Concept Graph Neural Networks for Surgical Video Understanding
This addresses AI-augmented surgery by enhancing interpretation of multiple entities and concepts in surgical videos, though it appears incremental as it builds on existing graph network and temporal analysis methods.
The paper tackles surgical video understanding by integrating conceptual knowledge into temporal analysis using temporal concept graph networks, demonstrating improved recognition and detection on complex benchmarks like critical view of safety verification and Parkland grading scale estimation.
We constantly integrate our knowledge and understanding of the world to enhance our interpretation of what we see. This ability is crucial in application domains which entail reasoning about multiple entities and concepts, such as AI-augmented surgery. In this paper, we propose a novel way of integrating conceptual knowledge into temporal analysis tasks via temporal concept graph networks. In the proposed networks, a global knowledge graph is incorporated into the temporal analysis of surgical instances, learning the meaning of concepts and relations as they apply to the data. We demonstrate our results in surgical video data for tasks such as verification of critical view of safety, as well as estimation of Parkland grading scale. The results show that our method improves the recognition and detection of complex benchmarks as well as enables other analytic applications of interest.