Automatic Teeth Recognition Method from Dental Panoramic Images Using Faster R-CNN and Prior Knowledge Model

Dentists have to check a large number of panoramic X-ray images every day. It takes much time and may cause human-made error. There are some studies in the literature to detect teeth by using deep learning-based techniques automatically. However, due to the limitations of object detection performance, it may provide multiple candidates. This paper proposes an automated teeth recognition method that refines the multiple detected candidates in panoramic X-ray images based on an optimization method with a prior knowledge model. The proposed method detects teeth candidates through faster R-CNN, and then determines the appropriate candidates by optimizing an objective function. The objective function evaluates the teeth's relative position concerning other teeth based on the prior knowledge model. Use 1000 images, the proposed method's accuracy was 0.94, and was higher than 0.92 with a method without optimization method. The experimental results verified that the proposed method could perform better than the stand-alone faster R-CNN technique.