Digital Diagnosis of Mouth Disease Using Deep Learning Algorithms

The ability to identify and diagnose oral and mouth problems has significantly increased with the use of digital instruments. Those with potentially malignant oral diseases have an increased risk of developing lip or oral cavity cancer. Poor brushing, hormone fluctuations, and flossing practices that let plaque, a sticky layer of germs, are typically the causes. Since the mouth is thought to be a reflection of one's overall health, there is presently no personal gadget available to track one's dental health. Mouth disease to be correctly diagnosed and treated, an accurate forecast is necessary. It is possible to prevent dental disorders that gradually weaken tooth roots by identifying them early on. The goal of this effort is to develop a low-cost, multimodal, personal oral sensing device that perceives and classifies data automatically, enabling the physician to diagnose patients early and treat them effectively. The mouth disease prediction consists of preprocessing and classification process. The first step in mouth disease prediction wiener filter applied to filter the noise in testing image of dataset. After that augmentation involved to expand the number of images from the limited images. Finally, the comparison algorithms such as federated learning, multilayer perceptron (MLP) and deep belief networks (DBN) to produce the outputs as precision, recall, accuracy was analyzed. The 85% accuracy of the deep belief network predict mouth disease for more efficient.