SSA-Net: Spatial self-attention network for COVID-19 pneumonia infection segmentation with semi-supervised few-shot learning

•SSA is presented for COVID-19 lesion segmentation from CT images.•Self-attention learning is utilized to obtain more contextual information.•Spatial convolution is used to accelerate the training convergence.•A semi-supervised few-shot iterative method is proposed for multi-class segmentation.•The iterative method is based on a re-weighting module and a trust module.