End-to-End Spatiotemporal Analysis of Color Doppler Echocardiograms: Application for Rheumatic Heart Disease...

Area of Science:

• Cardiology
• Medical Imaging
• Artificial Intelligence

Background:

• Rheumatic Heart Disease (RHD) is a major global health issue, particularly in low- and middle-income countries.
• Early RHD detection via echocardiography is vital but limited by specialist physician scarcity in resource-limited settings.

Purpose of the Study:

• To introduce RADAR (Rapid AI-Assisted Echocardiography Detection and Analysis of RHD), a novel AI approach for early RHD detection.
• To provide a scalable, autonomous solution for RHD screening in underserved regions.

Main Methods:

• RADAR utilizes AI for end-to-end spatiotemporal analysis of color Doppler echocardiograms.
• It identifies key imaging views, analyzes cardiac cycle phases using convolutional neural networks, and examines blood flow patterns.
• The system was trained and validated on 1,022 echocardiogram videos from Ugandan children.

Main Results:

• RADAR achieved high performance on validation sets (accuracy 0.92, sensitivity 0.94, specificity 0.90).
• Independent testing demonstrated clinically acceptable performance (accuracy 0.79, sensitivity 0.87, specificity 0.70).
• RADAR outperformed existing methods in RHD detection.

Conclusions:

• RADAR shows significant potential for improving RHD diagnosis in resource-limited settings.
• The AI approach can enhance health equity by increasing access to timely and accurate RHD screening.
• This technology can aid vulnerable children by facilitating early intervention for RHD.