Home /Research /A Standardized Temporal Segmentation Framework and Annotation Resource Library in Robotic Surgery
SURGICAL

A Standardized Temporal Segmentation Framework and Annotation Resource Library in Robotic Surgery

Busisiwe Mlambo, Mallory Shields, Simon P. Bach, Armin Bauer, Andrew J. Hung, Omar Yusef Kudsi, Felix Neis, John F. Lazar, Daniel Oh, Robert X. Perez, Seth A. Rosen, Naeem Soomro, Michael Stany, Christian von Wagner, Lilia Purvis, Benjamin Mueller, Sadia Yousaf, Casey Troxler, Alfred Song, Emily Summers

Year
2025
Citations
7
Access
Open access

Abstract

Objective: To develop and share the first clinical temporal annotation guide library for 10 robotic procedures accompanied with a standardized ontology framework for surgical video annotation. Patients and Methods: A standardized temporal annotation framework of surgical videos paired with consistent, procedure-specific annotation guides is critical to enable comparisons of surgical insights and facilitate large-scale insights for exceptional surgical practice. Existing ontologies and guidance not only provide foundational frameworks but also provide limited scalability in clinical settings. Building on these, we developed a temporal annotation framework with nested surgical phases, steps, tasks, and subtasks. Procedure-specific annotation resource guides consistent with this framework that define each surgical segment with formulaic start and stop parameters and surgical objectives were iteratively created across 7 years (January 1, 2018, to January 1, 2025) through global research collaborations with surgeon researchers and industry scientists. Results: We provide the first resource library of annotation guides for 10 common robotic procedures consistent with our proposed temporal annotation framework, enabling consistent annotations for clinicians and large-scale data comparisons with computer-readable examples. These have been used in over 13,000 annotated surgical cases globally, demonstrating reproducibility and broad applicability. Conclusion: This resource library and accompanying ontology framework provide critical structure for standardized temporal segmentation in robotic surgery. This framework has been applied globally in private studies examining surgical objective performance metrics, surgical education, workflow characterization, outcome prediction, algorithms for surgical activity recognition, and more. Adoption of these resources will unify clinical, academic, and industry efforts, ultimately catalyzing transformational advancements in surgical practice.

Keywords

AnnotationSegmentationComputer scienceResource (disambiguation)Robotic surgeryArtificial intelligenceInformation retrievalComputer vision

Related papers

Browse all SURGICAL papers