Improved signal-to-noise ratio in EPI sequences with highly asymmetric spin echo and highly asymmetric STEAM preparations (HASE-EPI and HASTEAM-EPI)

Objective

Alternative versions of echo-planar imaging (EPI) sequences with standard spin echo (SE) or stimulated echo acquisition mode (STEAM) preparations are proposed, allowing for improved signal-to-noise ratio (SNR), especially in high-resolution imaging. The suitability of the proposed sequences, dubbed "highly asymmetric SE-EPI" (HASE-EPI) and "highly asymmetric STEAM-EPI" (HASTEAM-EPI), is tested in vivo, for anatomical imaging with $T_2$ weighting and diffusion-weighted imaging (DWI).

Materials and methods

In HASE-EPI and HASTEAM-EPI, echo formation occurs prior to the EPI readout, rather than at k-space centre as in standard SE-EPI and STEAM-EPI. This allows for a constant preparation period, independent of the spatial resolution. The proposed sequences are compared to their standard counterparts, via simulations and experimentally via in vivo anatomical imaging and DWI.

Results

HASE-EPI and HASTEAM-EPI yield SNR improvements for large matrix sizes and fully sampled EPI readout. Simulations and in vivo results show a signal gain in HASE-EPI versus SE-EPI or HASTEAM-EPI versus STEAM-EPI for white and gray matter, particularly for higher spatial resolution with full readout in anatomical imaging and DWI. However, simulations also show that in the case of very long EPI-readout trains, HASE-EPI and HASTEAM-EPI are more prone to pixel broadening due to relaxation effects.

Discussion

In contrast to commonly used SE-EPI or STEAM-EPI, the proposed sequences facilitate the acquisition of anatomical imaging and DWI data with improved SNR, especially for full EPI readouts. However, the applicability of HASE-EPI and HASTEAM-EPI should be carefully assessed: while signal gains are less pronounced when using parallel imaging and/or reduced matrix sizes, there may be pixel broadening for very long EPI-readout trains.