SSM-Net: Semi-supervised multi-task network for joint lesion segmentation and classification from pancreatic EUS images

Jiajia Li, Pingping Zhang, Xia Yang, Lei Zhu, Teng Wang, Ping Zhang, Ruhan Liu, Bin Sheng, Kaixuan Wang

Abstract: Highlights•A semi-supervised network is designed to classify and segment pancreatic mass.•Saliency-aware contrastive loss is introduce to capture unlabeled data features.•Attention modules are proposed to classify and segment pancreatic lesions.•A large EUS-based pancreas image dataset is collected, containing about 500 patients.