Gum Disease Detection in the Front Part of the Mouth Using Convolutional Neural Network Through the Use of Keras with TensorFlow as Backend

Gum disease is one of the most common problems when it comes to oral health. Majority of people are unaware of their status of the gums whether it falls under which of the categories namely healthy, gingivitis, periodontitis, or advance periodontitis. People should be informed about the status of their gums so they could do preemptive action or maintenance. The number of individuals which have their own smartphones that are used daily both for necessity and entertainment are increasing. It is with ease for the users the accessibility of the application by installing the file in their phones. The objective of this study is to develop a smartphone application based on convolutional neural network that will allow the users to capture images of their gums and detect even the earliest stage of gum disease -- gingivitis. The application was created using Android Studio which serves as the integrated development environment for Android application development, rendering the application to be specifically used for Android OS. Keras framework running on top of TensorFlow was used to define and train the CNN model. The experiment was done on 200 gum images and 50 physical examinations. The accuracy of the application was verified by four dentists. The result of the study shows an accuracy of 83.5% in classifying gum diseases.