Akademik Veri Yönetim Sistemi
Comptes Rendus de L'Academie Bulgare des Sciences, cilt.74, sa.2, ss.269-277, 2021 (SCI-Expanded)
© 2021 Academic Publishing House. All rights reserved.The aim of this study was to detect the impacted teeth in panoramic radiography images by Object Learning (Faster R-CNN) technique, which is one of the Deep Learning (DL) techniques and Inception v2, ResNet-50, ResNet-101 transfer learning architectures. In order to obtain the appropriate data from the image, Faster R-CNN algorithm and Inception v2, ResNet-50, ResNet-101 transfer learning methods were used. Model performance was measured with Confusion Matrix, F1-Score, Sensitivity (Recall) and Precision values. It was observed that detection performances of the three methods used in the impacted teeth detection in panoramic radiography images in the test data set are close to each other and were over 95%. Using Faster R-CNN – Inception v2 architecture, 121 of 45 panoramic radiography images in the test dataset were correctly identified, yielding a success rate of approximately 96% (F1-Score). As a result of this study, the high rate of success in detecting affected teeth in panoramic radiographic images of adult patients suggests that artificial intelligence may be an aid tool for dentists. As a result, we believe that this would make significant contributions to the development of artificial intelligence in the field of healthcare.