Non Invasive Approach for Dental Caries Detection

• Main Authors: Mehta, Leena Rohan, Borse, Megha
• Format: Conference Proceeding
• Language: English
• Subjects: accuracy; confusion matrix; Convolution; Convolutional neural network; Convolutional neural networks; dental caries; Dentistry; Diseases; precision; Testing; Training
• ISSN: 2831-5022
• DOI: 10.1109/PuneCon58714.2023.10450116

In developing nation like India, dental health is highly ignored health domain. This is due to the lack of awareness, lack of education, socio-economic background and poverty. Hence an automated system of dental caries detection is necessary which will reduce the burden on dental practitioners. The presented system focuses on dental caries detection using convolutional neural network with four layers of convolution. Web based database of 74 single tooth digital RGB images are used for the study. 60 images were used for training and 14 were used for testing. Data augmentation is done since the database is low. System with three convolution layers have been implemented initially. The results are compared. With three convolution and pooling layers 78.57% accuracy is obtained whereas with four layers of convolution and pooling layers 85.71% accuracy is obtained. This model gives the platform for future research in the dental disease's detection using deep learning approaches using multiclass classification.