LapSeg3D: Weakly Supervised Semantic Segmentation of Point Clouds Representing Laparoscopic Scenes
This addresses the problem of automating surgical tasks for medical professionals, but it is incremental as it builds on existing DNN and clustering methods for a specific domain.
The paper tackled semantic segmentation of laparoscopic surgical scenes for task automation by proposing LapSeg3D, a DNN-based method that achieved an F1 score of 0.94 for gallbladder segmentation on ex-vivo porcine liver datasets.
The semantic segmentation of surgical scenes is a prerequisite for task automation in robot assisted interventions. We propose LapSeg3D, a novel DNN-based approach for the voxel-wise annotation of point clouds representing surgical scenes. As the manual annotation of training data is highly time consuming, we introduce a semi-autonomous clustering-based pipeline for the annotation of the gallbladder, which is used to generate segmented labels for the DNN. When evaluated against manually annotated data, LapSeg3D achieves an F1 score of 0.94 for gallbladder segmentation on various datasets of ex-vivo porcine livers. We show LapSeg3D to generalize accurately across different gallbladders and datasets recorded with different RGB-D camera systems.