Detection of Gingiva Phenotype using Deep Learning

Abstract

In the field of computer science, technologies are evolving in a very rapid phase in order to solve real world problems in more efficient ways. There is huge scope in the field of medical science to solve a problem by leveraging computer science techniques and principles. Similarly, here also we tried to solve one of the problems of the dentistry field by using the state-of-the-art technologies like Artificial Intelligence and Machine Learning. Orthodontic problems are very painful and it indirectly causes various diseases. Gingiva related issues also fall in the orthodontics category and are sometimes ignored until it's too late to recover from it. The treatment of this issue is very painful and sometimes costly. Identification of gingiva is the very first step for the treatment of any Dental disease and traditional methodologies used to measure the gingival tissue are very painful and costly. To overcome this problem, we have developed an advanced algorithm which harnesses the power of Deep Convolutional Neural Network and Transfer learning to solve this issue. We synthesized the set of ROI extracted intraoral images and then fed all the images to deep CNN. Here we tried to provide an end-2-end solution to the medical practitioners in order to identify gingival phenotype from just a snap of an image. In this way by leveraging the computer science-based technologies we made this process painless and less costly.