Recently, a 4D non-breath-hold multiphase, steady-state imaging technique (MUSIC) was proposed using Ferumoxytol as an intravascular contrast agent [1] to address the issue of poor intra-cardiac definition and the necessity of breath holding in conventional CE-MRA. MUSIC provides accurate anatomical assessment of the heart and related blood vessels in children with congenital heart disease (CHD). Although left ventricular volume measured from 6-9 cardiac phases (temporal resolution: 80-50ms) of MUSIC correlates well with standard 2D CINE [1], we suspected that MUSIC with limited cardiac phases may miss the true end-systolic (ES) and end-diastolic (ED) phase, resulting in less accurate volume measurement. Also, cardiac wall motion assessment may be constrained. In this study, we compared ventricular volume measurement and cardiac wall motion assessment between MUSIC and an improved acquisition strategy that generated more cardiac phases. We hypothesized that increased number of cardiac phases on MUSIC would provide better functional assessment of the heart.Data Acquisition: The original MUSIC sequence was modified in such a way that ky-kz plane was sampled with (ky,kz) phase encodings grouped in spiral-like arms with rotating gold-angle ordering (Fig.1a). Each spiral arm started at k-space center (ky=ky=0) and represented a superior-inferior (SI) projection of imaging volume. These frequently sampled SI projections enabled estimation respiratory and cardiac motion through self-gating (Fig.1b) [2]. Data Binning and Image Reconstruction: Based on estimated respiratory and cardiac signals, acquired data were retrospectively binned into 18 cardiac phases that doubled the temporal resolution of original MUSIC. Images were reconstructed jointly for all cardiac phases by minimizing the following cost function [3]: $$argmin_x ||DFSx-y||_2+\mu_1||Wx||_1+\mu_2||Rx||_1,$$ where $$$S$$$ is the coil sensitivity map estimated using ESPIRiT [4], $$$F$$$ and $$$D$$$ are Fourier transform and under-sampling operations, $$$x$$$ and $$$y$$$ represent the reconstructed images for all cardiac phases and k-space measurement. Randomized shifting wavelets $$$W$$$ and Total Variation (TV) $$$R$$$ along cardiac dimension were used as the regularization term. Experiment Setup: 8 pediatric CHD patients were scanned under general anesthesia and controlled ventilation. Each patient received a Ferumoxytol bolus injection (4 mg-Fe/kg). Original MUSIC and the modified sequence were performed during the steady-state distribution phase of Ferumoxytol. Sequence parameters included: TR/TE: 2.9/0.9ms, FA: 25°, isotropic resolution: 0.8-1.0mm without interpolation, 6-9 cardiac phases for original MUSIC. Due to concerns about likely blood oxygen desaturation during ventilator controlled breath-hold, 2D CINE was not performed in these 8 critically ill patients with possible cardiopulmonary instability. Image Evaluation: Based on manually picked ES and ED phase from all available cardiac phases, ES volume (ESV), ED volume (EDV) and ejection fraction (EF) were derived directly from the 3D segmented left ventricle using Mimics (V17.0, Materialize). An error percentile of ESV, EDV and EF between original MUSIC and 18-phase reconstruction were also calculated. Subjective score of myocardial border and cardiac wall motion using criteria listed in Table 1 were visually assessed on both sets of images by an experienced radiologist. The reported scores represent mean standard deviation, and a paired student's t-test was used for statistical analysis, where $$$P<0.05$$$ suggested statistical significance.Fig.2 shows selected cardiac phases of reformatted short-axis view from original MUSIC and 18-phase reconstruction. Manually picked ES and ED phases are also highlighted. Due to the limited cardiac phases available, cardiac contraction on original MUSIC is less continuous compared with the 18-phase reconstruction. Table 2 summarizes calculated ESV, EDV, EF, error percentile and reader’s score. Due to the absence of 2D CINE, we were not able to directly demonstrate that 18-phase reconstruction provides more accurate volume measurement. However, as can be seen from Table 2, ESV, EDV and EF measured from original MUSIC is significantly over-estimated, under-estimated and under-estimated, respectively, compared with those from 18-phase reconstruction. This indirectly confirms that original MUSIC missed the true ES and ED phase. As for subjective score, original MUSIC has slightly sharper myocardial border compared with 18-phase reconstruction, but no significant difference was detected. In terms of cardiac wall motion assessment, 18-phase reconstruction is however significantly better than original MUSIC.With increased number of cardiac phases available, ES and ED phases can be better captured, providing more accurate calculation of ESV, EDV as well as EF. In addition, cardiac wall motion can be assessed with higher confidence providing more cardiac phases. Further systematic evaluation and comparison of the high temporal resolution reconstruction with gold standard 2D CINE is required and warranted.