Comparison of T1-weighted 3D high-resolution anatomical sequences for the brain at 3 Tesla: FLASH, MP-RAGE and MDEFT

Introduction: The outcome of automated quantitative morphometric population studies relies greatly on the quality of the cerebral anatomical images. For these applications, it is crucial that the images be uniform and have high spatial resolution, SNR (signal-to-noise ratio) and CNR (contrast-to-noise ratio) between brain tissues. This study evaluates the performance of three different T1-weighted pulse sequences at 3T (Tesla) in terms of SNR and CNR efficiency, as well as signal intensity NU (nonuniformity): FLASH (fast low angle shot), MP-RAGE (magnetization prepared rapid acquisition by gradient echo), and MDEFT (modified driven equilibrium Fourier transform). In both MP-RAGE and MDEFT, the 3D-FLASH readout is preceded by a magnetization preparation period to modify the contrast characteristics in a time-compact sequence design. The MP-RAGE preparation consists of an inversion pulse followed by a delay TI before the readout. The MDEFT preparation consists of two pulses: first a saturation pulse immediately followed by spoiler gradients to remove transverse magnetization then, after a time delay τ1, an inversion pulse is applied followed by an additional delay τ2. The total preparation time is TI= τ1+τ2. The inner centric 3D phase encoding loop of the readout in MDEFT is divided into 2 segments. Our objective was to determine the optimal 3T sequence for automated image analysis, and to identify further challenges that need to be addressed.