Quality Assessment of Kidney Transplants by Machine Learning

  • Research type

    Research Study

  • Full title

    Can an artificial neural network trained for image analysis perform histological quality assessment of kidney transplant biopsies?

  • IRAS ID

    244147

  • Contact name

    Gavin Pettigrew

  • Contact email

    gjp25@cam.ac.uk

  • Duration of Study in the UK

    0 years, 9 months, 1 days

  • Research summary

    Renal transplantation is the best treatment for patients with end-stage kidney disease (ESKD), but the use of poorer quality transplants - typically those retrieved from older donors (>65 years) - is associated with poorer outcomes. One way to offset these risks is to use a biopsy of the kidney to generate a ‘Remuzzi’ score that guides the decision to use the organ. The PITHIA trial (PB-PG-1215-20033), led by Mr Gavin Pettigrew, will determine the impact of a national, digital histopathology service (the first of its kind), on UK kidney transplant numbers. It is anticipated that this service will result in 120 extra kidney transplants in the UK each year, saving up to £3,500,000 for the health service.

    Recent advances in computing power mean that now, computers can ‘learn’ to analyse and classify objects accurately. In 2017, 29 of the 38 participating computer programs at the annual ImageNet competition could classify everyday objects more accurately than a human performing the same task (Computer error <5%; human error ~6-8%). Automatic analysis of the biopsy by a computer would offer several benefits. It would reduce the burden - financial and logistical - of maintaining a national biopsy service, minimise delays in processing, and could help reduce human error.

    We aim to develop computer program based on modern image analysis techniques that automatically reproduces the ‘Remuzzi’ score currently assigned by Renal Histopathologists in assessing biopsies from kidneys retrieved for transplantation. The aim is to produce a software that is ready to integrate into the National Digital Histopathology Service, should it be implemented nationally at the end of the PITHIA trial.

    This initial project is a proof of concept study, that will use existing samples previously taken from NHSBT organ donors currently stored through the Cambridge University Hospitals Tissue Bank.

  • REC name

    East of England - Cambridge South Research Ethics Committee

  • REC reference

    19/EE/0022

  • Date of REC Opinion

    11 Feb 2019

  • REC opinion

    Favourable Opinion