Semi-Supervised Method for Determining the Maxillary and Mandibular Boundaries on Panoramic Radiographs

Abstract

X-ray imaging plays an important role in the detection and diagnosis of dental disorders that can not be detected with the eye. Examination, evaluation and accurate diagnosis of panoramic radiography images used in anomaly detection is a process requiring serious experience and expertise. In order to simplify this process and to increase accuracy, some studies have been carried out on both the imaging techniques and the development of the computer aided diagnosis systems by transferring these images to the digital medium. One of these studies is segmentation of various structures in dental X-ray images. In this study, instead of using random masks as initial inputs, we developed a special mask identification method for each image and gave the mask which is the output of this method as initial input to the segmentation algorithm to obtain more suitable elliptical forms closer to the actual(ground truth) results and to the maxillary and mandibular structures.