The role of deep learning for periapical lesion detection on panoramic radiographs.

Researchers

Journal

Modalities

Models

Abstract

This work aimed to detect automatically periapical lesion on panoramic radiographs (PRs) using deep learning.454 objects in 357 PRs were anonymized and manually labeled. They are then pre-processed to improve image quality and enhancement purposes. The data was randomly assigned into the training, validation, and test folders with ratios of 0.8, 0.1, and 0.1 respectively. The state of art ten different deep learning based detection frameworks including various backbones were applied to periapical lesion detection problem. Model performances were evaluated by mean average precision (mAP), accuracy, precision, recall, F1 score, precision-recall curves, area under curve (AUC) and several other Common Objects in Context (COCO) detection evaluation metrics.Deep learning based detection frameworks were generally successful in detecting periapical lesions on panoramic radiographs. Detection performance, mAP, varied between 0.832 and 0.953 while accuracy was between 0.673 and 0.812 for all models. F1 score was between 0.8 and 0.895. RetinaNet performed the best detection performance, similarly ATSS provided F1 score of 0.895 as highest value. Testing with external data supported our findings.This work showed that deep learning models can reliably detect periapical lesions on panoramic radiographs. Artificial intelligence based on deep learning tools are revolutionizing dental healthcare and can help both clinicians and dental healthcare system.