An Algorithm Helps Detect Blood Flow Patterns Correlated With a Heart Attack Precursor

Image of a heart

Coronary microvascular disease (CMD) is a complication of diabetes and metabolic syndrome, and studies suggest the disease may be a precursor to a heart attack — a common cause of death among people with either condition.

Because of CMD’s potential to provide an early warning that a person has or is developing heart disease, Nationwide Children’s Hospital researchers are working to develop a non-invasive method to identify CMD earlier. Currently, the disease is diagnosed invasively or through a process of elimination.

Aaron J. Trask, PhD, a principal investigator in the Center for Cardiovascular Research at the Abigail Wexner Research Institute, and his lab team are focusing on transthoracic Doppler echocardiography (TTDE) to assess changes in blood flow patterns as a means to detect CMD. Their research suggests flow patterns correlate with coronary microvascular data.

To improve TTDE image analysis, Trask’s lab developed an algorithm with off-the-shelf software.  Applied to images of healthy mouse models and mouse models of type 2 diabetes, the algorithm significantly reduced inconsistencies by and among users and reduced analysis time 30-fold.

The research, published in Scientific Reports, is the first step toward determining if coronary flow patterns may be useful in diagnosing CMD, says Dr. Trask, who is also an assistant professor of Pediatrics at The Ohio State University College of Medicine.

“Coronary microvascular disease is a subtle disease. If we develop a reliable computer program, a physician can utilize it to assess a patient’s coronary flow pattern and other measurements obtained during an echocardiogram to diagnose CMD with some degree of accuracy,” Dr. Trask says. “A patient could begin treatments to prevent worsening of heart disease before the disease progresses towards the heart attack stage.”

The research started when, while using TTDE, Dr. Trask’s lab noticed coronary flow patterns differ between healthy mice and mouse models of diabetes. Ian Sunyecz, then an undergraduate researcher, found some patterns correlate with changes in the coronary microvasculature.

To dig deeper, Sunyecz led development of an algorithm using MATLAB software. Research Assistant Kishan Patel applied machine learning to increase the algorithm’s power to predict whether CMD is present.

The lab is now collaborating with Christopher Bartlett, PhD, and William Ray, PhD, principal investigators in the Battelle Center for Mathematical Medicine, on a broader study of machine learning in predicting CMD from coronary flow patterns. The labs are also investigating cardiac function and other non-invasive parameters that, in combination with flow patterns, may increase the ability to predict and diagnose CMD.

Dr. Trask says the technology could potentially be applied to other diseases or procedures that alter coronary flow.

Reference:

Sunyecz IL, McCallinhart PE, Patel KU, McDermott MR, Trask AJ. Defining coronary flow patterns: comprehensive automation of transthoracic Doppler coronary blood flow. Scientific Reports. 2018 Nov 22; 8(1): 17268.