Deep learning software reduces variability in cardiovascular imaging

San Francisco-based tech company Bay Labs this week announced the success of its deep learning software, EchoMD AutoEF, in reducing variability in cardiovascular imaging.

The software is able to evaluate left ventricular ejection fraction (EF) with a variability of around 8.29 percent, according to a release. That’s nearly 1 percent lower than the average variability of cardiologist readers using the Simpson’s biplane method to estimate EF.

“Historically, there have been challenges with variability and reproducibility in reporting of the ejection fraction, especially when the EF is not normal,” Richard Bae, MD, director of the Echocardiography Laboratory at the Minneapolis Heart Institute and the co-author of a related study, said in the release. “Our study showed that the EchoMD AutoEF algorithms can aid interpretation enormously and have less variability than cardiologists reported in literature.”

Bae said the efficacy of the AI algorithms will allow physicians to spend more time on what matters, like putting imaging results into context for patients and guiding them through treatment.

“Bay Labs EchoMD AutoEF was shown to automatically provide accurate EF calculations,” Bay Labs CEO Charles Cadieu said. “Our hope is that this will assist cardiologists in their decision-making.”