Randomized trial demonstrates AI's effectiveness in predicting surgical phases in SILC, indicating improved training and safety.
Key Points
This study aims to develop an automated system for recognizing and predicting surgical phases in single-incision laparoscopic cholecystectomy using artificial intelligence.
Developed a deep learning model based on 122 labeled surgical videos from 148 total videos collected across two medical centers.
Tested model performance on 26 additional videos by comparing with surgeon-annotated ground truth.
Measured performance using accuracy, precision, recall, Jaccard Index, and mean absolute error for phase prediction.
Trans-SVNet model achieved overall accuracy of 0.933, precision of 0.939, and recall of 0.939 in classifying surgical phases.
In terms of phase transition prediction, achieved an overall inMAE of 37 seconds, eMAE of 34 seconds, and pMAE of 50 seconds.
Increasing training data significantly improved model performance.