Fast and Robust Single-Shot Cine Cardiac MRI Using Deep Learning Super-Resolution Reconstruction

Area of Science:

• Cardiovascular Imaging
• Artificial Intelligence in Medicine
• Medical Image Reconstruction

Background:

• Cardiac MRI is crucial for diagnosing cardiovascular diseases.
• Standard multishot balanced steady-state free precession (bSSFP) cine sequences are time-consuming and sensitive to patient motion.
• Deep learning (DL) offers potential for accelerated image acquisition and improved quality.

Purpose of the Study:

• To compare the diagnostic quality of DL-reconstructed single-shot (SSH) cine cardiac MRI with standard multishot cine.
• To evaluate the impact of DL on scan time and image quality metrics.
• To assess performance in specific patient subgroups, including those with arrhythmias.

Main Methods:

• Prospective study comparing DL-SSH cine with standard cine in patients undergoing cardiac MRI.
• Acquisition at 1.5 Tesla using breath-holding and ECG-gating.
• Subjective (readers' assessment) and objective (edge definition, SNR, CNR) image quality evaluation.
• Assessment of functional left ventricular (LV) parameters.

Main Results:

• DL-SSH cine acquisition time was 65% shorter than standard cine (92s vs 265s).
• DL-SSH cine demonstrated comparable or superior subjective image quality, particularly in endocardial edge definition.
• No significant differences in objective metrics like apparent signal-to-noise ratio (aSNR).
• Strong correlation for LV functional parameters (ejection fraction r=0.95).
• Improved image quality ratings for DL-SSH cine in patients with arrhythmia or unreliable breath-holding.

Conclusions:

• DL reconstruction of SSH cine cardiac MRI accelerates acquisition while achieving non-inferior diagnostic quality.
• DL-SSH cine offers superior diagnostic quality in challenging patient populations, such as those with arrhythmias.
• This technology holds promise for more efficient and effective cardiac MRI examinations.