Automated and Adaptive Surgical Robotic Training Using Real-Time Style Feedback Through Haptic Cues

Evaluating robotic surgical skills efficiently and effectively while providing surgical residents with intuitive and easy to understand feedback is an important problem. This dissertation focuses on the design, implementation, and experimental validation of a style-based surgical skill assessment haptic guidance and training method that addresses key challenges related to surgical training systems. The first topic of this dissertation is finding differences in the style of movement of an expert versus a novice while performing a task. We leverage crowd-sourced assessment to define the stylistic attributes of surgical motion and find features within movement and physiological data to classify style. We show through an experimental human subject study that stylistic behaviors can distinguish between different levels of expertise and have more discriminating power than standard scoring systems for surgical robot simulators. The second topic of this dissertation is to automate the stylistic behavior detection in near-real-time to be able to provide feedback to the user during task performance. For this purpose, a data driven model was developed and tested for each proposed stylistic behavior adjective. These models enable the detection of a deficiency as soon as it occurs. The final step in a training system is to provide meaningful feedback to the user. Since each individual learns in a different way, it is important to customize training for each individual. The third topic of this dissertation is to provide the trainees with adaptive feedback based on their stylistic performance. For this purpose, the deficiency of a user’s stylistic behavior is detected and is used to provide an appropriate force feedback to help correct movement. Three types of force feedback for each stylistic behavioral adjective are evaluated in this study and the best feedback for each style is found through an experimental human subject study. The work in this dissertation provides a groundwork for adaptive, automatic and real-time surgical skill training. This method can also be extended for coaching in other areas other than surgical applications, such as sports.