Medical Imaging with Deep Learning 2019 MIDL 2019 – Extended Abstract Track Odontogenic cysts and tumors detection in panoramic radiographs using Deep Convolutional Neural Network(DCNN) Tae-Hoon Yong 1 louisyong9512@gmail.com Sang-Jeong Lee 2 sjlee89@snu.ac.kr Won-Jin Yi 3 wjyi@snu.ac.kr 1 Department of Computer Engineering, School of Engineering, Hongik University, Seoul, Korea 2 Department of Biomedical Radiation Sciences, Graduate School of Convergence Science and Tech- nology, Seoul National University, Seoul, Korea 3 Seoul National University Dental hospital, Department of Oral and Maxillofacial Radiology, Seoul, South Korea Abstract Diseases that require surgery, such as cysts or tumors that occur in the oral maxillofacial region, have been often missed or misdiagnosed despite the importance of early detection. Computer-assisted diagnostics using a deep convolution neural network (DCNN), a machine learning technology based on artifcial neural networks, can provide more accurate and faster results. In this study, we will investigate a method for automatically detecting fve diseases that frequently occur in the oral maxillofacial region using DCNN in panoramic radiographs. Keywords: Object Detection, Region Proposal Networks, Dental panoramic images 1. Introduction A lot of studies are currently being conducted on the diagnosis using deep learning, and it is actively used in the medical feld. In particular, the technique of predicting the region of a disease in a radiological image conduct a great help to a radiologist. Among other things, the deep learning technology to detect objects with CNN (Krizhevsky et al., 2012) backbone such as Region proposals with Convolutional Neural Network (RCNN) (Girshick et al., 2014), Fast RCNN (Girshick, 2015) and Faster RCNN (Ren et al., 2015) is a major innovation in the existing medical system. Recently, a new technology has been announced that has improved performance called YOLO (Redmon et al., 2016). YOLO has the advan- tage of faster image analysis than Faster RCNN. In this paper, we will perform detecting cyst and tumor diseases using YOLO-V3 networks and compare the results. © 2019 T.-H. Yong, S.-J.L. & W.-J.Y. .