SCL-SOD: A hybrid self-supervised contrastive learning framework for salient object detection | Synapse