Towards Surgical Skill Modeling in Cardiac Ablation Using Deep Learning

The most widely practised method for treating cardiovascular problems is minimally invasive surgery (MIS). The rising interest in surgical robots and simulations has led to a greater demand for more objective methods of skill evaluation. Traditionally, the performance of novices is evaluated using surgeons’ skills through a specific and streamlined ablation task. To this end, an experimental setup was proposed to provide a simulated ablation procedure through a mechanical system. It is equipped with the synthetic heartbeat mechanism of the heart with the capability of measuring the contact forces between a catheter’s tip and a force sensor. Using a commercially available catheter for ablation, the task was to maintain the force within a safe range while the tip of the catheter is touching the surface of the sensor. Accomplishing multiple experiments by novices and experts, a deep recurrent neural network was considered to extract the model of skills by solving a binary classification problem. The results of the trained model showed that the proposed pipeline was able to properly distinguish the novices’ from experts’ maneuvers with 95% accuracy.