Principal Investigator: Mr Hugh O’Brien
Institution: King's College LondonTags: 49156, Cardiac, cardiac scar, CT, Imaging, MRI
Imaging cardiac scar tissue is important to assist in the diagnosis and treatment of many heart conditions. The current standard method is an MRI scan with gadolinium contrast agent. Many patients, such as those with kidney disease, are unable to get these scans due to the pressure it puts on the renal system. Others with cardiac implants such as pacemakers cannot receive it due to image artefacts from the metal in the implant interfering with the scanner.
This study aims to use data from patients who have had cardiac MRI scans to develop an algorithm which can predict scar in the heart wall without contrast agent. Scar tissue presence causes differences in shape and density of heart tissue which are difficult to manually identify but can be automatically identified using machine learning methods. Detecting these signs of scar without contrast agent would therefore be possible. We aim to make this method viable with other scanning methods such as CT or echocardiography since they also would be able to detect the signs of scar from the shape and density of the heart wall tissue. Since these scans are routinely carried out in cardiac clinics the clinicians would receive additional information from scans the patient is already receiving using this method. Such a method could be used to inform whether the patient would require an MRI or further clinical follow up.