Skip to content

Computer Vision for Diagnosis of Parkinson’s Disease

Type of Project: Research and Development

Project Lead: Prof Rami Qahwaji

Project Summary: This RDF-funded Project is led by Prof Rami Qahwaji (PI) and includes Prof Raed Abd-Alhameed, Dr John Buckley, Dr Amr Rashad Abdullatif and Eng Ramzi Jaber (Funded PhD Studentship).

Parkinson’s disease is the second most common neurodegenerative disease, with a lifetime risk of around 2% [Ascherio and Schwarzschild, 2016], and causes muscle rigidity, tremors, and changes in movement and gait. There are 127,000 people diagnosed with Parkinson’s in the UK, the prevalence rising with an ageing population. Neurologists depend on observation of the patients to make clinical decisions about diagnosis, disease monitoring and treatment [Clarke et al., 2016]. However, visual observation is highly subjective and depends mainly on the clinician’s experience. Moreover, small or subtle changes in movement patterns and gait, which could be early signs of Parkinson’s disease, are usually missed by clinicians.

Bradykinesia is one of the cardinal symptoms associated with Parkinson’s disease. Typically characterised as a movement disorder, bradykinesia can be represented according to the degree of motor impairment. The inability to initiate movement effectively can stem from a combination of slowness of voluntary and involuntary or delayed movement. Parkinson’s disease (PD) can further cause tremors and stiffness which must also be considered in clinically assessing the degree of motor impairment. The assessment criteria for PD is therefore well defined due to the symptomatic nature associated with Parkinson’s. In the early stages of Parkinson’s irregularities in movement may be mild.

Artificial intelligence (AI) and computer vision can be utilised in the diagnosis of PD. Visual computing can be particularly useful in extracting key features of movement involved in facial expressions, gait and hand movements. Finger tapping is one of the most common clinical observations used to determine signs of Parkinsonian movement and involves tapping the index finger and thumb together. Visual computing offers a non-invasive and simple way of quantifying such movements in terms of pattern recognition, amplitude, rhythm and speed.

The early detection of this disease could lead much improved outcomes for patients. This project involves a multidisciplinary collaboration with Stefan Williams and Jane Alty, neurologists at Leeds General Infirmary, to develop visual computing and AI technology for detecting movement disorders and the diagnosis of early signs of Parkinson’s disease: also monitoring of neurological conditions by GPs and clinicians.

This project has NHS ethical approval (HRA approval; IRAS project ID:224848; REC reference:17/LO/0692; study title: Can computer processing of video distinguish Parkinson’s disease patients from controls? - A pilot study)





Project Aims: Using the technical and clinical expertise of the team, this project will exploit cutting-edge AI and visual computing technologies to develop a novel, usable and effective tool for the detection of early signs and monitoring of disease progression. Straightforward video capture followed by advanced computational analysis will deliver these results.

For more information, please contact Prof Rami Qahwaji



Motion tracking software identifying the finger and thumb of a human hand