Pith. sign in

REVIEW

Dynamic Scene Graph Representation for Surgical Video

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2309.14538 v2 pith:QM67PMM5 submitted 2023-09-25 cs.CV

Dynamic Scene Graph Representation for Surgical Video

classification cs.CV
keywords surgicalscenegraphgraphsvideosworkflowanatomicalcrucial
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Surgical videos captured from microscopic or endoscopic imaging devices are rich but complex sources of information, depicting different tools and anatomical structures utilized during an extended amount of time. Despite containing crucial workflow information and being commonly recorded in many procedures, usage of surgical videos for automated surgical workflow understanding is still limited. In this work, we exploit scene graphs as a more holistic, semantically meaningful and human-readable way to represent surgical videos while encoding all anatomical structures, tools, and their interactions. To properly evaluate the impact of our solutions, we create a scene graph dataset from semantic segmentations from the CaDIS and CATARACTS datasets. We demonstrate that scene graphs can be leveraged through the use of graph convolutional networks (GCNs) to tackle surgical downstream tasks such as surgical workflow recognition with competitive performance. Moreover, we demonstrate the benefits of surgical scene graphs regarding the explainability and robustness of model decisions, which are crucial in the clinical setting.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.