Accurate Periodontal Disease Classification from Dental Radiographs Using Deep Learning Models

Abstract

Dental radiography is useful for clinical diagnosis, treatment, and quality assessment. Much effort has gone into developing digitalized dental X-ray image analysis systems to improve clinical quality. The preprocessing of the dataset, procedures, and result evaluation of dental treatment performed using periapical X-ray images taken before and after the operation. I propose a tool pipeline for automated clinical quality evaluation to assist dentists in making clinical decisions. I use Deep Learning technique to detect the disease from the X-Ray images. The dataset contains 525 dental X-Ray images. X-Ray images are labelled as Normal and Diseased by designated dental experts. This research explores the application of deep learning models for dental disease detection, focusing on two advanced architectures: ResNet101 and ResNet152. The study involves training and evaluating these models on a curated dental dataset to assess their performance in classifying dental images. ResNet101, configured with a batch size of 256 and a learning rate of 0.001, achieved a training loss of 0.002 and a test loss of 0.015. The model demonstrated perfect training accuracy of 100% and a commendable test accuracy of 98.35%, indicating strong learning and generalization capabilities. In comparison, ResNet152, with a batch size of 64 and a higher learning rate of 0.01, exhibited a training loss of 0.02 and an exceptionally low-test loss of 0.001. The model achieved a training accuracy of 100% and a test accuracy of 99.15%, showcasing superior generalization to unseen data. The results highlight the effectiveness of both models in dental disease detection, with ResNet152 showing a marginally better performance on test data.