Clinical Research Office. A partnership between Sheffield Teaching Hospitals NHS Foundation Trust and the University of Sheffield

New artificial intelligence tool speeds up diagnosis for thousands of NHS heart patients

A pioneering artificial intelligence tool measuring the heart's function in seconds could improve future heart disease care

  • Pioneering artificial intelligence tool automatically performs tasks that would normally involve lengthy manual analysis to provide fast and accurate measurements of the heart's function when reporting on MRI heart scans
  • The technology, which has a high degree of accuracy, has been tested on thousands of images, and validated in over 5,000 anonymised patient scans in Sheffield
  • Researchers have shown that the technology frees up vital NHS resource and aids earlier diagnosis by giving accurate and comprehensive analysis of the heart's function in seconds in a significant proportion of cases

 

A pioneering artificial intelligence (AI) tool which provides a quick and comprehensive analysis of the heart’s function could improve future cardiovascular care by aiding earlier diagnosis and giving more detailed information about the heart’s function.

Developed by researchers at Sheffield Teaching Hospitals NHS Foundation Trust and the University of Sheffield, The AI segmentation of cardiac MRI to automate the measurement of cardiac function and volume technology tool automatically detects chambers of the heart on images taken from MRI heart scans – performing tasks that would normally involve lengthy manual analysis within seconds.

MRI heart scans are requested by doctors to check on patient’s heart health, giving detailed information on how the heart is pumping. This allows for a diagnosis to be given, or treatments started or adjusted.

The current process for reading these results is time-consuming and resource-intensive, with doctors and cardiac imaging specialists first having to draw contours on the scan images of the heart and then undertake complex volumetric and mathematical calculations to work out blood flow in and out of the heart.

Researchers estimate that the AI tool will save doctors and expert imaging specialists up to 30 minutes per scan, freeing up vital NHS resource whilst also aiding earlier diagnosis.

Consultant Cardiothoracic Radiologist at Sheffield Teaching Hospitals and Senior Lecturer at the University of Sheffield Dr Andrew Swift said: “Getting answers quickly and accurately will reduce even further the time it takes for patients to begin receiving the right treatment. Obtaining complex measurements showing how well both the left and right side of the heart is pumping is a time-intensive manual task. The AI segmentation of cardiac MRI to automate the measurement of cardiac function and volume technology overcomes this problem. It has the potential to free up hospital staff to deal with more patients rather than spend time on image analysis. This is an excellent example of innovation from within the NHS and a proud legacy of the clinical and technical expertise we have here in Sheffield.”

The tool has been shown to have a high degree of accuracy comparable, if not superior, to manual analysis in a significant proportion of cases.

The technology has been extensively tested on thousands of images, and validated in over 5,000 anonymised patient scans at Sheffield Teaching Hospitals and further tested on scans from over 30 hospitals in the UK over the past three years, with the team now aiming to make it available to the wider NHS thanks to a £5,000 funding boost from a Medipex NHS Innovation Award win.

Professor Wendy Tindale OBE, Director of Innovation at Sheffield Teaching Hospitals, said: “We have a long and proud history of pioneering new research and innovation that can be adopted for wider patient benefit. We are delighted that the talents of our scientific and clinical teams in identifying and looking for solutions for healthcare problems have been recognised in this prestigious regional awards scheme.”

Around 10 to 20 MRI cardiac scans are usually processed at a workstation by radiographer a day.

The technology has been developed by Dr Andrew Swift, Dr Samer Alabed, Dr Kavita Karunasagaraar and Dr Pete Metherall with support from MRI radiographers and clinical scientists at the Sheffield 3D-lab and in collaboration with Dr Rob van der Geest at Leiden University.