Akademik Veri Yönetim Sistemi
ORAL SURGERY, ORAL MEDICINE, ORAL PATHOLOGY, ORAL RADIOLOGY, AND ENDODONTICS, cilt.1, ss.1-26, 2023 (SCI-Expanded)
In this study, we aim to compare two different deep learning method of Faster R-CNN and YOLO-V4 for the classification of teeth in dental panoramic radiography.
Our method is based on deep learning models trained on a semantic segmentation task. This study was carried out on 1200 panoramic radiographs selected by retrospectively. In the classification process, a total of 36 classes were determined, including 32 teeth and 4 impacted teeth. The test data were not introduced to the systems during the training phase of the methods.
The YOLO-V4 method has an 0.9990 of average precision, 0.9918 of sensitivity, and 0.9954 of F1 score and Faster R-CNN method has 0.9367 of average precision, 0.9079 of sensitivity, and 0.9221 of F1 score. The experimental evaluations showed that the YOLO-V4 method outperformed the Faster R-CNN method in terms of the accuracy of the predicted tooth in the tooth classification process. In addition, the YOLO-V4 method was 4 times faster than the Faster R-CNN method for classifying teeth.
Misinterpretation may be inevitable during dental radiographic analysis. Intelligent systems can assist dentists in clinical decision making, save time, and reduce the negative effects of stress and fatigue in daily practice.