1351965 - Utilization of Deep Learning Image Reconstruction to acquire rapid cine Images in a patient with severe arrhythmia

61-year-old female with history of atrial fibrillation who developed new onset ventricular arrhythmia with shortness of breath.  The patient has a history of mitral valve prolapse with mitral regurgitation.  The patient underwent coronary CTA demonstrating normal epicardial coronary arteries.  She was referred for cardiac MRI to evaluate for etiology of ventricular arrhythmia, ventricular function, and degree of mitral valve prolapse/mitral regurgitation.  

Diagnostic Techniques and Their Most Important Findings:

Standard steady state free procession (SSFP) images were acquired in the short axis plane.  An additional short axis stack was acquired utilizing deep learning (DL) cine (acquisition over 1R-R interval).  DL Cine uses an unrolled reconstruction network that consists of 8 iterative blocks including both data consistency update and trainable convolutional neural network [1,2]. DL Cine network is trained on 232 fully sampled 2D FIESTA CINE dataset, with a retrospective variable density k-t sampling scheme [3]. l1 loss between ground truth and DL Cine reconstruction images is used as the loss function for network training. Real-Time short axis cine images were also acquired to image quality comparison.  Multiplanar late gadolinium enhancement imaging, 4D Flow, and additional multiplanar cine imaging through the mitral valve was also performed. 

Standard SSFP images in our patient are severely marred by the patient’s severe arrythmia.  Accurate evaluation of wall motion or function is severely limited on the standard SSFP and real time cine images.  Leverage of DL image reconstruction to acquire rapid, 1R-R cine images allows accurate evaluation of wall motion and functional analysis despite patient’s severe arrhythmia.  The calculated right and left ventricular function was similar to patients echocardiogram performed prior to the cardiac MRI. 

Learning Points from this Case:

Although the image quality is slightly reduced compared to standard SSFP, the ability to accurately evaluate wall motion and ventricular function utilizing rapid 1R-R DL Cine, makes this an excellent option for patients with arrhythmias.  Additional research needs to be performed to evaluate rapid, 1R-R DL cine in other imaging planes, additional types of arrythmias, and faster heart rates.