Computer-implemented detection and processing of oral features
Described herein are computer-implemented methods for analyzing an input image of a mouth region from a user to provide information regarding a disease or condition of the mouth region, a computing device configured to receive the input images from a user; and a trained machine learning system. In some embodiments, the computing device is configured to transmit an oral health score to the user.
1 . A system for analyzing a mouth region to determine a disease or condition of the mouth region, the system comprising:
a trained machine learning system comprising at least one processor and trained models, wherein the models are trained using a dataset of training images, wherein the dataset is partitioned into a first subset of training images and a second subset of validation images, the dataset comprising one or both of: a dental caries feature and a periodontitis feature, the trained machine learning system further configured to:
receive one or more images of the mouth region;
pre-process the one or more images to extract image features;
analyze the extracted image features to generate a prediction based on a recognized feature within each of the one or more images; and
generate an oral health score for each of the one or more images corresponding to the recognized feature associated with one or both of: the dental caries feature and the periodontitis feature.
2 . The system of claim 1 , wherein the dataset comprises images having one or more of: a resolution of about 32×32 to about 2048×2048; a greyscale; and a rectangular shape.
3 . The system of claim 1 , wherein each image of the dataset is cropped to provide a cropped image having one or both of: the dental caries feature and the periodontitis feature.
4 . The system of claim 3 , wherein the models are trained to recognize both the dental caries feature and the periodontitis feature.
5 . The system of claim 1 , wherein the machine learning system further processes one or more of the images of the dataset to provide additional images for the dataset, and the processes performed on the subset comprise one or more of: adding noise, adjusting a contrast, adjusting a brightness, blurring, sharpening, flipping, rotating, adjusting a white balance, adjusting a color, or equivalents thereof.
6 . The system of claim 5 , wherein the processes may be performed dynamically at a time of training.
7 . The system of claim 1 , wherein pre-processing the one or more images to extract image features further comprises one or more of: adjusting a resolution of each of the one or more images; or converting each of the one or more images into a greyscale image.
8 . The system of claim 1 , wherein each of the trained models are stacked to provide the prediction.
9 . The system of claim 1 , further comprising a user interface configured for interaction with a user using one or both of: an application residing on a smartphone or a website associated with a computing device.
10 . The system of claim 9 , wherein the user interface is configured for interaction with a user by providing visual aids to assist the user in capturing the one or more images of the mouth region.
11 . The system of claim 10 , wherein the visual aids include one or more of: frames, lines, points, geometric shapes, or combinations and equivalents thereof, in order to align, angle, or distance of an image sensor to different areas inside the mouth region.
12 . The system of claim 1 , wherein each of the one or more images is of a different area in the mouth region.
13 . The system of claim 1 , wherein the oral health score comprises a score for each individual tooth for each input of the one or more images.
14 . The system of claim 1 , wherein the oral health score comprises a score for each gum region for each of the one or more images.
15 . The system of claim 1 , wherein the oral health score comprises a score for each individual tooth and gum region for each of the one or more images.
16 . The system of claim 1 , wherein the processor is configured to transmit the oral health score to a user.
17 . The system of claim 1 , wherein the oral health score comprises an indication of one or more of: dental caries, periodontitis, gingivitis, fillings, toothbrush abrasion, dental erosion, teeth sensitivity, oral cancer, cracked or broken teeth, mouth sores, halitosis, abscess, congenital tooth conditions, tongue disease, and one or more cosmetic conditions.
18 . A method for analyzing a mouth region to provide information regarding a disease or condition of the mouth region, the method comprising:
at a trained machine learning system comprising at least one processor and trained models, wherein the models are trained using a dataset of training images, wherein the dataset is partitioned into a first subset of training images and a second subset of validation images, and wherein the dataset comprises one or both of: a dental caries feature and a periodontitis feature:
receiving one or more images;
pre-processing the one or more images to extract image features;
analyzing the extracted image features to generate a prediction based on a recognized feature within each of the one or more images; and
generating an oral health score for each of the one or more images corresponding to the recognized feature associated with one or both of the dental caries feature and the periodontitis feature.
19 . The method of claim 18 , wherein the dataset are images having one or more of: a resolution of about 32×32 to about 2048×2048; a greyscale; and a rectangular shape.
20 . The method of claim 18 , further comprising processing a subset of the dataset to provide additional images for the dataset by one or more of: adding noise, adjusting a contrast, adjusting a brightness, blurring, sharpening, flipping, rotating, adjusting a white balance, adjusting a color, or equivalents thereof.
21 . The method of claim 20 , wherein the processes may be performed dynamically at a time of training.
22 . The method of claim 18 , wherein pre-processing the one or more images further comprises converting the one or more images from a 3-channel image to a 1-channel image.
23 . The method of claim 18 , wherein pre-processing the one or more input-images further comprises adjusting a resolution to resize the one or more images.
24 . The method of claim 18 , further comprising outputting, from the processor, visual aids in order to capture the one or more images.
25 . The method of claim 24 , wherein the visual aids comprise one or more of: frames, lines, points, geometric shapes, or combinations and equivalents thereof, in order to align, angle, or distance of an image sensor to different areas inside the mouth region.
26 . The method of claim 18 , further comprising transmitting, using the processor, the oral health score to a user.
27 . The method of claim 18 , wherein the oral health score comprises an indication of one or more of: dental caries, periodontitis, gingivitis, fillings, toothbrush abrasion, dental erosion, teeth sensitivity, oral cancer, cracked or broken teeth, mouth sores, halitosis, abscess, congenital tooth conditions, tongue disease, and one or more cosmetic conditions.