arrangement. In some embodiments, a training data set for training the first deep neural network may comprise a plurality of groups of training data, each group of training data comprises one lateral cephalometric radiograph and a target position(s) in the selected direction(s) of the selected tooth/teeth of a first dentition in the lateral cephalometric radiograph. In some embodiments, the target position(s) in the selected direction(s) of the selected tooth/teeth may comprise one of the following: a target position in sagittal direction of an anterior tooth, a target position in vertical direction of an anterior tooth, a target position in vertical direction of a posterior tooth, a target position in sagittal direction of a posterior tooth, and any combination of the above. In another aspect, the present application provides a computer-implemented method for generating a digital data set representing a target tooth arrangement based on the target position(s) in the selected direction(s) of the selected tooth/teeth, the method comprises: preliminarily aligning the first dentition of the first 3D digital model using a trained second deep neural network; and performing a first optimization on the preliminarily aligned first dentition, where the first optimization comprises: moving the selected tooth/teeth in the preliminarily aligned first dentition to the target position(s) in the selected direction(s). In some embodiments, the first optimization may further comprise: optimizing the dental arch curve of the first dentition, and eliminating collision and/or closing diastema in the first dentition. In some embodiments, the moving the selected tooth in first dentition model to the target position in the selected direction may be performed simultaneously with the optimization of the dental arch curve of the first dentition model and the collis