Coronary heart disease (CHD) is the single largest cause of death in the United States. Coronary angiograms, x-rays that reveal the flow of blood through the heart’s pathways, are often used to diagnose and guide CHD treatment. However, they are largely reliant on physician interpretation making them potentially subject to human error. As a result, some patients are exposed to unnecessary procedures while others may fail to receive proper treatment.
Big problems require big solutions. Brahmajee Nallamothu, MD, MPH, an interventional cardiologist, set his sights on significantly improving coronary angiogram interpretation to enhance patient outcomes and reduce unnecessary procedures. Determined to find U-M collaborators, Nallamothu attended one of MCIRCC’s networking events, the Innovation Underground. MCIRCC’s Executive Director, Kevin Ward, MD, introduced Nallamothu to MCIRCC’s Data Science Director, Kayvan Najarian, PhD—a highly experienced researcher in signal processing and machine learning.
What began as a casual night of beers ended with two investigators from different backgrounds collaborating to achieve one common goal: improving the diagnosis and treatment of CHD.
Through initial funding from the Michigan Translational Research and Commercialization (MTRAC) for Life Sciences Innovation Hub, and additional funding from the American Heart Association, Najarian and Nallamothu are developing AngioAid, a fully automated, computer-based platform capable of analyzing angiogram videos and generating standardized assessments in real-time.
Using machine learning and advanced algorithms, AngioAid will analyze large data sets of angiogram videos to develop outcome prediction models. The platform has the potential to identify specific areas of coronary arteries with suspected disease for more intense study. It can also estimate the width of blood vessels and quantitatively determine the presence and percentage of blockages in each blood vessel.
Although there are existing computer-based techniques that assist with the interpretation of coronary angiograms, they require substantial input from clinicians, interrupt workflow, and require significant resources.
Nallamothu believes AngioAid will effectively guide clinical diagnosis in real-time as well as provide quality assurance without the limitations of human input.
In addition to developing the automated platform, the team is also producing a publicly available database of coronary angiogram videos to encourage the development of new algorithms to advance the field. The team hopes this database will lead to new educational and support resources for trainees and cardiologists in addition to new decision-support tools.
CHD affects hundreds of thousands of Americans annually and is the most common and deadly heart disease diagnosis in the United States. Nallamothu and Najarian hope AngioAid will benefit the millions of patients who undergo coronary angiograms each year and ultimately save lives.