The core clinical feature of Parkinson’s disease is bradykinesia. However, the most common method of clinical assessment, finger tapping, has poor inter-rater reliability, even among movement disorder specialists. Many technologies have been devised to objectively measure finger tapping, but virtually all involve specialised equipment, which may explain why none are in widespread use. One method involves patients tapping a smartphone screen, but this cannot detect tapping amplitude or decrement (key features of bradykinesia assessment).
Computer vision takes static or moving images from a camera, and then applies computing algorithms to automatically extract useful information. It is widely used in commercial applications, such as facial detection and recognition of facial expression. Crucially, the only hardware required is a simple camera with a computer processor, and such items are ubiquitous, e.g. smartphones. We report a computing method, including the technique of ‘optical flow’, which uses video from a smartphone to detect the pixels of the hand and track their movement during finger tapping. It has the potential to detect and measure bradykinesia without the need for specialised equipment. We present striking videos and early results comparing computer vision measurements of finger tapping with clinical ratings for Parkinson’s patients and controls.