Matched Filter EPI Increases BOLD-Sensitivity in Human Functional MRI

INTRODUCTION Post processing of MR images is omnipresent: for fMRI in particular, smoothing of the raw images is frequently applied to gain SNR on the spatial scale of the BOLD response (‘matched filter’). This shaping of the point spread function (PSF) acts complementary to sampling itself, where the choice of a k-space trajectory imposes an intrinsic filter onto the data. We combine both approaches to pursue SNR optimization through the choice of an optimal acquisition strategy, incorporating the prior knowledge about the target smoothing kernel. Specifically, the implications for temporal SNR and fMRI t-maps in the case of Gaussian weighting are considered for a 2D gradient-velocity modulated EPI.