Dental Cavity Detection Using CNN models From X-ray Images

Dental caries, often known as tooth decay, stands as one of the most widespread and persistent dental issues, impacting countless individuals throughout their lives. It remains a recurring challenge for oral health worldwide. Dentists treat this cavity by clinical and visual inspection along with the x-ray images. Most of the time it may be difficult to identify cavity in the photograph and may be mistaken because of the shadows for various reasons, which also include poor x-rays quality. With the rise of machine learning in the field of dentistry, images serve a major role in finding the diverse variety of dental issues over the population. They can be specially utilised for finding the decay, that measures dental health. In this study, deep learning algorithms are utilized to automatically detect dental caries in intraoral periapical (IOPA) x-ray pictures. The proposed technique uses newly trained model which includes DenseNet121, Xception, VGG 16, Inception and Renet50 for extraction of essential features from the images and to classify them as caries or non-caries. Experiments are conducted on 4000 augmented images. The obtained results demonstrate that models like DenseNet121(99%), Xception(84%) and VGG 16(83%) produces a high accurate detection of caries and non-caries.