UC San Francisco researchers developed a multiview deep learning network that improves diagnostic accuracy for major cardiac conditions from echocardiography data.
Key Details
- 1Researchers from UCSF created a 'multiview' deep neural network (DNN) architecture for echocardiograms.
- 2The model integrates data from multiple imaging views, instead of using a single 2D view.
- 3The DNNs were trained and tested on disease detection tasks for left/right ventricular abnormalities, diastolic dysfunction, and valvular regurgitation.
- 4Performance was superior to single-view DNNs, demonstrating improved diagnostic accuracy on real-world echo datasets.
- 5Findings are published in Nature Cardiovascular Research (March 17, 2026; DOI: 10.1038/s44161-026-00786-7).
- 6The research was funded by the National Institutes of Health.
Why It Matters
This approach addresses a long-standing limitation in imaging AI by combining information across multiple views, which better reflects clinical echocardiography practice. Improved model accuracy in echo interpretation could advance automated cardiac diagnostics and be translatable to other multi-view medical imaging modalities.
Source
EurekAlert
Related News
•EurekAlert
AI-Powered OCT Enables Rapid 'Optical Biopsy' for Early Endometrial Cancer Detection
A team at Washington University has developed a catheter-based 3D OCT system with AI to quickly and noninvasively detect early endometrial cancers.
•EurekAlert
AI Clinical Reasoning in Diagnostics and Digital Fatigue in Healthcare
Recent JMIR features explore large language models in clinical diagnostics and digital fatigue among healthcare professionals.
•EurekAlert
KAIST, MIT, Microsoft Develop Efficient AI Image Upsampling for Robotics
KAIST, MIT, and Microsoft have created 'Upsample Anything,' a training-free AI method to restore high-resolution visual data from compressed images with up to 16x improved GPU memory efficiency.