Computer vision of smartphone video has potential to detect functional tremor

Functional neurological disorders (FND) constitute more than 15% of referrals to neurology clinics [ [1] ], and functional tremor is the most common functional movement disorder [ [2] ]. Physical features of a functional tremor include: tremor present at rest, posture and action; variability in frequency and direction; and reduction or abolition of tremor with distraction [ [2] ]. Tremor judgement by eye is inherently subjective and imprecise [ [3] ], and a need for objective tests is recognised [ [4] ]. Although laboratory accelerometery can distinguish functional tremor from other tremors [ [4] ], it is a limited resource. Smartphone accelerometers can measure tremor frequency and discriminate tremor type [ [5] ], but a clinical test whereby patients hold a smartphone is not one that has entered routine practice. An alternative ubiquitous item of hardware that could be used to assess tremor is the camera (present in smartphones, personal computers, CCTV). Computer vision technology uses algorithms to detect and interpret the contents of camera images [ [6] ]. It is widely used commercially, e.g. facial recognition, but there are only a few reports of its application within neurology. Here we describe early results for computer vision of functional tremor.