Methods and systems for forming a three-dimensional model of dentition

Patent metadata

Field	Value
Publication number	US-12376943-B2
Application number	US-202318340025-A
Country	US
Kind code	B2
Filing date	Jun 22, 2023
Priority date	Dec 23, 2019
Publication date	Aug 5, 2025
Grant date	Aug 5, 2025

Abstract

Provided herein are systems and methods for optimizing a 3D model of an individual's teeth. A 3D dental model may be reconstructed from 3D parameters. A differentiable renderer may be used to derive a 2D rendering of the individual's dentition. 2D image(s) of an individual's dentition may be obtained, and features may be extracted from the 2D image(s). Image loss between the 2D rendering and the 2D image(s) can be derived, and back-propagation from the image loss can be used to calculate gradients of the loss to optimize the 3D parameters. A machine learning model can also be trained to predict a 3D dental model from 2D images of an individual's dentition.

First claim

What is claimed is:

1. A method of forming a three-dimensional (3D) model of an individual's dentition, the method comprising:
obtaining one or more two-dimensional (2D) original images depicting at least a portion of the individual's dentition;
extracting features from the one or more 2D original images;
providing the extracted features to a trained neural network that is trained to construct a parametric 3D dental model using the extracted features and a set of network weights, wherein the network weights are based on results from a loss function comparing a plurality of 2D original images of historical patient dentitions with corresponding one or more 2D differentiable renderings generated based on 3D parameters associated with the patient dentitions; and
outputting, from the trained neural network, the parametric 3D dental model of the individual's dentition.

2. The method of claim 1 , wherein the network weights are based on gradients of loss determined from the results from the loss function.

3. The method of claim 1 , wherein results of the loss function are derived from a pixel by pixel comparison between the plurality of 2D original images of the historical patient dentitions and the corresponding one or more 2D differential renderings.

4. The method of claim 1 , wherein the 3D parameters for a particular dentition of the historical patient dentitions comprise one or more of: a local translation for each of a plurality of teeth of the particular patient dentition, a local rotation for each of a plurality of teeth of the particular patient dentition, a global translation for a jaw of the particular patient dentition, and a global rotation for a jaw of the particular patient dentition.

5. The method of claim 1 , wherein extracting features from the one or more 2D original images comprises extracting one or more of: a tooth mask for an upper jaw or a lower jaw; tooth segmentation data; tooth numbering data; and dental edge information.

6. The method of claim 1 , further comprising comparing the parametric 3D dental model to an orthodontic treatment plan of the individual and determining whether the individual's dentition is tracking the orthodontic treatment plan.

7. The method of claim 6 , further comprising modifying the orthodontic treatment plan based on the comparison between the parametric 3D dental model and the orthodontic treatment plan.

8. A system for forming a three-dimensional (3D) dental model of an individual's dentition, the system comprising:
one or more processors; and
a memory coupled to the one or more processors, the memory storing computer-program instructions, that, when executed by the one or more processors, perform a computer-implemented method comprising:
obtaining one or more two-dimensional (2D) original images depicting at least a portion of the individual's dentition;
extracting features from the one or more 2D original images;
providing the extracted features to a trained neural network that is trained to construct a parametric 3D dental model using the extracted features and a set of network weights, wherein the network weights are based on results from a loss function comparing a plurality of 2D original images of historical patient dentitions with corresponding one or more 2D differentiable renderings generated based on 3D parameters associated with the patient dentitions; and
outputting, from the trained neural network, the parametric 3D dental model of the individual's dentition.

9. The system of claim 8 , wherein the network weights are based on gradients of loss determined from the results from the loss function.

10. The system of claim 8 , wherein results of the loss function are derived from a pixel by pixel comparison between the plurality of 2D original images of the historical patient dentitions and the corresponding one or more 2D differential renderings.

11. The system of claim 8 , wherein the 3D parameters for a particular dentition of the historical patient dentitions comprise one or more of: a local translation for each of a plurality of teeth of the particular patient dentition, a local rotation for each of a plurality of teeth of the particular patient dentition, a global translation for a jaw of the particular patient dentition, and a global rotation for a jaw of the particular patient dentition.

12. The system of claim 8 , wherein extracting features from the one or more 2D original images comprises extracting one or more of: a tooth mask for an upper jaw or a lower jaw; tooth segmentation data; tooth numbering data; and dental edge information.

13. The system of claim 8 , further comprising comparing the parametric 3D dental model to an orthodontic treatment plan of the individual and determining whether the individual's dentition is tracking the orthodontic treatment plan.

14. The system of claim 13 , further comprising modifying the orthodontic treatment plan based on the comparison between the parametric 3D dental model and the orthodontic treatment plan.

15. A non-transitory, computer-readable medium including contents that are configured to cause one or more processors to perform a method comprising:
obtaining one or more two-dimensional (2D) original images depicting at least a portion of the individual's dentition;
extracting features from the one or more 2D original images;
providing the extracted features to a trained neural network that is trained to construct a parametric 3D dental model using the extracted features and a set of network weights, wherein the network weights are based on results from a loss function comparing a plurality of 2D original images of historical patient dentitions with corresponding one or more 2D differentiable renderings generated based on 3D parameters associated with the patient dentitions; and
outputting, from the trained neural network, the parametric 3D dental model of the individual's dentition.

16. The non-transitory, computer-readable medium of claim 15 , wherein the network weights are based on gradients of loss determined from the results from the loss function.

17. The non-transitory, computer-readable medium of claim 15 , wherein results of the loss function are derived from a pixel by pixel comparison between the plurality of 2D original images of the historical patient dentitions and the corresponding one or more 2D differential renderings.

18. The non-transitory, computer-readable medium of claim 15 , wherein the 3D parameters for a particular dentition of the historical patient dentitions comprise one or more of: a local translation for each of a plurality of teeth of the particular patient dentition, a local rotation for each of a plurality of teeth of the particular patient dentition, a global translation for a jaw of the particular patient dentition, and a global rotation for a jaw of the particular patient dentition.

19. The non-transitory, computer-readable medium of claim 15 , wherein extracting features from the one or more 2D original images comprises extracting one or more of: a tooth mask for an upper jaw or a lower jaw; tooth segmentation data; tooth numbering data; and dental edge information.

20. The non-transitory, computer-readable medium of claim 15 , further comprising comparing the parametric 3D dental model to an orthodontic treatment plan of the individual and determining whether the individual's dentition is tracking the orthodontic treatment plan.

21. The non-transitory, computer-readable medium of claim 20 , further comprising modifying the orthodontic treat

Assignees

• Align Technology Inc

Inventors

• Li Guotu
• Cramer Christopher E

Classifications

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06T17/20 Primary

  Finite element generation, e.g. wire-frame surface description, {tesselation} · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T2207/30036

  Dental; Teeth · CPC title