Artificial intelligence-based automated laparoscopic cholecystectomy surgical phase recognition and analysis

Area of Science:

• Medical technology
• Artificial intelligence in surgery
• Computer vision

Background:

• Laparoscopic surgical video analysis is advanced by AI and computer vision.
• Lack of multi-center studies on deep learning for laparoscopic cholecystectomy phase recognition.
• This study applies AI to recognize and analyze phases in multi-center laparoscopic cholecystectomy videos.

Purpose of the Study:

• To develop and validate a deep learning model for accurate phase recognition in laparoscopic cholecystectomy.
• To assess the model's performance on multi-center data.
• To explore the potential of AI in analyzing surgical video data.

Main Methods:

• An observational cohort study included 163 laparoscopic cholecystectomy videos from four centers.
• A deep learning model was trained on 90 videos and validated on 10 videos against surgeon annotations.
• Model performance was evaluated using precision, recall, F1 score, and overall accuracy.

Main Results:

• Surgeon annotations showed a mean concordance correlation coefficient of 92.38%.
• The deep learning model achieved an overall phase recognition accuracy of 91.05%.
• Analysis of 63 videos revealed significant individual variance in surgical phase durations, with acute cholecystitis cases showing prolonged durations.

Conclusions:

• A multi-center deep learning model can accurately identify laparoscopic cholecystectomy phases.
• AI demonstrates potential for large-scale surgical data analysis with future clinical applications.
• Continued refinement of AI can enhance its utility in surgical practice.