Partnership could transform kidney treatment
The University of Cambridge has been working with SAS to revolutionise kidney treatment in the UK. Artificial Intelligence (AI) and computer vision can automate the process of scoring biopsies for kidneys to better select kidneys for transplantation.
The aim is to increase the number of transplants and improve the function of those kidneys used. This has the potential to save lives and transform the quality of life for more than 100 people each year who would otherwise require dialysis. Dialysis is expensive and using transplants for this many people will save the National Health Service (NHS) around £3.5m annually.
The partnership followed the creation of the PITHIA trial (Pre-Implantation trial of Histopathology In renal Allografts), led by Mr Gavin Pettigrew, Reader in Experimental and Clinical Transplantation at the University of Cambridge, and supported by the Office for Translational Research (OTR). It has been funded by the NIHR (National Institute for Health Research) and assesses the impact of a national, digital histopathology service - the first of its kind - on UK kidney transplant numbers.
Renal transplantation is the best treatment for patients with end-stage kidney disease. To meet the increasing demand in the UK for transplantation, more and more elderly deceased donors are being considered. However, kidney function deteriorates with age, and kidney transplants from elderly donors are associated with higher risks of early failure and are considered ‘marginal’.
Early failure of a kidney graft is a disastrous outcome for the recipient, because they undergo a major operation with no lasting benefit; and as a result, mortality rates are high for this group of patients. Consequently, many organs from elderly donors are either declined or discarded as unsuitable.
One way to potentially offset these risks is to perform a biopsy of the kidney that assesses the degree of chronic injury or age-related-damage to the kidney. This information is used to guide the decision to implant the organ. However, this is not current practice because of the time and specialist histopathology skills needed.
By minimising human involvement, an assessment of the digitalised kidney biopsy image using computer vision technology could enable the introduction of National Digital Pathology scheme. Such a scheme could allow greater numbers of kidneys to be safely transplanted from elderly donors, leading to a significant increase in the number of kidney transplants performed each year in the UK.
Mr Pettigrew said: “It has been a tremendous experience working closely with SAS in developing a solution to AI interpretation of renal biopsies. Being able to bring our clinical experience to open and productive discussions at all stages in the development of the project, has undoubtedly been extremely productive. It means all parties understand how the service works, and the eventual translation to the clinic will be very much a blend of human expertise and state of the art computational neural network technology.
“Each kidney transplant saves approximately £30,000 annually over the costs of dialysis and improves both survival rates and patients’ quality of life. So, if successful, we expect that automated digital pathology services will be widely adopted, even beyond the UK.”
It is believed the new approach would continue to drive better quality decisions and reduce errors, because there would be access to the collective experience of everybody compared to the experience and judgement of someone acting alone or as part of a small group. Access to this knowledge would be available at all hours, helping to make better decisions more quickly.
“SAS software can align the AI with the human process of specialist medical review and inform the decision-making process by highlighting the relevant features in the biopsy image,” explained Simon Tilley, Director of Life Sciences for SAS EMEA.
“It is technology augmenting normal clinical decision-making, in that medical experts can see how each organ has been scored. It is both interpretable and transparent, as it is possible to see how each score for each component has been arrived at.”
To find out more about the research and the impact of this new service, you can register for the University of Cambridge’s Symposium on AI and Clinical Imaging by taking place on 27th May from 2pm BST.