EczemaNet Study: Computer scoring of eczema severity in all skin tones
Research type
Research Study
Full title
EczemaNet Study: Feasibility of machine learning algorithm to assess eczema severity across a range of skin tones using digital photographs
IRAS ID
325468
Contact name
Reiko Tanaka
Contact email
Sponsor organisation
Imperial College London
Clinicaltrials.gov Identifier
N/A, N/A
Duration of Study in the UK
2 years, 5 months, 29 days
Research summary
Background:
Eczema is an itchy skin condition that affects 20% of UK children. Constant itching and sleep loss result in poor quality of life that affects the whole family.
Following the pandemic, many more consultations for eczema are taking place by video calls and by sending in photographs. However, assessing eczema severity on photographs is challenging. Doctors and nurses need better support for telephone/video assessment of eczema severity to provide the best treatment advice for people with eczema.
We have developed a computer software (called EczemaNet) that helps doctors and nurses to assess eczema severity from photographs. The software, also known as a machine learning algorithm, works well on white skin but is less accurate for assessing darker skin tones as the software was mainly developed using images of eczema on white skin.Aim:
We will train the computer software so that it can score eczema severity in children and young people of ALL skin types.Methods:
We will collect over 5,000 digital images of eczema in children and young people with a range of darker skin tones, pair these images with face-to-face assessments of eczema severity by clinical staff, and then expand the image bank by making over 200,000 computer-generated images.
These images are then used to 'train' the computer software to estimate eczema severity reliably on digital images. The computer software will learn to score the seven visible aspects of eczema that doctors look for when assessing eczema (dryness, cracking, redness/colour change, scratch marks, oozing, thickening, and swelling).
We will also get feedback on user-friendliness from children/carers and doctors.REC name
West of Scotland REC 5
REC reference
23/WS/0098
Date of REC Opinion
18 Jul 2023
REC opinion
Further Information Favourable Opinion