# US 2021/0279871 A1 · Dental imaging system utilizing artificial intelligence > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2021/0279871 A1 - Document ID: US20210279871A1 - Status: Published application - Kind: A1 - Application number: 17/329,969 - Filed: 2021-05-25 - Published: 2021-09-09 - Earliest claimed: 2018-04-25 - Assignees: Sota Precision Optics, Inc. - Applicants: SOTA Precision Optics, Inc. - Inventors: Dustin JOHNSON - Web page: https://patentgrep.com/patent/US20210279871A1 ## CPC - A61B 6/51: for dentistry - A61B 6/5294: {involving using additional data, e.g. patient information, image labeling, acquisition parameters} - A61C 7/002: {Orthodontic computer assisted systems} - A61C 9/004: {Means or methods for taking digitized impressions} - A61C 9/0046: {Data acquisition means or methods} - A61C 9/0053: {Optical means or methods, e.g. scanning the teeth by a laser or light beam} - G06F 18/214: Generating training patterns; Bootstrap methods, e.g. bagging or boosting - G06N 3/04: Architecture, e.g. interconnection topology - G06N 3/045: Combinations of networks - G06N 3/0464: Convolutional networks [CNN, ConvNet] - G06N 3/048: Activation functions - G06N 3/08: Learning methods - G06N 3/084: Backpropagation, e.g. using gradient descent - G06N 3/09: Supervised learning - G06T 5/60: using machine learning, e.g. neural networks - G06T 7/0012: {Biomedical image inspection} - G06T 7/11: Region-based segmentation - G06T 7/12: Edge-based segmentation - G16H 30/20: for handling medical images, e.g. DICOM, HL7 or PACS - G16H 30/40: for processing medical images, e.g. editing - G16H 40/20: for the management or administration of healthcare resources or facilities, e.g. managing hospital staff or surgery rooms - G16H 50/20: for computer-aided diagnosis, e.g. based on medical expert systems - G16H 50/70: for mining of medical data, e.g. analysing previous cases of other patients - G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks - G06T 2207/10116: X-ray image - G06T 2207/20081: Training; Learning - G06T 2207/20084: Artificial neural networks [ANN] - G06T 2207/30036: Dental; Teeth - G06T 2207/30168: Image quality inspection - G06V 2201/03: Recognition of patterns in medical or anatomical images - A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D) - A61C: DENTISTRY; APPARATUS OR METHODS FOR ORAL OR DENTAL HYGIENE (non-driven toothbrushes A46B; {tongue scrapers A61B17/24;} preparations for dentistry A61K6/00; preparations for cleaning the teeth or mouth A61K8/00, A61Q11/00) - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL - G16H: HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING ## Abstract A system and a method for training and utilizing deep convolutional neural networks to perform diagnostic and image enhancement operations on images of dentition uses digital images and other auxiliary parameters as inputs for a convolutional neural network. The neural network can output a tooth segmentation map, tooth identifiers, a probability map indicating the presence of caries, cavities or other dental anomalies/conditions, and recommended parameters for use in image enhancement algorithms. ## Claims ### Claim 1 A computer implemented method of training an artificial neural network for automatic dental image manipulation, enhancement, labeling, tagging, and recognition of abnormalities in dentition, caries and/or cavities, the method comprising: receiving at least one set of training images and their associated labels, wherein the associated labels include a quality score for each image in the at least one set of training images; passing the set of training images and the associated labels to an artificial neural network; using a technique of backpropagation to determine statistical weights corresponding to a likelihood that values of each pixel or a pooled sub-region of pixels in an input image collectively represent a structure or state that a neural network model is intended to identify; and determining image characteristics of an acquired image. ### Claim 2 (depends on 1) The method of claim 1, further comprising: adjusting the image characteristics to improve the quality of the acquired image. ### Claim 3 (depends on 1) The method of claim 1, further comprising: adjusting the image characteristics dependent upon a specific field of dentistry or specific problems related to the dentition in question. ### Claim 4 (depends on 1) The method of claim 1, further comprising: alerting a user if the image quality is below a predetermined acquired image quality score. ### Claim 5 (depends on 1) The method of claim 1, wherein the at least one set of training images are human dental images. ### Claim 6 (depends on 1) The method of claim 1, wherein the at least one set of training images are animal dental images. ### Claim 7 (depends on 1) The method of claim 1, wherein: the target artificial neural network that is passed the training data set utilizes one or more of the following: one or more convolution layers, one or more rectified linear units (ReLUs), one or more pooling layers, one or more fully connected layers, and includes at least one input layer, one hidden layer, and one output layer. ### Claim 8 (depends on 7) The method of claim 7, wherein: the at least one set of training images includes images that have been labeled according to one or a combination of the following: an image orientation, the teeth contained within the image, parameters associated with image enhancements that have been applied to the image, what anomalies in the dentition are present. ### Claim 9 (depends on 8) The method of claim 8, wherein: the at least one set of training images includes images that have been labeled with pixel coordinates of rectangles containing locations of the teeth. ### Claim 10 (depends on 8) The method of claim 8, wherein the at least one set of training images includes images that have been labeled with a segmentation map corresponding to locations of the teeth. ### Claim 11 (depends on 8) The method of claim 8, wherein: the step of receiving at least one set of training images is representative of one or more of the following scenarios: i) the at least one set of training images includes images oriented in such a way that each possible multiple of 90° rotations are represented, with labels indicating an amount of rotation for each image; ii) the at least one set of training images includes images representative of each tooth typically present in the dentition of both anatomically normal and abnormal adults and/or children, with labels indicating what teeth are present in each image; iii) the at least one set of training images includes images representing a statistically significant portion of an image enhancement parameter space with labels indicating enhancement parameters and a qualitative rating of those parameters; iv) the at least one set of training images includes images representing dental anomalies and/or dental conditions, in addition to healthy dentition, with labels indicating the presence of an anomaly and/or a dental condition, or a plurality of anomalies and/or dental conditions; and v) images in the at least one set of training images have been perturbed or manipulated to augment the size of the training set to the benefit of the resulting model. ### Claim 12 (depends on 8) The method of claim 8, wherein: the one or more convolution layers of the target artificial neural network can apply convolution to one or more of: consecutively overlapping sub-regions of each image in the at least one set of training images, or to consecutive, non-overlapping sub-regions of each image in the at least one set of training images, or to non-consecutive overlapping or non-overlapping sub-regions of each image in the at least one set of training images. ### Claim 13 (depends on 1) A computer system having non-transitory computer readable media with imaging software embedded therein, the imaging software including code which when executed on a processor of the system, causes the system to implement the artificial neural network trained by the method of claim 1. ### Claim 14 (depends on 13) The computer system of claim 13 wherein: the artificial neural network includes parameters that dictate an operation and behavior of the network, which are trained to minimize error during the process of backpropagation; and the neural network outputs a distribution map indicating a per-pixel or per-sub-region likelihood of each pixel or sub-region being part of a structure or state that the neural network is intended to locate and/or identify. ### Claim 15 (depends on 14) The computer system of claim 14 wherein: the neural network outputs a bounding box or segmentation map derived from the per-pixel or per-sub-region likelihood of each pixel or sub-region being part of the structure or state that the neural network is intended to locate and/or identify. ### Claim 16 (depends on 14) The computer system of claim 14, wherein: the neural network outputs labels or tags corresponding to the state or structure that the neural network is intended to locate and/or identify. ### Claim 17 (depends on 14) The computer system of claim 14, wherein: the neural network outputs one or more image filter parameters optimized for an input image and representative of contrast, gamma, sharpness, and/or other related enhancements.