AI Automates Detection of Mitral Regurgitation on Echocardiograms for Minimally Invasive Procedure

Mitral valve regurgitation is the most common but often missed heart valve disorder. Out of the four valves in the heart for facilitating blood movement throughout the body, the mitral valve, situated on the heart's left side, fails to close properly in some individuals, leading to blood flowing backward, a condition known as mitral valve regurgitation. This issue hampers blood circulation and may evolve into more severe complications such as shortness of breath, arrhythmia, and heart failure. Accurately determining the severity of this condition is crucial to deciding whether patients might adopt a watch-and-wait strategy or require immediate intervention. Researchers have now developed an artificial intelligence (AI) program to detect the presence and severity of mitral valve regurgitation from echocardiograms that could help identify patients for a minimally invasive procedure or surgery.

To develop the new program, investigators at Smidt Heart Institute at Cedars-Sinai (Los Angeles, CA, USA) utilized over 58,000 transthoracic echocardiograms—a type of ultrasound imaging used to evaluate heart conditions, including mitral regurgitation. The effectiveness of this program was evaluated using echocardiograms from 1,800 patients at Cedars-Sinai and an additional 915 from Stanford Healthcare. The findings, published in Circulation, show that the AI model demonstrated high accuracy in identifying moderate to severe cases of mitral valve regurgitation. After analyzing videos from more than 50,000 echocardiogram studies, the deep learning model effectively identified the most relevant and important videos for evaluating the severity of mitral regurgitation.

“This could improve how we identify patients with mitral regurgitation, which is becoming more prevalent in our aging population, and to personalize treatment even more so than we already do,” said Raj Makkar, MD, associate director of the Smidt Heart Institute, where the team has also performed more than 1,500 robotic mitral valve repairs with a near 100% success rates.

Related Links:
Smidt Heart Institute