Attention/Distraction Estimation for Surgeons during Laparoscopic Cholecystectomies

Due to the large cognitive strain put upon surgeons' attention during laparoscopic cholecystectomies, automated interface systems that provide meaningful information and feedback without incurring increased cognitive load are necessary. This work outlines (i) a method based on visual data analysis for estimating a surgeon's visual distraction from the surgical monitor and (ii) an approach for using this information to provide unobtrusive reminders to (novice) surgeons during critical sections of the surgical workflow. A large corpus of sensor data was collected during 38 real-world laparoscopic cholecystectomies, and the algorithms were trained and optimized using this data. The algorithm for detecting visual distraction achieved a recognition rate of 93.8%, and a temporal attention-sensitive algorithm for triggering feedback based on thresholds was successfully devised. Further insights into patterns of attention/distraction of surgeons could be derived by means of qualitative analysis.