Eye Tracking As a Tool to Evaluate Colonoscopy Skill

Background and Object: Colonoscopy requires a high level of both motor skills and spatial cognitive ability. However, few studies have focused on the cognitive factors of colonoscopy skill. This feasibility study investigated whether measurement of eye tracking during colonoscopy is an effective means of objectively evaluating colonoscopy skill level.
Methods and Procedures: Observation and manipulation tasks involving a colonoscope were prepared in order to determine gaze point during naked-eye colonoscopy. For all experimental tasks we used a Fujinon EVE 250 LPS colonoscope and a colon simulator made of silicon rubber (LM-044B, Koken, Japan). The study subject was a surgeon with 25 years’ experience of performing colonoscopies. The colonoscopic image was displayed on a 32-inch LCD monitor. Eye movements and gaze points were determined during insertion of the colonoscope from rectum to cecum. Gaze points were measured by the Eyemark Recorder EMR-8B (NAC Imaging Technology, Japan). This system detects the movement of both eyes at 30 Hz using two infrared cameras. The gaze points are displayed and recorded on the image of the actual visual field obtained from a separately equipped color camera. Obtained data were analyzed by EMR-d Factory software (NAC Imaging Technology). Gaze points toward the center axis of the colonic lumen (i.e. the direction in which the colonoscope should be inserted) were classified as being within the “target” area and other gaze points in the colonoscopic view were classified as being within the “peripheral” area. Gaze points outside the display or failure to detect gaze points were classified as “other”.
Results: Eye tracking analysis revealed that during the insertion task from rectum to cecum, which took 299 seconds, the gaze point was located in the “target” area for 151 seconds, in the ”peripheral” area of the colon for 80 seconds, and in “other” areas for 68 seconds. Mean gaze point residence time for the “target” area was 0.52 seconds and 0.28 seconds for the “peripheral” area. The subject rarely looked at the controlling portion of the colonoscope or the external view of the simulator.
Discussion and Conclusion: The subject spent 50% of the insertion time looking at the “target” area with relatively limited eye movements. He spent 27% of the time looking at the “peripheral” area with frequent and rapid eye movements in the horizontal direction as he scanned the area to determine the direction of insertion. The majority of the remaining time consisted of eye tracking errors such as blinking. These time ratios are expected to differ from those of novice subjects. The results of this preliminary study indicate that eye tracking measurement is a feasible tool for evaluating the skill level of colonoscopists.

Session: Poster

Program Number: P188

View Poster

« Return to SAGES 2010 abstract archive