AI Revolutionizes Standard Cardiac Imaging

Multiview deep neural networks improve diagnostic accuracy for heart conditions compared to single-view models.

Mar. 18, 2026 at 3:04am by Ben Kaplan

Got story updates? Submit your updates here. ›

Researchers from UC San Francisco developed a new "multiview" deep neural network (DNN) architecture that can analyze multiple imaging views of echocardiograms simultaneously to better detect cardiovascular conditions like left and right ventricular abnormalities, diastolic dysfunction, and valvular regurgitation. The multiview DNN approach outperformed traditional single-view DNN models, demonstrating that integrating information across multiple imaging views can improve the accuracy of AI-based cardiac disease diagnosis.

Why it matters

Heart disease is the leading cause of adult death worldwide, making accurate and efficient cardiovascular disease diagnosis a critical global health priority. Echocardiograms are one of the most commonly used imaging tools, but current AI models have been limited to analyzing only single views at a time. This new multiview DNN architecture represents a significant advancement that could enhance the diagnostic capabilities of AI in echocardiography and potentially other medical imaging modalities.

The details

The researchers developed a new "multiview" deep neural network (DNN) architecture that can analyze multiple imaging views of echocardiograms simultaneously, rather than the traditional approach of using only a single view. They trained demonstration DNNs using this multiview architecture to detect disease states for three cardiovascular conditions: left and right ventricular abnormalities, diastolic dysfunction, and valvular regurgitation. Compared to single-view DNNs, the multiview DNNs demonstrated improved diagnostic accuracy, suggesting they were better able to capture complex 3D cardiac anatomy and physiology by integrating information across multiple imaging perspectives.

  • The study was published on March 17, 2026.

The players

Geoffrey Tison

A cardiologist and co-director of the UCSF Center for Biosignal Research, who was the senior study author.

Joshua Barrios

An assistant professor in the UCSF Division of Cardiology, who was the first author of the study.

UCSF Health

The academic medical center of the University of California, San Francisco, which is world-renowned for its graduate-level health sciences education and biomedical research.

Got photos? Submit your photos here. ›

What they’re saying

“Until now, AI has primarily been used to analyze one 2D view at a time—from either images or videos—which limits an AI algorithm's ability to learn disease-relevant information between views.”

— Geoffrey Tison, Cardiologist and co-director of the UCSF Center for Biosignal Research (Nature Cardiovascular Research)

“Our multi-view neural network architecture is explicitly designed to enable the model to learn complex relationships between information in multiple imaging views.”

— Joshua Barrios, Assistant professor in the UCSF Division of Cardiology (Nature Cardiovascular Research)

What’s next

The researchers suggest that future research should examine how multiview DNN architectures may assist other medical tasks or imaging modalities beyond echocardiography.

The takeaway

This new multiview DNN approach represents a significant advancement in the use of AI for cardiac imaging, demonstrating that integrating information across multiple imaging perspectives can improve the accuracy of disease diagnosis. As heart disease remains a leading global health concern, this innovation has the potential to enhance cardiovascular care and management through more precise and efficient AI-powered diagnostic tools.