Computer vision of neurological examination signs - Version 1

  • Research type

    Research Study

  • Full title

    Computer vision of neurological examination signs to augment assessment and clinical diagnosis of neurological disease

  • IRAS ID

    353750

  • Contact name

    Stefan Williams

  • Contact email

    stefan.williams2@nhs.net

  • Sponsor organisation

    Leeds Teaching Hospitals Trust

  • Duration of Study in the UK

    5 years, 0 months, 1 days

  • Research summary

    Neurological diseases are varied and can present in many ways. Neurological examination signs are the findings a clinician makes when they meet with a patient and perform an examination. Examination signs, along with the history provided by the patient, help the clinician to establish a diagnosis or assess the severity of a condition. Examples of neurological examination signs that can be observed include: a tremor of the hands; an abnormal walking pattern; involuntary twitching of muscles.

    These observable neurological examination signs can be captured on video for analysis by computer software. This analysis is termed “computer vision”. There is research to show that computer vision can recognise some of these signs in video. However, current computer vision studies in this area have largely looked at isolated signs in individuals with a known diagnosis or a healthy control. This is different to what a clinician experiences, which is seeing a patient with one, or more, undiagnosed neurological conditions causing their signs. Further, many studies to date have used high-quality video equipment which would be a limiting factor to applying this technology in a clinic.

    This feasibility study would aim to look at neurological examination signs in adults who have been referred, or are under the care of, a hospital neurology team. We will video neurological examinations of participants, performed by a clinician, using a standard smartphone camera. We will apply computer vision techniques to the videos to identify and assess neurological signs.

    The aim is to see if the computer can ‘learn’ what certain neurological examination signs look like, and how this looks different to other neurological signs, or people without neurological signs. This could potentially lead in the future to the development of new tools to help clinicians better diagnose and assess neurological conditions.

  • REC name

    London - Dulwich Research Ethics Committee

  • REC reference

    25/PR/0169

  • Date of REC Opinion

    3 Apr 2025

  • REC opinion

    Further Information Favourable Opinion