Can optical flow distinguish Parkinson’s patients from controls ?

  • Research type

    Research Study

  • Full title

    Can computer processing of video distinguish Parkinson’s disease patients from controls ? (A pilot study)

  • IRAS ID

    224848

  • Contact name

    Stefan Williams

  • Contact email

    stefan.williams@yhahsn.nhs.uk

  • Sponsor organisation

    The Leeds Teaching Hospitals NHS Trust

  • Duration of Study in the UK

    0 years, 8 months, 0 days

  • Research summary

    A cardinal feature of Parkinson's disease is an impairment of movement that is termed bradykinesia (‘slowness of movement’). In standard clinical practice, bradykinesia is assessed by asking the patient to repetitively tap their forefinger against their thumb. This brief clinical examination plays a fundamental role in multiple aspects of Parkinson’s disease care, including diagnosis, disease monitoring and medication prescribing. However, it relies on a visual judgement from the clinician’s observation, which means that it is inherently imprecise – the naked eye cannot accurately or reliably measure small changes in hand movements. Some of these considerations also apply to tremor.

    We have developed a method that uses computer analysis to track movement in a simple video of a patient hands. This could allow detection and measurement of bradykinesia and tremor in a way that is automatic and objective, and potentially more sensitive and reliable than the human eye. Ultimately, we would like to develop a widely available version of this technology, such as a smartphone app .

    This is a pilot, exploratory study, to demonstrate the proof of concept that computer optical flow can detect bradykinesia and tremor in video. 20 patients with Parkinson’s disease, 20 age matched controls and 20 younger age controls will be recruited from Leeds Teaching Hospitals NHS Trust (with written consent). Anonymous video of hands (resting and moving) will be filmed using both a smartphone and a higher frame rate camera, in a process lasting around 20 minutes. No patient identifiable details will be included in the video. Any clinical details will be kept within the Leeds Teaching Hospitals Trust. Anonymous videos will be processed by computing academics at Bradford and Edge Hill Universities, without any patient identifiable information. Exploratory analysis will be undertaken, looking for correlations between computer measurements from the videos and the diagnosis of Parkinson’s disease.

  • REC name

    London - Fulham Research Ethics Committee

  • REC reference

    17/LO/0692

  • Date of REC Opinion

    24 Apr 2017

  • REC opinion

    Favourable Opinion