Systems and methods for constructing a dental arch image using a machine learning model

1. A method comprising:

receiving, by a model generation engine, a two-dimensional image of a dental arch of a user;

executing, by the model generation engine, a machine learning model that is trained to receive the two-dimensional image as input and output a three-dimensional representation of the two-dimensional image, wherein in at least one iteration during training, the model generation engine:

extracts one or more features defining at least a tooth location of a tooth of the dental arch of the user from an iteration of the three-dimensional representation of the two-dimensional image generated using the machine learning model; and

determines a loss based on the one or more features extracted from the iteration of the three-dimensional representation of the two-dimensional image and a corresponding one or more features extracted from the two-dimensional image; and

outputting, by the model generation engine, an output three-dimensional representation of the two-dimensional image responsive to executing the machine learning model using the two-dimensional image of the dental arch as input.

2. The method of claim 1 , wherein the output three-dimensional representation of the two-dimensional image is a triangular model representing the dental arch of the user.

3. The method of claim 2 , wherein the triangular model is a stereolithography file.

4. The method of claim 1 , wherein the output three-dimensional representation of the two-dimensional image is a point cloud representing the dental arch of the user.

5. The method of claim 4 , further comprising generating, by the model generation engine, a plurality of points of the point cloud corresponding to one or more features having a probability of being present in the two-dimensional image.

6. The method of claim 5 , further comprising generating, by the model generation engine, a three-dimensional (3D) model based on the plurality of points of the point cloud.

7. The method of claim 1 , wherein the output three-dimensional representation of the two-dimensional image is a three-dimensional (3D) model of the dental arch of the user.

8. The method of claim 7 , further comprising manufacturing, based on the 3D model, a dental aligner specific to the user and configured to reposition one or more teeth of the user.

9. The method of claim 7 , wherein the 3D model is a first 3D model, and wherein the method further comprises generating a merged model by merging the first 3D model with a second 3D model of the dental arch of the user.

10. The method of claim 7 , wherein the 3D model is a first 3D model, the method further comprising comparing the first 3D model with a second 3D model.

11. The method of claim 10 , wherein the second 3D model is generated based on a dental impression of the dental arch of the user.

12. The method of claim 1 , wherein the two-dimensional image is multiple two-dimensional images, and wherein the multiple two-dimensional images are received by the model generation engine from a mobile device associated with the user.

13. The method of claim 1 , wherein the one or more features extracted from the two-dimensional image is a feature map representing the one or more features of the two-dimensional image.

14. The method of claim 13 , wherein the feature map includes a plurality of portions, and wherein each portion of the plurality of portions is classified based on a respective feature within the portion of the two-dimensional image as corresponding to a respective characteristic of a dentition of the user.

15. The method of claim 1 , wherein the one or more features are extracted from the two-dimensional image by executing a feature extractor model.

16. The method of claim 1 , wherein the one or more features are extracted from the iteration of the three-dimensional representation of the two-dimensional image by executing a feature extractor model.

17. The method of claim 1 , wherein the machine learning model is trained until the loss calculated based on the one or more features extracted from the iteration of the three-dimensional representation of the two-dimensional image and the corresponding one or more features extracted from the two-dimensional image satisfies a threshold.

18. The method of claim 1 , wherein the machine learning model is trained until a threshold number of iterations are performed.

19. A method comprising:

identifying, by an image detector based on a two-dimensional image of a dental arch of a user, one or more features in a portion of a plurality of portions of the two-dimensional image;

extracting, by a model generation engine, one or more features defining at least a tooth location of a tooth of the dental arch of the user from an iteration of a three-dimensional representation of the two-dimensional image, the three-dimensional representation of the two-dimensional image generated by executing a machine learning model based on the two-dimensional image;

updating, by the model generation engine, the machine learning model based on a loss calculated based on the one or more features in the portion of the plurality of portions of the two-dimensional image, a corresponding probability of the one or more features, and the one or more features extracted from the iteration of the three-dimensional representation of the two-dimensional image; and

outputting, by the model generation engine, the three-dimensional representation of the two-dimensional image based on the updated model.

20. The method of claim 19 , wherein the three-dimensional representation of the two-dimensional image is a triangular model representing the dental arch of the user.

21. The method of claim 20 , wherein the triangular model is a stereolithography file.

22. The method of claim 19 , wherein the three-dimensional representation of the two-dimensional image is a point cloud representation comprising one or more points in space corresponding to the respective probability of the one or more features.

23. The method of claim 22 , wherein the one or more points in space are determined by applying the machine learning model to the one or more features in the portion of the plurality of portions of the two-dimensional image.

24. The method of claim 22 , further comprising, generating, by the model generation engine based on the one or more points in space, a three-dimensional (3D) model of the dental arch of the user.

25. The method of claim 23 , further comprising manufacturing, based on the 3D model, a dental aligner specific to the user and configured to reposition one or more teeth of the user.

26. The method of claim 19 , wherein the one or more features in the portion of the plurality of portions of the two-dimensional image is a feature map representing one or more features of the two-dimensional image.

27. The method of claim 26 , wherein the feature map includes the plurality of portions, and wherein each portion of the plurality of portions is classified based on a respective feature within the portion of the two-dimensional image as corresponding to a respective characteristic of a dentition of the user.

28. The method of claim 19 , wherein the model generation engine updates the machine learning model based on a difference between the corresponding probability of the one or more features in the portion of the plurality of portions and the corresponding probability of the one or more features identified from the three-dimensional representation of the two-dimensional image.

29. A system comprising:

a processing circuit comprising a processor communicably coupled to a non-transitory computer readable medium, wherein the processor is configured to execute instructions stored on the non-transitory computer readable medium that cause the processor to:

receive a two-dimensional image of a dental arch of a user;

execute a machine learning model that is trained to receive the two-dimensional image as input and output a three-dimensional representation of the two-dimensional image, wherein in at least one iteration during training, the processing circuit:

extracts one or more features defining at least a tooth location of a tooth of the dental arch of the user from an iteration of the three-dimensional representation of the two-dimensional image generated using the machine learning model; and

determines a loss based on the one or more features extracted from the iteration of the three-dimensional representation of the two-dimensional image and a corresponding one or more features extracted from the two-dimensional image, wherein the one or more features extracted from the iteration of the three-dimensional representation of the two-dimensional image correspond to one or more features having a probability of being present in the two-dimensional image and the one or more features extracted from the two-dimensional image correspond to the one or more features having a probability of being present in the two-dimensional image; and

output an output three-dimensional representation of the two-dimensional image responsive to executing the machine learning model using the two-dimensional image of the dental arch as input.

30. The system of claim 29 , wherein the one or more features are extracted from the iteration of the three-dimensional representation of the two-dimensional image by executing a feature extractor model.