Robotic surgery is an innovative minimally invasive technique which allows surgeons to perform complex procedures with assistance of a 3-D view camera and remote mechanical arms. It offers potential benefits of smaller incisions, reduced infection risks, decreased postoperative pain, and shortened patient recovery time over traditional surgery (Mack, 2001). Yet increased technique complexity and the physical separation between the robotic console and the patient can impose high mental workload on surgeons and lead to increased medical errors (Catchpole et al., 2018). Previous studies have measured workload in robotic surgery mainly through subjective tools such as the NASA-TLX (Lee et al., 2014), yet this type of assessment is discrete and has potential bias. With advances in wireless sensors and signal analytics, physiological measures are becoming more feasible in the operating room and can provide objective approaches to continuously monitor surgeons' workload. Eye-tracking metrics have been used as workload measurements in many domains including laparoscopic surgery (Marquart, Cabrall, & de Winter, 2015; Zheng, Jiang, & Atkins, 2015; Zheng et al., 2012). However, few studies have applied eye-tracking metrics in the robotic surgery environment, and it remains unknown if the new environment will impact the eye metrics' capacity. The aim of this study is to examine the feasibility of measuring mental workload in robotic surgery with eye-tracking metrics. The Da Vinci Skill Simulator (dVSS, Intuitive Surgical, Inc. Sunnyvale, CA) was used to simulate 5 tasks that are designed for training: Camera Targeting, Peg Board, Ring & Rail, Suture Sponge and Dots & Needles (Hung et al., 2011). Each task had two levels of difficulty and hence 10 exercises in total. Eight participants with no robotic surgery experience (residents and medical students) were recruited voluntarily to practice the 10 exercises in repetitive training sessions. Sessions were scheduled periodically over two months based on the simulator's availability. Each session was restricted to 45 minutes. Participants completed the NASA-TLX survey after every completion of an exercise. Throughout all exercises, a wearable eye tracker, Tobii Pro Glasses 2.0, (Tobii Technology AB, Danderyd, Sweden) was used to sample pupil diameter and gaze point at 50Hz. Three metrics were derived from the eye-tracking signals: pupil diameter, gaze entropy, and fixation duration. Gaze entropy is a measure of the randomness in gaze pattern (Di Nocera, Camilli, & Ter-enzi, 2007) and fixation duration is the percentage of time spent in fixation during a task. Over the study period, 156 cases of exercises have been collected. Mixed effect models were used to test how task difficulty impact on eye tracking metrics with participants treated as a random effect. Increasing difficulty was observed to significantly affect pupil diameter for all tasks (p < .05). The positive coefficients suggested that average pupil diameter in level 2 was larger than that in level 1 and the Cohen's d were very large for 4 of the tasks (Table 1). For gaze entropy, significant effect of difficulty level was observed in task Camera Targeting, Ring & Rail, and Suture Sponge; Cohen's d were large for these tasks. The positive coefficients suggested that participants' gaze patterns became larger in higher level of difficulty (Table 1). Pearson's correlation was used to test the relationship between eye metrics and NASA-TLX ratings, and gaze entropy showed a significant correlation (ρ = .34,p < .001). For both tests, fixation duration did not reveal a significant result. The findings generally support the use of eye-tracking metrics as measurement for mental workload in robotic surgery. Yet pupil diameter was not correlated with subjective ratings as it did in other studies, suggesting the impact of different task environments. Eye-tracking techniques can help identify high workload steps during robotic surgery training and provide feedback for live surgery. Future research should apply eye-tracking measurements in live robotic surgery to validate the results.
展开▼