Increased SNR and activation in Hadamard-encoded fMRI through physiological noise removal and phase correction

Introduction: Hadamard-encoded (H-encoded) fMRI [1] is a method that acquires a slab of multiple subslices with each RF excitation [2]. The subslices can then be extracted using temporal filtering, analogous to UNFOLD [3]. By setting the subslice width equal to that of a conventional fMRI scan, SNR is increased because of the larger slab thickness. However, since H-encoding uses complex data for subslice extraction, noise from physiological sources such as respiration and cardiac pulsatility can have a large effect on image integrity and functional activation. In this abstract, we introduce both physiological noise removal and phase correction in H-encoded fMRI, and compare the resulting functional activation with that of a conventional fMRI scan. Theory: In this work, each slab is composed of 2 subslices for H-encoded fMRI. The RF pulses consist of a frequency-modulated Hamming-weighted sinc function [1] r(t ) = [exp(−iπf0t )+ i(−1) n exp(iπf0t )]sinc( f0t )[0.54 + 0.46cos(2πt /T )] , where t ≤ T /2 , where T is the duration of the pulse, f0 is the bandwidth of the sinc, and n = 0,1, ..., N −1 is the time frame of the pulse for N frames. A low-pass filter is used to extract the subslice whose excitation phase does not alternate in time. To extract the other subslice, the original timeseries data is first multiplied by i(−1)n−1, then low-pass filtered.