Cardiovascular disease continues to be the leading cause of morbidity and mortality in the United States. Magnetic resonance imaging (MRI) of scarred or fibrotic heart tissue plays a major diagnostic and prognostic role in patients with coronary heart disease (CHD) or non-ischemic cardiomyopathy (NICM). Cardiac MRI is a non-invasive, multifaceted imaging modality and is the clinical gold standard for scar and fibrotic cardiac tissue imaging with use of gadolinium-based contrast injection. However, administration of such gadolinium-based contrast agents (GBCA) prolongs the scan time, increases scan cost, and is contra-indicated in patients with impaired kidney function? a highly prevalent comorbidity in CHD patients. Until recently, GBCA was presumed to be safe in patients with normal kidney function; however, there are emerging data on long-term GBCA retention in the body.
We aim to develop two complimentary approaches to reduce GBCA use in myocardial, or cardiac muscle, scar imaging. Initially, we will develop a quantitative risk-benefit model to identify NICM patients with a low chance of having scarred myocardium. Concurrently, we will develop a GBCA-free cardiac MR myocardial tissue probe platform based on AI (MyoProbe.ai) to quantify scarred regions of the heart. To accomplish this, we will develop and evaluate an individualized, patient-specific scar prediction model to reduce GBCA use in NICM patients with different etiologies by training the model to learn to identify whether the patient is likely to have scarring based on non-contrast images. If it is unlikely that a patient has scarring, contrast administration can be avoided. To develop and evaluate MyoProbe.ai for GBCA-free quantification of myocardial scar in CHD patients, we will use AI to integrate signal intensity and heart motion data from MRI images to accurately locate and quantify scar tissue. This information can then be used by cardiologists to diagnose and treat the patient. We will rigorously validate our risk-benefit model and AI myocardial probe platform using retrospectively and prospectively collected cardiac MRI images from multiple healthcare centers, MRI vendors, and magnetic field strengths. Our dataset will include different types of NICM and CHD patient populations to ensure that our work is applicable to patients with many different types of NICM and CHD.
Cardiac magnetic resonance imaging (MRI) of scarred/fibrotic heart tissue with gadolinium- based contrast agents (GBCA) plays a major diagnostic and prognostic role in patients with coronary heart disease (CHD) or non-ischemic cardiomyopathy (NICM). However, administration of GBCAs prolongs the scan time, increases scan cost, contaminates surface and drinking water, is retained in the body long-term, and is contra-indicated in patients with impaired renal function? a highly prevalent comorbidity in CHD patients. The goal of our study is to reduce or eliminate GBCA use in cardiac MR scar imaging using artificial intelligence (AI) by concurrently developing a quantitative model to identify NICM patients with a low likelihood of scarring and a GBCA-free cardiac MR AI platform to quantify scar tissue.
