Faster fat-water imaging with a novel multislice time-shifted GRASE-Dixon sequence

Introduction Recently, the gradientand spin-echo (GRASE) technique has been used to increase efficiency in Dixon fat-water imaging [1,2]. In these sequences, oscillating gradient lobes are carefully positioned between spin refocusing pulses to acquire multiple echoes with the necessary fat-water phase shifts. As it is necessary to collect at least two echoes with approximately π fat-water phase shift between them, the minimum time between refocusing pulses is of the order of 10 ms [1,2]. This determines the minimum acquisition time. Here we show that by using a dual slice spin-echo refocusing technique [3], double-echo GRASE and the two-point Partially-Opposed-Phase Dixon method (POP) [4], we can double the rate at which slices are acquired for Dixon imaging. The non-CPMG condition in Ref [3] is overcome by a novel time-shifting of pulses acting on different slices. Our sequence is similar to the SER-Dixon sequence reported in Ref [5], but Ref [5] did not use a multiple spin-echo train. Morevoer, Ref [5] used a single refocusing pulse acting simultaneously on two slices, which would not be usable in a multiple spinecho train without violating the CPMG condition. Method The pulse sequence schematic is shown in Fig. 1. Exc1 and Exc2 are 90° pulses of 1.2 ms duration, separated by 2.3 ms (time for π fat-water phase shift at 1.5T), and modulated to select different slices. Each 90° pulse is refocused by its own 1.2 ms slice selective pulse (Ref1 or 2) after a time τ. The read gradients are adjusted so that each readout lobe encodes an in-phase spin-echo from one slice, and the partially-opposed-phase echo from the other slice; e.g. in Fig. 1, the first readout lobe encodes echoes E2,o and E1,i corresponding to the opposed-phase echo from the second slice and the in-phase echo from the first slice. The second readout encodes echoes E1,o and E2,i which are the opposed-phase echo from slice 1 and the in-phase echo from slice 2. The time between E1,i and E1,o was 1790 us, giving a fat-water angle, α = 140°, at 1.5T. The time between E2,i and E2,o was 2810 us, giving α = 220°. The time between two echoes in each gradient lobe, e.g. E2,i and E1,o was 510 us. The choice of fat-water angles gives the same effective number of signal averages NSA >1.95 for both slices when reconstructed with the POP algorithm [4]. Note that both slices share the same phase encoding and spoilers. Spoilers were constant throughout the echo train and the same as for a RARE sequence. Also, because spin-echo refocusing reverses the direction of the k-space trajectory, the read gradient polarities need to be reversed after every pair of refocusing pulses. Phase errors due to fast switching of readout gradients were corrected using a separate reference scan as for EPI. As the bandwidth was quite high (1 kHz/pixel), chemical shift misregistration was neglected as a first approximation. The sequence was implemented on an Siemens Sonata 1.5T scanner. In-vivo images were acquired with body and head coils. Scan parameters were: TE = 10.6 ms, ETL = 9, τ = 5300 us, and matrix size per slice = 130 × 128.