The project is a collaboration with Dr Rob Van Der Geest Leiden University Medical Centre, and Dr Declan O’Regan at Imperial College. The aim is to develop AI tools to measure heart health using MRI/CT scans. One of the 42 of >500 successful applications.
Heart and lung diseases together account for close to half of all deaths in the UK, 26% and 20%, respectively. Picturing the working of the patient’s heart is vital to good diagnosis and treatment. Cardiac magnetic resonance imaging (CMRI) and computed tomography (CT) scans are key to diagnosis. Currently doctors draw on the images of the heart to clarify their view of its function. This information we describe as ‘physiological parameters’, unfortunately this procedure is time-consuming and two sets of drawings may give different results.
The application of computer AI (artificial intelligence) to image analysis is a fast developing field of research. Recently, an AI program trained on large numbers of scans with different types of heart disease has been shown to make accurate drawings on the scan images automatically for the left ventricle. This project uses both human and machine intelligence, in a process called ‘human in the loop’ to create an AI tool to measure the health of the heart in large cohorts of patients with different cardiac and pulmonary conditions. This method should generate improved ‘physiological parameters’ for the four chambers of the heart.
The AI assessment of the heart will provide important measurements for predicting how well treatments are working and patient survival in individuals with cardiac disease. The automatic ‘physiological parameters’ produced by this deep learning (human plus AI) approach could revolutionise disease assessment and improve treatment delivery and patient care. Key advantages are rapid assessment, minimised human error and a more consistent and comprehensive assessment of the heart.
AI tools and products are the first projects to be funded by the NIHR’s and will receive a share of funding totalling over £50 million. The funding forms part of the and the award is managed by the in partnership with and NIHR.