Artificial intelligence based detection of diseases and anatomical structures related to the oral region

Recent advancements in the fields of computer software and hardware technology have paved the way for the widespread integration of artificial intelligence and deep learning methodologies across diverse disciplines. These developments have resulted in significant advancements, particularly in domains such as image processing, object detection and segmentation. The present thesis focuses on the utilization of artificial intelligence to facilitate the segmentation of diseases and anatomical structures related to the oral region. In the contemporary context, dental health is acknowledged as a pivotal component of human health, with panoramic radiography images being extensively utilized within the domain of dental practice. However the majority of studies in the literature analyze images by cropping method to detect dental diseases and structures, in this study, segmentation operations were performed on the whole mouth region without any preprocessing on panoramic X-ray images. In the course of the study, ranges of artificial intelligence-based methodologies were employed for the identification of impacted teeth, decayed teeth and dental infections. In addition, as there is a need for a dataset to apply segmentation using panoramic radiography in the literature, a new dataset was created and annotated with expert dentists. This thesis aims to fill the gaps in the literature and presents a new dataset and optimal segmentation methods for artificial intelligence-based segmentation of panoramic radiography images. The accuracy and efficiency of the applied methods were analyzed in comparison with other current models. The results obtained have shown that the applied methods are successful and have the potential to contribute to diagnostic and treatment processes in the field of dentistry.