1354525 - |B1+|-prepared imaging for efficient cardiac transmit field mapping

Cardiac MRI at 3T offers increased SNR, but is often affected by artifacts, including B1+ field inhomogeneities [1]. Cardiac B1+ mapping remains particularly challenging due to motion, and few methods have reported robust results [2]. Recently, Bloch-Siegert (BS) shift based B1+ mapping [3] has shown promising results in the heart [4]. However, long echo required for BS-pulse may lead to artifacts, particularly in the presence of imperfect shimming or implanted devices. In this work, we introduce a new |B1+| preparation module to allow for time efficient BS based B1+ mapping.

Methods:

The proposed |B1+| preparation comprises a 90° flip down, two adiabatic refocusing pulses and 90° flip up pulses. Off-resonant BS Fermi pulses with alternating sign frequency shifts are played between the excitation and refocusing pulses (see Fig. 1B). Three types of images are acquired: 1) one leading to a |B1+|-dependent phase shift φBS; 2) equivalent preparation leading to a net 0 phase shift, to compensate for MT effects; and 3) a saturation prepared image without the |B1+| preparation to capture the effect of the imaging readout. Acquired images can then be related to φBS (see Fig. 1C), enabling B1+ magnitude map reconstruction [3]. Each image acquisition was preceded by a saturation pulse after detection of the R-wave to eliminate the effect of preceding readouts.

A homogeneous phantom and a healthy female subject (female, 24 years old) were imaged using a 3T scanner (Ingenia, Philips) using cardiac-triggered breath-held sequences. The |B1+|-prepared sequence used a single-shot bSSFP readout, with the following preparation parameters: duration 43 ms, 4 Fermi pulses with off-resonance fRF=7 kHz and tBS=4 ms, acquired with two averages. Reference was performed using a segmented spoiled GRE acquisition with fRF=7 kHz and tBS=8 ms [4]. Total breath-hold durations were 12s and 5s for the original and proposed sequences, respectively. Standard deviation (SD) maps were estimated in phantom (10 replicates) and in-vivo (3 replicates). SD maps were normalized to 10 ms scan duration and 1.3x1.3x10 mm resolution for comparison between in-vivo scans. Mean SD values were estimated in manually drawn ROI’s over the heart.

Results:

Reference and |B1+| prepared maps achieve high visual quality, displaying central brightening in the phantom images (Fig. 2). Despite higher resolution, the proposed technique achieves lower noise variability (SD 0.6% compared to 4.5% in the reference). Accordingly, in-vivo maps show lower normalized noise variability (Fig. 3). For selected cardiac ROI, SD was improved from 7.1% to 3.7% in SAX view, from 6.2% to 3.5% in 2CH view and from 8.8% to 6.0% in 4CH view, when compared with the reference.

Conclusion:

The proposed sequence provides a fast and SNR efficient way to measure transmitted RF field distortion in the heart. The preparation scheme allows for combination with various imaging modules, and reduced scan time is promising for fast 3D |B1+| mapping in the heart.