Machine learning in cardiovascular magnetic resonance: basic concepts and applications.

Area of Science:

• Cardiovascular Magnetic Resonance (CMR)
• Machine Learning (ML) in Medical Imaging
• Deep Learning Applications

Background:

• Machine learning (ML) is increasingly impacting cardiovascular magnetic resonance (CMR).
• ML, especially deep learning, offers potential to enhance CMR imaging efficiency, quality, analysis, and patient evaluation.

Purpose of the Study:

• To review major areas in CMR where ML can assist clinicians and engineers.
• To highlight ML developments in image acquisition, reconstruction, analysis, diagnostics, and prognostics within CMR.

Main Methods:

• Review of recent developments in ML applied to CMR.
• Discussion of ML applications in image acquisition & reconstruction, image analysis, diagnostic evaluation, and prognostic information derivation.

Main Results:

• ML has significantly reduced time for CMR image segmentation and analysis.
• Fully automated, accurate, and reproducible quantification of ventricular mass and volume is now commercially available.
• Active research focuses on reducing acquisition/reconstruction time, improving resolution, and analyzing perfusion/myocardial mapping.

Conclusions:

• ML is transforming CMR, with current impact primarily in automated image analysis.
• Future research directions include optimizing image acquisition and advanced analysis techniques.
• Validation through large cohort studies, open-source practices, and controlled trials is crucial for safe and broad ML application in CMR.