Statistical shape models (SSM) describe complex shape variations derived from a training population in a compact way. Thus, they are well suited for robust reconstruction of unknown shapes, especially insituations where data is affected by noise, artefacts, or only partially contains the unknown shape. Important applications are, e.g., image segmentation as well as reconstructive surgery. The aim of this work is to provide an automatic method to reconstruct and complete partial liver surfaces.
This work won the Shape Challenge Best Prize, sponsored by sponsored by Siemens.