Improving cardiac care through innovative diagnostic systems

New grant will support the development of a ground-breaking precision medicine tool

Coronary heart disease (CHD) results from the buildup of plaque in coronary arteries, which reduces the flow of oxygen-rich blood to heart muscle. CHD can cause shortness of breath, chest pain, and can also lead to heart attack or heart failure, making it the single largest cause of death in the United States.

Coronary angiograms are x-ray pictures that reveal how blood flows through the heart’s arteries and can highlight any obstructions. They provide essential information for diagnosing CHD and determining how to treat and manage the disease.Yet when cardiologists assess these images, they have only their own eyes to visually evaluate the severity of potential blockages – estimations which drive critical decisions about whether a patient will need additional testing, surgery, or other major interventions.

Research suggests that without standard tools to make these measurements, cardiologists interpret angiograms incorrectly in up to one out of five cases – meaning that some patients do not receive potentially life-saving treatments, while others undergo unnecessary invasive (and costly) procedures.

Now, a team of University of Michigan researchers is developing a fully automated, computer-based tool to rapidly analyze angiogram videos and generate standardized assessments in real-time.

“This system will remove the variability in interpretation from one physician to the next, providing objective data needed for making accurate clinical decisions, and all without human input,” says Brahmajee Nallamothu, M.D., M.P.H., a University of Michigan interventional cardiologist,  co-developer of the new tool, and faculty lead for the Michigan Integrated Center for Health Analytics and Medical Prediction (MiCHAMP). Kayvan Najarian, Ph.D., a biomedical engineer with expertise in biomedical signal processing and image analysis, and director of the Michigan Center for Integrative Research in Critical Care (MCIRCC) Biosignal-Image and Computational (BIC) Core program, is co-lead on the project.

HOW IT WOULD WORK

The tool will build on a widely used computer modeling technique known as machine learning, which uses algorithms to analyze large data sets from a variety of sources to develop models and predict outcomes. In this project, the tool under development instructs its algorithms to focus on particular areas of the images that require more intense study, and then measure important vascular features and calculate the severity of blockages within the vessels.

Support from a new Innovative Development grant from the American Heart Association (AHA) Institute for Precision Cardiovascular Medicine, Nallamothu and Najarian are developing the tool based on successful pilot work at U-M funded by the Michigan Translational Research and Commercialization (MTRAC) for Life Sciences Innovation Hub. The team will validate and refine the algorithm by providing feedback from a database of hundreds of angiogram videos representing a diverse range of cardiovascular disease states.

The AHA competitive grants and fellowships foster cross-disciplinary learning, research and collaboration by bringing together scientists with unique backgrounds in data science and computer engineering. As part of the AHA Data Grant Portfolio, Amazon Web Services (AWS) will provide grant recipients, like Nallamothu, credits for computational analysis and data storage on the AHA Precision Medicine Platform powered by Amazon Web Services. 

“The promise of precision cardiovascular medicine and care can be realized when research and technology come together to deliver new insights,” said Jennifer Hall, Ph.D., Chief of the AHA Institute for Precision Cardiovascular Medicine. “The AHA and AWS collaboration will unite the global research community to accelerate discovery in cardiovascular health and usher in a new era of tailored prevention and treatment that will help patients and lessen the global burden of cardiovascular disease.”

The tool will break new ground by extending machine learning and image processing techniques already in use with still images to analyze videos, an advancement made possible through collaboration between engineers and physicians.

The team will also produce a publicly available, curated library of coronary angiogram videos intended to encourage the development of new algorithms to advance the field.

While there are existing computer-based techniques to assist cardiologists in interpreting coronary angiograms, these need substantial input from clinicians, interrupt workflow, and require significant resources.

“Our goal is to produce user-friendly software for fully-automated angiogram analysis that can be integrated into routine clinical practice, which could ultimately benefit the millions of patients undergoing coronary angiograms each year,” Nallamothu says.

The team also hopes that the tool could lead to novel scalable and cost-effective quality review and decision-support tools, as well as new educational and support resources for trainees and practicing cardiologists to help improve surgical accuracy.

The team will contribute their data from their works within the AHA Precision Medicine Platform. The AHA Precision Medicine Platform, powered by Amazon Web Services, is a cloud-based data marketplace where scientists from around the world can access, store, share and analyze research data.

MiCHAMP is a multidisciplinary research collaborative harnessing existing national, regional and local healthcare data to develop medical prediction models that address complex clinical problems. Through rich interactions with data science methodologists, informaticians, and clinical researchers, investigative teams aim to transform patient care to improve quality and decrease healthcare costs.

MCIRCC is one of the world’s first comprehensive research enterprises devoted to transforming critical care medicine by accelerating science and moving it from bench to bedside. To do this, MCIRCC brings together integrative teams comprised of U-M scientists, clinicians, and engineers with industry partners and funding sources to develop and deploy cutting-edge solutions that elevate the care, outcomes, and quality of life of critically ill and injured patients and their families.