Initial experience with a cardiac multi-contrast real-time cine prototype integrating sparse sampling and iterative reconstruction

Methods All patients were submitted to a conventional cardiac magnetic resonance study (Magnetom Aera, Siemens AG Healthcare, Germany) that included shortand long-axis steady-state free-precession (SSFP) segmented cine measurements (spatial resol.: 1.5x1.5 mm; slice thickness: 7 mm; temporal resol.: 40 ms; 7 heart beats (HB)/slice), modified Look-Locker inversion recovery post-contrast T1 mapping (spatial resol.: 1.6x1.6 mm; slice thickness: 8 mm; 17 HB/slice), and segmented spoiled gradient-echo late gadolinium enhancement (LGE) images (spatial resol.: 1.6x1.6 mm; slice thickness: 8 mm; 8-10 HB/slice). Followed by multi-TI real-time cine performed in the same cardiac planes (spatial resol.: 2.1x2.1 mm; slice thickness: 8 mm; temporal resol.: 45 ms; 4 HB/slice). The multi-TI cine prototype has been described in detail elsewhere, but briefly it consists of an inversion recovery highly accelerated SSFP 2D real-time cine sequence, featuring sparse sampling and k-t regularization. Using an offline reconstruction algorithm based on a registration and motion-propagation strategy, a full-length cine can be reconstructed for each acquired TI (fig. 1) and also a pseudo-T1 map cine. Results A total of 12 consecutive patients (61% male, 50±19 yrs) were included. All sequences were successfully performed and reconstructed, rendering good-quality images on subjective analysis. In all patients, a multi-TI cine, with ideal myocardial nulling, could be produced for simultaneous cardiac function and LGE analysis. Figure 2 illustrates a case of myocardial infarction with evident apical fibrosis on LGE and post-contrast T1 map, associated with akinesia of the involved segments on standard cines. On multi-TI cine, both abnormalities could be fully appreciated. In a subject with myocarditis, subtle mesocardial LGE without segmental contractility dysfunction was also adequately depicted on multi-TI cines (fig. 1). Conventional cine and LGE together required more time and breath-holds than multi-TI cines (745±210 seconds and 13±1 BH vs. 357±39 seconds and 3±0 BH respectively).