Deeplab v3+ Based Automatic Diagnosis Model for Dental X-ray: Preliminary Study

Abstract

Recently, deep learning (DL) based semantic segmentation approach has been widely applied in medical image analysis. The semantic segmentation approach based DL technique was employed in the diagnosis of dental conditions with digital panoramic radiography (DRP). The purpose of this study is to investigate the accuracy of the semantic segmentation of Deeplab v3+ in the diagnosis of 5 different dental disease - apical, abrasion, caries, impaction, perio. DPR database (512×748-pixel, including 86 panoramic radiography) was used for semantic segmentation (DeepLab v3+). To validate the performance, the confusion matrix (maximum 97 %) was estimated. In addition, significant classification and semantic segmentation results were assessed. From the result of this study, the DL model could be a useful tool for the dentist to identify dental diseases as a clinical aid software.