Poster: A Pilot Study On Camera-based Neurological Deficit Detection

The BANDIT - Brain Attack Neurological Deficit Identification Tool - project aims at developing an automatic tool to quantitatively assess stroke-related neurological deficits such as facial weakness and limb drift. In this paper, we first introduce the BANDIT project by describing the main framework and then present a patient video data acquisition protocol that was conducted in a real-world hospital setting. We also discuss the inherent bias and variations within our dataset that may create challenges for the algorithm design. The experiences and lessons gained from our study could be beneficial for other researchers who conduct camera-based research in the connected health area.