Unsupervised identification of surgical robotic actions from small non homogeneous datasets

TL;DR

This paper introduces a fast unsupervised algorithm for identifying surgical actions using minimal, heterogeneous datasets, significantly outperforming existing methods in complex, real-world surgical training scenarios.

Contribution

It presents a novel unsupervised approach that leverages automatically extracted features to identify surgical actions without manual annotations, effective on small and diverse datasets.

Findings

Achieved 58% F1-score on non-expert surgical data, outperforming 24% of previous methods.

Robustly handles noise, short actions, and non-repetitive workflows.

Demonstrates effectiveness with limited and heterogeneous datasets.

Abstract

Robot-assisted surgery is an established clinical practice. The automatic identification of surgical actions is needed for a range of applications, including performance assessment of trainees and surgical process modeling for autonomous execution and monitoring. However, supervised action identification is not feasible, due to the burden of manually annotating recordings of potentially complex and long surgical executions. Moreover, often few example executions of a surgical procedure can be recorded. This paper proposes a novel fast algorithm for unsupervised identification of surgical actions in a standard surgical training task, the ring transfer, executed with da Vinci Research Kit. Exploiting kinematic and semantic visual features automatically extracted from a very limited dataset of executions, we are able to significantly outperform state-of-the-art results on a dataset of non-expert executions (58\% vs. 24\% F1-score), and improve performance in the presence of noise, short actions and non-homogeneous workflows, i.e. non repetitive action sequences.