Development of an artificial neural network/ convoluted neural network and deep learning based predictive model of difficulty grades of laparoscopic cholecystectomy

**Background:**Laparoscopic cholecystectomy, a routinely performed high volume surgical procedure, is a suitable surgical procedure for the application of machine learning techniques, using analysis of ultrasound images. However, the adoption of AI in surgical practice has been limited by factors such as surgeon awareness, complex algorithms, and the lack of synchronized preoperative ultrasound and intra-operative images.

 **The Study:**This pilot study aims to investigate the role of artificial intelligence (AI) in predicting difficulty grades for Laparoscopic Cholecystectomy whereby it is  hypothesized that AI-based analysis of preoperative ultrasound images can predict operative difficulty by training a computer system with both preoperative and intra-operative images.

 Aims & Objectives: The research objectives include collecting and feeding data of ultrasound images to develop an AI algorithm in the first phase, and testing the algorithm by predicting difficulty grades based on preoperative ultrasound images in the second phase. Being conducted at a tertiary care teaching hospital, approximately 400 cases of laparoscopic cholecystectomy will be analyzed.

 Methodology: Involves creating a data pool of preoperative and intra-operative images, training a machine learning platform using artificial/ convolutional neural networks, and subsequently assessing the accuracy of preoperative ultrasound findings in predicting operative difficulty.

 It is expected to establish the potential role of AI-based image analysis in laparoscopic cholecystectomy and pave the way for future research in this field. The findings will contribute to evidence-based medicine and have implications for clinical practice, providing a foundation for further exploration and development of AI tools in surgical prediction.