1. We proposed a new self-supervised learning framework named SNU-Net. SUN-Net uses contrastive learning to generate supervision signals for the network to learn CT images feature. 2. By extensive comparative experiments, SNU-Net shows the effectiveness of benchmark PTX segmentation tasks.
Slides
Slides
SNU-Net : A Self-Supervised Deep Learning Method for Pneumothorax Segmentation on Chest CT(application/pdf)