AI-driven CT scan analysis offers new insights into heart health risks
Researchers at Case Western Reserve University, University Hospitals and Houston Methodist will harness the power of artificial intelligence (AI) to more accurately predict risk of heart failure and other cardiovascular events, including estimating when an adverse event might occur, by developing an AI model that "learns" from patient scans.
Cardiovascular disease is the leading cause of death worldwide, claiming over 17 million lives every year, according to the American Heart Association. Accurately identifying individuals at high risk remains a crucial unmet need. This initiative bridges the gap by developing advanced AI tools to analyze images from calcium-scoring computed tomography (CT) scans, a widely used diagnostic tool.
The scans, which indicate how much plaque is in a patient's arteries, also contain information about the aorta, heart shape, lung, muscles and liver.
The National Institutes of Health awarded two grants, totaling $4 million, to the collaboration to develop the AI model.
"This project represents a significant leap forward in personalized health care," said project leader Shuo Li, a Case Western Reserve professor of biomedical engineering and computer and data sciences. "It has the potential to set new standards for cardiovascular disease prevention and management, as well as advance the forefront of using AI to analyze images for transformational health care."
Innovative approach to predicting heart failure
The project creates AI-driven predictive models capable of interpreting combined data from calcium-scoring CT scans, clinical risk factors and demographics. Led by Li and Sadeer Al-Kindi, an imaging cardiologist and associate professor of cardiology at Houston Methodist DeBakey Heart and Vascular Center, the team aims to uncover deeper insights into the interplay between heart health and body composition. This would allow clinicians to identify at-risk patients with unprecedented accuracy.
"Accurate risk prediction allows us to tailor preventative treatments, reducing the burden of cardiovascular diseases and improving patient outcomes," Al-Kindi said. "By identifying risk of heart failure and other events early, this project can potentially redefine care protocols, save lives and lower health care costs."
Seamlessly adding AI to a clinician's toolkit
By leveraging existing screening CT data in two large health systems (Houston Methodist and University Hospitals), this research underscores the potential of AI to address longstanding clinical challenges in a cost-effective and scalable way.
"Our goal is to develop a non-invasive, accurate and personalized method for predicting cardiovascular disease risk," Li said. "This innovation will seamlessly integrate into existing clinical workflows, enhancing decision-making while minimizing the need for invasive diagnostic procedures."
Using low-cost, non-invasive screening
A calcium-scoring CT is a low-cost, non-invasive heart scan that identifies how much calcified plaque is in the coronary arteries. The plaque in the heart's arteries can narrow or block them and can predict someone's risk of heart attack.
The AI model will learn to extract novel insights from CT images and use these measurements to estimate risk of cardiovascular events in large cohorts. These measurements include coronary calcium, heart shape, body composition, bone density and visceral fat, in addition to age and other factors. AI can correlate outcomes with these risk factors much faster and more comprehensively.
"A clearer understanding of how these novel imaging-based risk factors combine will advance the knowledge of cardiometabolic disease phenotypes and support doctors in making appropriate and timely therapeutic recommendations," said Sanjay Rajagopalan, a professor and director of the Cardiovascular Research Institute at the Case Western Reserve University School of Medicine and chief of cardiovascular medicine at University Hospitals Harrington Heart & Vascular Institute.
The research team includes other key contributors from diverse disciplines at Case Western Reserve: David Wilson, the Robert Herbold Professor of biomedical engineering and radiology, and Pingfu Fu, professor of biostatistics in the Department of Population and Quantitative Health Sciences at the medical school.
Provided by Case Western Reserve University