Fig. 1: An AI system that decodes intraoperative surgical activity from videos.
From: A vision transformer for decoding surgeon activity from surgical videos
a, Surgical videos commonly collected during robotic surgeries are decoded via SAIS into multiple elements of intraoperative surgical activity: what is performed by a surgeon, such as the suturing subphases of needle handling, needle driving and needle withdrawal, and how that activity is executed by a surgeon, such as through discrete gestures and at different levels of skill. b, SAIS is a unified system since the same architecture can be used to independently decode different elements of surgical activity, from subphase recognition to gesture classification and skill assessment.
