Neural network-based generation and placement of tooth restoration dental appliances
Techniques are described for automating the design of dental restoration appliances using machine learning models. An example computing device receives transform information associated with a current dental anatomy of a dental restoration patient, provides the transform information associated with the current dental anatomy of the dental restoration patient as input to a machine learning model trained with transform information indicating placement of a dental appliance component with respect to one or more teeth of corresponding dental anatomies, the dental appliance being used for dental restoration treatment for the one or more teeth, and executes the machine learning model using the input to produce placement information for the dental appliance component with respect to the current dental anatomy of the dental restoration patient.
1 . A computing device comprising:
an input interface configured to receive transform information associated with a current dental anatomy of a dental restoration patient; and
an engine configured to:
provide the transform information associated with the current dental anatomy of the dental restoration patient as input to a machine learning model trained with transform information indicating placement of a dental appliance component with respect to one or more teeth of corresponding dental anatomies, the dental appliance component being used for dental restoration treatment for the one or more teeth; and
execute the machine learning model using the input to produce placement information for the dental appliance component with respect to the current dental anatomy of the dental restoration patient.
2 . The computing device of claim 1 ,
wherein the transform information indicating the placement of the dental appliance component with respect to the one or more teeth of the corresponding dental anatomies comprises a single matrix transform corresponding to each of the one or more teeth of the corresponding dental anatomies,
wherein the training data further comprise a ground truth transform indicating a post-completion placement of the dental appliance component, and
wherein to produce the placement information for the dental appliance component with respect to the current dental anatomy of the dental restoration patient, the engine is configured to output a matrix transform for the dental appliance component with respect to dental restoration treatment of the current dental anatomy.
3 . The computing device of claim 2 , wherein each single matrix transform corresponding to each of the one or more teeth is a respective 4×4 transform corresponding to each of the one or more teeth, and wherein the matrix transform output for the dental appliance component is a 4×4 transform output for the dental appliance component.
4 . The computing device of claim 3 , wherein the 4×4 transform output for the dental appliance component with respect to the dental restoration treatment comprises a {translation, rotation, scale} tuple indicating change information for a {position, orientation, size} tuple describing the placement information for the dental appliance component with respect to the current dental anatomy of the dental restoration patient.
5 . The computing device of claim 4 , wherein the one or more teeth comprise two maxillary central incisors, wherein the engine is further configured to train the machine learning model by:
converting each 4×4 transform for each of the two maxillary central incisors to a respective 1×7 quaternion vector; and
concatenating the 1×7 quaternion vectors corresponding to the two maxillary central incisors to form a 1×14 feature vector, and
wherein to provide the transform information as the input to the machine learning model, the engine is configured to provide the 1×14 feature vector as the input to the machine learning model.
6 . The computing device of claim 5 , wherein the dental appliance component is a center clip.
7 . The computing device of claim 6 , wherein to train the machine learning model, the engine is configured to use a backpropagation algorithm with the respective 4×4 transforms corresponding to the one or more teeth of the corresponding dental anatomies.
8 . A method comprising:
receiving transform information associated with a current dental anatomy of a dental restoration patient;
providing the transform information associated with the current dental anatomy of the dental restoration patient as input to a machine learning model trained with transform information indicating placement of a dental appliance component with respect to one or more teeth of corresponding dental anatomies, the dental appliance component being used for dental restoration treatment for the one or more teeth; and
executing the machine learning model using the input to produce placement information for the dental appliance component with respect to the current dental anatomy of the dental restoration patient.
9 . The method of claim 8 ,
wherein the transform information indicating the placement of the dental appliance component with respect to the one or more teeth of the corresponding dental anatomies comprises a single matrix transform corresponding to each of the one or more teeth of the corresponding dental anatomies,
wherein the training data further comprise a ground truth transform indicating a post-completion placement of the dental appliance component, and
wherein producing the placement information for the dental appliance component with respect to the current dental anatomy of the dental restoration patient comprises outputting a matrix transform for the dental appliance component with respect to dental restoration treatment of the current dental anatomy.
10 . The method of claim 9 , wherein each single matrix transform corresponding to each of the one or more teeth is a respective 4×4 transform corresponding to each of the one or more teeth, and wherein the matrix transform output for the dental appliance component is a 4×4 transform output for the dental appliance component.
11 . The method of claim 10 , wherein the 4×4 transform output for the dental appliance component with respect to the dental restoration treatment comprises a {translation, rotation, scale} tuple indicating change information for a {position, orientation, size} tuple describing the placement information for the dental appliance component with respect to the current dental anatomy of the dental restoration patient.
12 . The method of claim 11 , wherein the one or more teeth comprise two maxillary central incisors, the method further comprising training the machine learning model by:
converting each 4×4 transform for each of the two maxillary central incisors to a respective 1×7 quaternion vector; and
concatenating the 1×7 quaternion vectors corresponding to the two maxillary central incisors to form a 1×14 feature vector,
wherein providing the transform information as the input to the machine learning model comprises providing the 1×14 feature vector as the input to the machine learning model.
13 . The method of claim 12 , wherein training the machine learning model comprises using a backpropagation algorithm with the respective 4×4 transforms corresponding to the one or more teeth of the corresponding dental anatomies.
14 . The method of claim 13 , wherein the dental appliance component is a center clip, and wherein the placement information for the center clip identifies a midpoint between two teeth of the current dental anatomy of the dental restoration patient.
15 . The method of claim 14 , wherein the dental appliance component is a rear snap clamp, and wherein the placement information for the rear snap claim identifies a midpoint between the two teeth of the current dental anatomy of the dental restoration patient.
16 . The method of claim 15 , wherein the two teeth of the current dental anatomy of the dental restoration patient comprise a molar and a canine of the current dental anatomy of the dental restoration patient.
17 . The method of claim 15 , wherein the two teeth of the current dental anatomy of the dental restoration patient comprise a molar and a premolar of the current dental anatomy of the dental restoration patient.
18 . The method of claim 11 , wherein the machine learning model is a neural network.
19 . A non-transitory computer-readable storage medium encoded with instructions that, when executed, cause one or more processors of a computing system to:
receive transform information associated with a current dental anatomy of a dental restoration patient;
provide the transform information associated with the current dental anatomy of the dental restoration patient as input to a machine learning model trained with transform information indicating placement of a dental appliance component with respect to one or more teeth of corresponding dental anatomies, the dental appliance component being used for dental restoration treatment for the one or more teeth; and
execute the machine learning model using the input to produce placement information for the dental appliance component with respect to the current dental anatomy of the dental restoration patient.
20 . The non-transitory computer-readable storage medium of claim 19 , wherein the machine learning model is a neural network.