An efficient deep neural network framework for COVID-19 lung infection segmentation

Ge Jin, Chuancai Liu, Xu Chen

Abstract: Highlights•A novel network for Coronavirus Disease infected area segmentation from Computed Tomography slices is presented.•The proportion loss is proposed to mitigate the class imbalance, but without losing generality.•A novel semi-supervised framework based on adversarial learning is proposed.•The comprehensive experiments is experimented to demonstrate that the method effectively.