Abstract Introduction Robotic-assisted surgery offers ergonomic and clinical advantages over traditional techniques but demands rigorous skill acquisition. Current assessments rely on subjective assessments or system-dependent objective metrics. This study aims to develop an objective, reproducible, and system-agnostic method for assessing robotic surgical performance using hand motion data. Methods We recruited participants, novices (2 h robotic training) and intermediates (20 h), to perform three tasks (dissection, running and interrupted suture) on a phantom model. A ZED depth camera was fixed on the robotic console to capture trainees' hands and AprilTags tracked hand movements on both the DaVinci and HUGO. We used MUTUAL to synchronously capture endoscopic and hand views. 3D positions of both hands were extracted over time in Python. For motion analysis, we used path length distance for efficiency, mean absolute relative phase for bimanual coordination, and convex hull volumes for hand workspace. Results Our system-agnostic pipeline enabled data collection from the HUGO and DaVinci. Task-dependent variations were shown in bimanual coordination through differences seen in hand motion distances and workspace volumes. Temporal relationship was evaluated using time-series of distance and volume for each hand, providing insight into bimanual coordination required in varying skills. Fluctuations between repetitions indicated the need for additional sessions to better define the learning curve. Self-assessed ratings in post-session questionnaires improved for all skill domains (instrument handling, hand movements, and suturing) after repetitions. Conclusions This study developed an objective and efficient methodology for assessing robotic performance using hand motion data to support surgical training and intra-operative performance evaluation.
Liu et al. (Sun,) studied this question.