Deep learning-based localization of the biliary tract in laparoscopic images acquired during surgical robotic procedures

Laparoscopic cholecystectomy is a minimally inva- sive procedure whereby the gallbladder is removed using la- paroscopic techniques. With more than 500,000 cholecystectomies performed per year, great interest has developed in laparoscopic cholecystectomy. The major advantages of laparoscopic chole- cystectomy with respect to traditional cholecystectomies are the short hospital stay and early return to normal activity. Morbidity is low, but there is a concern about bile duct injuries. This study proposes a deep learning-based algorithm for localization of the biliary tract during laparoscopic surgical procedures. A dataset has been collected, consisting of videos of standard surgical practice, and has been used to train the deep learning algorithm.