Objective video-based tremor assessment for movement disorders using open-source software

Abstract

Tremor is an involuntary rhythmic muscle movement assessed subjectively by specialists. To improve accuracy and mitigate bias, tremor must be video recorded and rated by multiple experts. Existing video-based motion tracking techniques can be applied to quantify tremor assessment; though, such methods rely on sophisticated and expensive instrumentation as well as specialized skin markers. This paper describes a low-cost markerless method using accessible hardware and open-source software. In a cohort of 8 subjects with tremor undergoing deep brain stimulation therapy, we show our video-based technique has strong concordance (r = 0.93, p < 0.001) with expert tremor ratings. This makes it suitable for point-of-care assessment as well as use in future structured clinical trials.