Background: Experts consistently exhibit more efficient gaze behaviors than non-experts during motor tasks. In surgery, experts have been shown to gaze more at surgical targets than surgical tools during simple simulations and when watching surgical recordings, suggesting a proactive control strategy with greater use of feedforward visual sampling. To investigate such expert gaze behaviors in a more dynamic and complex laparoscopic surgery simulation, the current study measured and compared gaze patterns between surgeons and novices who practiced extensively with laparoscopic simulation. Methods: Three surgeons were assessed in a testing visit and five novices were trained and assessed (at pre-, mid-, and post-training points) in a 5-visit protocol on the Fundamentals of Laparoscopic Surgery peg transfer task. The task was adjusted to have a fixed action sequence to allow recordings of dwell durations based on pre-defined areas of interest (AOIs). Individualized learning curves of novices were analyzed using an inverse function model, and group-level differences were tested using analysis of variance on both behavioral performance and dwell duration measures. Results: Trained novices were shown to reach more than 98% (M = 98.62%, SD = 1.06%) of their behavioral learning plateaus, leading to equivalent behavioral performance to that of surgeons. Despite this equivalence in behavioral performance, surgeons continued to show significantly shorter dwell durations at visual targets of current actions and longer dwell durations at future steps in the action sequence than trained novices (ps < .03, Cohens ds > 2). Conclusion: This study demonstrates that, whereas novices can train to match surgeons on behavioral performance, their gaze pattern is still less efficient than that of surgeons, suggesting that eye-tracking metrics might be more sensitive than behavioral performance in detecting surgical expertise. Such insight can be applied to develop training protocols so non-experts can internalize expert gaze templates to accelerate learning.
bioRxiv Subject Collection: Neuroscience