Multi-modal monocular endoscopic depth and pose estimation with edge-guided self-supervision | Synapse