Endoscopic Reconstruction

Endoscopic reconstruction aims to create accurate 3D models of internal organs and tissues from endoscopic images, primarily to improve surgical planning and execution. Current research heavily utilizes Gaussian splatting-based methods and neural radiance fields (NeRFs), focusing on efficient real-time rendering, accurate handling of dynamic scenes (including deformable tissues), and robust performance even with limited input views. These advancements are significantly impacting minimally invasive surgery by providing surgeons with enhanced visualization and potentially improving surgical outcomes and training.