Feasibility of Continuous Ventricular Volumetric Quantification in Arrhythmias using Real-Time 3D CMR-MOTUS

BACKGROUND: Conventional cardiovascular magnetic resonance (CMR) cine sequences rely on binning reconstructions that average multiple heartbeats, an assumption that breaks down in arrhythmic patients where beat-to-beat variations lead to motion artifacts and loss of clinically relevant functional information. While two-dimensional (2D) real-time imaging can capture individual heartbeats, a stack of 2D slices is sub-optimal to map the full complexity of incoherent cardiac dynamics during arrhythmia. We investigated the feasibility of three-dimensional (3D) real-time motion field reconstruction for continuous beat-to-beat volumetric quantification in patients with premature ventricular contractions (PVC) using a free-running CMR protocol. METHODS: We extended CMR-MOTUS to jointly reconstruct real-time 3D motion fields and a motion-corrected reference image from continuously acquired data without breath-holds or electrocardiogram (ECG) gating. A variable-density Cartesian sampling trajectory (OPRA) was used with a 3D spoiled gradient echo or balanced steady-state free precession sequence. The real-time volumetric beat-to-beat changes were quantified by propagating a single manual segmentation on the reference image, through all time frames using the reconstructed motion fields. The method was evaluated on a cardiac motion phantom with ground-truth static acquisitions and in 10 healthy volunteers and 10 patients with PVC. All in vivo datasets were acquired post contrast enhancement. The ejection fraction (EF) was compared to ground-truth values for the phantom and to standard 2D real-time cine EF measurement techniques for the in vivo subjects. RESULTS: Reconstructed mean EF values of the phantom experiment were close to the ground-truth (Mean EF = 17.86 ± 0.33% versus 17.27%). In healthy volunteers, narrow beat-to-beat EF distributions reflected normal physiological consistency. In PVC patients, the method revealed broader, sometimes bimodal EF distributions. Simultaneously acquired ECG signals supported the temporal correspondence between volume irregularities and PVC episodes. CONCLUSIONS: 3D real-time joint motion field and image reconstruction from a free-running CMR protocol enables continuous beat-to-beat volumetric quantification in arrhythmic patients, revealing functional heterogeneity that conventional single-beat and averaging measurements (binning and gating) can obscure. Larger studies are needed before claims of clinical validation or replacement of standard 2D cine analysis.