Automatic Registration of Anatomical Structures of Stereo-Endoscopic Point Clouds

Abstract In image-guided surgery, imaging systems such as (stereo-)endoscopes allow intra-operative 3D reconstructions in form of point clouds. However, endoscopes provide only a narrow field view, resulting confined cloud. this paper, we present an analysis and registration pipeline for clouds acquired by stereo into fused representation. For coarse registration, TEASER is applied, while refinement conducted utilizing point-to-plane ICP. The tested on two datasets: head phantom using EinsteinVision ® 3.0 endoscope from the Stereo Correspondence And Reconstruction Endoscopic Data challenge. results both datasets show that anatomical structures stereo-endoscopes cloud promising contactless radiation-free method. non-rigid deformations are not yet incorporated evaluation method compared to reference data challenging.