Improving surgical motor skill assessment and acquisition via neuromodulation, neuroimaging, and machine learning

Abstract

Secondly, we explore the possibility to emulate the current standardized surgical skill metric employed in the field, namely the FLS score, by combining neuroimaging data acquired during the task execution and machine learning methodologies for potentially fast and bedside implementation. In this context, we have validated a deep neural network, Brain-NET, that accurately predicts performance scores from hemodynamic data from the brain obtained using functional near-infrared spectroscopy (fNIRS). Furthermore, we are also currently implementing deep learning approaches to improve and speed up the fNIRS data preprocessing workflow towards enabling real-time implementation.