How 4D Flow MRI and AI Can Transform Cardiovascular Disease Diagnostics
Oct. 4, 2023
By Danielle Henson
In a momentous achievement for the University of Louisville’s J.B. Speed School of Engineering, Dr. Amir Amini, professor and endowed chair of bio-imaging, has been awarded a substantial research grant from the National Institutes of Health (NIH). This prestigious grant, totaling $430,000 in funding for 2023, is set to propel Dr. Amini’s groundbreaking research in the field of cardiovascular imaging. Over the course of the award, Dr. Amini aims to revolutionize the diagnosis and understanding of aortic stenosis, a valvular heart disease that affects a significant portion of the aging population.
“It’s a significant achievement, and I want to emphasize that the credit goes to everyone including colleagues in the Cardiovascular Division at the School of Medicine who has contributed to this project,” said Dr. Amini. “To secure a grant from the NIH these days…there is fierce competition from prestigious institutions including Ivy League institutions. This accomplishment fills me with a sense of pride.”
Aortic stenosis, particularly the low-gradient subtype, poses a complex diagnostic challenge, impacting the lives of countless individuals. Dr. Amini’s research seeks to address this issue by combining cutting-edge 4D Flow MRI imaging technology with the power of artificial intelligence (AI). By doing so, he aims to provide a more accurate and efficient means of diagnosing the severity of aortic stenosis, ultimately improving patient outcomes.
A tenured faculty member in the department of electrical & computer engineering, Amini highlighted the significance of using AI in their work, stating, “We utilize a convolutional neural network in our research. This AI system takes imaged velocity data and generates pressure measurements, which are crucial in diagnosing vascular and valvular diseases. Our method has proven to be highly accurate. It’s likely the most effective approach available for deriving pressure values from velocity data. The network learns from training data and generalize to new data, effectively mapping velocity to pressure through training.”
The discordance between aortic valve area and mean gradient in aortic stenosis diagnosis is a critical problem affecting up to 40% of patients. Dr. Amini’s study will focus on developing efficient 4D Spiral flow imaging protocols and leveraging a Deep Learning framework to map velocities to pressures. This innovative approach will be applied to 50 subjects with potentially severe low-gradient aortic stenosis, with results validated against catheterization studies in a moderate aortic stenosis control group.