Pilot Study of Machine Learning on Digitised Transplant Biopsies v1
Research type
Research Study
Full title
Machine Learning Applied to Digitised Kidney Transplant Biopsies: a UK Multicentre Study
IRAS ID
251381
Contact name
Candice A Roufosse
Contact email
Sponsor organisation
Imperial College
Clinicaltrials.gov Identifier
N/A, N/A
Duration of Study in the UK
5 years, 0 months, 1 days
Research summary
When a patient’s kidney or kidney transplant function is found to be abnormal during a visit to the clinic, a biopsy of the organ is often taken to determine the cause of the injury and to direct treatment. This biopsy is examined down a microscope at high magnification (up to 1000x original size). In addition to identifying the cause of injury, the pathologist reading the biopsy will also systematically score a number of features relating to the degree of inflammation and the degree of scarring. These scores are used to estimate how well the patient will respond to treatment and how long the organ is likely to last. All of these are used to inform patient management. In recent years it has become possible and affordable to capture a digital image of the entire slide, to export this image to a computer, and for pathologists to read the biopsy on a screen rather than down a microscope. In addition, it is possible to develop mathematical models that can be used to analyse the scanned images automatically.
Our project will develop mathematical models to analyse biopsies of the kidney. We will scan and annotate a large series of kidney biopsies (over 1000) from multiple UK sites. The scanned images will be pseudonymised and shared with academic and industrial researchers who will develop mathematical models, both of features we already know to be important, and potentially of new features that are hard to evaluate down a microscope. Results of the mathematical analysis will be compared to results by the pathologist, obtained the standard way with a microscope. Both will be compared to how long and how well the graft functions after the biopsy. The ultimate aim is to automate certain aspects of evaluation of the biopsy, leading to increased precision and speed.REC name
East of Scotland Research Ethics Service REC 1
REC reference
21/ES/0080
Date of REC Opinion
6 Sep 2021
REC opinion
Further Information Favourable Opinion