Artificial Intelligence and Medical Image Processing for Early Detection of Gum Recession

Abstract

Periodontal disease is one of the most common oral diseases worldwide and can lead to irreversible tooth-supporting tissue destruction if not diagnosed at an early stage. Accurate grading of periodontal disease from panoramic dental radiographs (OPGs) is essential for appropriate treatment planning; however, manual assessment is time-consuming and subject to inter-observer variability. This study proposes a deep learning-based framework for the automatic classification of periodontal disease into Grade A, Grade B, and Grade C using OPG images. A dataset of 250 anonymized panoramic radiographs was collected and labeled according to the 2018 AAP/EFP classification by experienced periodontists. To improve model generalization, data augmentation techniques expanded the dataset to 2,100 images. Four deep learning models, namely ResNet50, InceptionV3, Xception, and the proposed ShuffleNet–ResNet50 hybrid model, were trained and evaluated using 10-fold cross-validation. Performance was assessed using multiple evaluation metrics, including accuracy, sensitivity, specificity, precision, F1-score, and receiver operating characteristic (ROC) analysis. Experimental results demonstrated that all evaluated models achieved high classification performance. Among them, ResNet50 produced the best overall results, achieving an accuracy of 96.76%, sensitivity of 95.15%, specificity of 97.60%, and an F1-score of 95.08%. Confusion matrix and ROC analyses further indicated that Grades A and C were classified with higher reliability than Grade B, which exhibited greater overlap with adjacent disease stages. The findings demonstrate the feasibility of deep learning models for automated classification of periodontitis Grades A, B, and C from panoramic dental radiographs.