MAFNet: A deep multi-scale attentive fusion network for virtual osteotomy of maxillofacial bones in CT images containing metal artifacts

Su Yang, Ji-Yong Yoo, Sang-Jeong Lee, Se-Ryong Kang, Jun-Min Kim, Jo-Eun Kim, Kyung-Hoe Huh, Sam-Sun Lee, Min-Suk Heo, Hoon Joo Yang, Won-Jin Yi

Abstract: Highlights•MAFNet predicts the mandible, maxilla, and skull regions for the Le Fort I osteotomy.•Multi-scale encoders are proposed to capture multi-scale context features.•Atrous feature fusion module is proposed to enlarge the receptive field of layers.•Multi-scale spatial attention module is used to focus on the salient regions.•MAFNet shows accurate and robust segmentation for the virtual Le Fort I osteotomy.