US 12,257,110 B1
System and method for constructing a three-dimensional model of teeth and adjoining tissues in an oral cavity of at least one user
Gerard Andre Philip Liberman Paz, San Diego, CA (US); Javier Ignacio Liberman Salazar, Santiago (CL); Felipe Ignacio Pesce Bentjerodt, Santiago (CL); Carlos Julio Santander Guerra, Santiago (CL); Cristobal Gaspar Ignacio Pizarro Venegas, Santiago (CL); Diego Facundo Lazcano Arcos, Santiago (CL); Andres Garabed Baloian Gacitua, Santiago (CL); and David Caro Benado, Santiago (CL)
Filed by Gerard Andre Philip Liberman Paz, San Diego, CA (US); Javier Ignacio Liberman Salazar, Santiago (CL); Felipe Ignacio Pesce Bentjerodt, Santiago (CL); Carlos Julio Santander Guerra, Santiago (CL); Cristobal Gaspar Ignacio Pizarro Venegas, Santiago (CL); Diego Facundo Lazcano Arcos, Santiago (CL); Andres Garabed Baloian Gacitua, Santiago (CL); and David Caro Benado, Santiago (CL)
Filed on Oct. 18, 2024, as Appl. No. 18/919,689.
Application 18/919,689 is a continuation in part of application No. 18/127,945, filed on Mar. 29, 2023, granted, now 12,154,222.
This patent is subject to a terminal disclaimer.
Int. Cl. A61C 13/34 (2006.01); A61B 1/00 (2006.01); A61B 1/24 (2006.01); G06T 7/593 (2017.01)
CPC A61C 13/34 (2013.01) [A61B 1/00016 (2013.01); A61B 1/24 (2013.01); G06T 7/593 (2017.01); G06T 2207/10028 (2013.01); G06T 2207/30036 (2013.01)] 18 Claims
OG exemplary drawing
 
1. A system to build a three-dimensional model of an at least one user's teeth and adjoining tissues, the system comprising:
at least one intraoral device configured to be arranged in the oral cavity of at least one user;
at least one set of cameras arranged in the at least one intraoral device at known positions thereon and configured to capture at least one stereo image comprising at least a pair of synchronized images of the same area of the oral cavity of the at least one user; and
at least one processing medium which receives the at least one stereo image, and comprising at least one trained neural network;
wherein the at least one processing medium is arranged for:
receiving the at least one pair of synchronized images;
estimating, using the at least one pair of synchronized images, a depth map using the trained neural network, by estimating a depth of each pixel in the pair of images and using the relative position of objects that appear in both images;
generating, using at least one location and mapping block, a three-dimensional model by sequentially integrating the at least one pair of images and the depth map together;
removing, from the three-dimensional model, projected points in the model that are unrelated to adjoining tissues;
regenerating the three-dimensional model after the points are removed; and
optimizing, in the regenerated three-dimensional model, poses and depths of adjoining tissues to minimize a reprojection error between the at least one pair of synchronized images and the projected points in the three-dimensional model, thereby producing a corrected three-dimensional teeth and adjoining tissues model.