Abstract
Functional magnetic resonance imaging (fMRI) studies in cognitive and clinical neuroscience rely on blood oxygenation level-dependent (BOLD) contrast, measured with single-shot gradient-echo-planar imaging. However, conventional schemes encompass the acquisition of single-echo fMRI, which samples a single echo at a single-echo time (TE), facing limitations in disentangling neural signals from artifacts. Multi-echo (ME) fMRI captures images at multiple echo times within a single repetition time (TR) period and enables the separation of BOLD and non-BOLD signal components. Previous studies have highlighted the benefits of ME-fMRI but often relied on comparisons with suboptimal single-echo data derived from ME acquisitions, limiting the validity of these evaluations. This study performs a more rigorous comparison between three datasets: the data acquired with an optimized single-echo (OSE) fMRI sequence at the highest possible temporal resolution, those acquired with an ME-fMRI sequence, and, as previously reported in the literature, the echo-2 time-series extracted from the ME-fMRI data itself. ME-fMRI vs. echo-2 comparison confirmed previous findings, which advantage the ME approach. However, the acquisition of multi-echo fMRI did not clearly outperform an optimized single-echo scheme. While OSE-fMRI exhibits benefits in terms of higher statistical power, ME-fMRI demonstrates superior performance at the single-subject level in terms of reliability (p < 0.05). Additional investigation and optimization could clarify the conditions under which one sequence may be preferred over the other.