Development of an Autonomous Platform for Surgical Simulation and Training

Simulation based training (SBT) has allowed for improvement of clinical skills by providing practical experience of performing surgical procedures without the risk of harming patients. Performance assessment of learners during SBT has proved to be effective in improving psycho-motor skills, encouraging active learning and in-turn increasing patient safety. However, training sessions require experienced physician instructors to conduct one-on-one training and assessment. Consequently, these sessions are long, subjective, difficult and expensive due to which they can’t be held regularly. This research addresses the problem through an autonomous training simulator that can provide objective, accurate and real-time measure of learners’ skill through deep neural networks. The simulator is a combination of a realistic manikin and a virtual platform that records and interprets surgical information collected from the physical actions performed on the manikin. As a proof of concept, the simulator was used to train for neonatal thoracentesis and pericardiocentesis, caused due to pneumothorax, pleural or pericardial effusion which have incidence rates of up to 2% when diagnosed. The simulator was successful at providing useful guidance and an accurate assessment of surgical performance during the trials and received an overwhelmingly positive feedback from the trainees and expert surgeons.