Surgery is a challenging domain for motor skill acquisition. A critical contributing factor in this difficulty is that feedback is often delayed from performance and qualitative in nature. Collection of highdensity motion information may offer a solution. Metrics derived from this motion capture, in particular indices of movement smoothness, have been shown to correlate with task outcomes in multiple domains, including endovascular surgery. The open question is whether providing feedback based on these metrics can be used to accelerate learning. In pursuit of that goal, we examined the relationship between a motion metric that is computationally simple to compute-spectral arc length-and performance on a simple but challenging motor task, mirror tracing. We were able to replicate previous results showing that movement smoothness measures are linked to overall performance, and now have performance thresholds to use in subsequent work on using these metrics for training.
INTRODUCTION BackgroundTraining in many motor domains can be a challenge. Consider learning to do some motor task with a well-defined "success" vs. "failure" outcome metric, such as shooting a free throw in basketball. A trainee can shoot repeatedly and easily determine for each shot whether the shot was successful, but it may be very difficult for the trainee, and even a coach, to determine why different shots resulted in success or failure. In a domain like surgery, this is even more of a challenge, as "success" vs. "failure" may not be known for weeks or even months after a procedure is performed. Furthermore, the "coach" in surgery is typically one or more other surgeons, whose expertise is valuable and time spent training is time taken away from other activities.Surgical skill is notoriously difficult to assess and evaluate (Moorthy, Munz, Sarker, & Darzi, 2003). The development of effective metrics to evaluate surgical skill is an active area of research (Reiley, Lin, Yuh, & Hager, 2011). The need for objective and quantitative assessment tools has been a topic of considerable interest and importance (Lin, Shafran, Yuh, & Hager, 2006;van Hove, Tuijthof, Verdaasdonk, Stassen, & Dankelman, 2010;Tsue, Dugan, & Burkey, 2007). Such need is driven by evidence that skill level can affect clinical outcomes after surgery (Reznick & MacRae, 2006). Assessment is often done informally through subjective feedback from other surgeons (Chaer et al., 2006;Bech et al., 2011;Riga et al., 2011) or based on a simple count of the number of times a procedure has been performed (Cronenwett, 2006;Schanzer et al., 2009).Collection of high-density motion information, if the relevant metrics can be linked successfully with task outcomes, offers the opportunity to provide more detailed feedback that may speed the learning process. A quantitative and motion-based approach to performance assessment in manual control tasks is starting to gain traction in the research community, especially in the domain of robotic surgery and the corresponding simulation environments that a...