Resting state fMRI (rs-fMRI) enables the investigation of functional connectivity, identifying integrated brain networks at rest (RSNs), as well as revealing specific functional connections to a pre-defined region (‘seed-based’ analysis)¹. However, rs-fMRI is highly affected by artefactual low frequency components, originating from head motion, scanner drift and physiological pulsatile motion; leading to activation error and overestimation. Multi-Echo-EPI (ME-EPI) acquisition, combined with independent component (ICA) and echo time (TE) dependent ², is an emerging methodology to de-noise and remove such artefacts³. In this investigation we assess the advantages of ME-EPI over conventional single echo protocols, by acquiring task-evoked and resting functional information in the same scanning session in 8 healthy volunteers, with both protocols.

The study was approved by the KCL human research ethics committee. All scans were acquired in a 3T GE-MR750 scanner in 8 healthy volunteers (2 female). rs-fMRI and motor-task fMRI data were collected in all scans, with single echo and ME-EPI sequences. The order was counterbalanced across subjects, but both rs-fMRI scans were collected first to avoid carry over effects of activation. For all fMRI acquisitions (denoted SE_TASK, ME_TASK, SE_RS and ME_RS), identical parameters were used: FA = 80°, TR = 2500ms, TE = 30ms (single echo) and TE = 12, 28, 44, 60ms, voxel-size: 3.3x3.3x4mm. A single, 1mm isotropic T1-weighted MPRAGE images was also acquired.

Motor Task

Participants were required to squeeze and release two soft balls in each hand upon a visual cue; once every 2.5s for 40s periods, alternating with 40s of rest (5 repeats), total acquisition time = 6:40 mins, 160 images.

Resting State

Participants were scanned in the supine position whilst observing a cross-hair on a dark screen for a total of 8.00mins, 192 images.

Image Processing

Motion correction and DARTEL-based spatial normalisation were performed with SPM-12, and a ME-ICA de-noising toolkit². Motion parameter regressors were used in the first level analysis. A random-effects analysis was employed to derive group activation maps for the motor task and a paired t-test was conducted across sequences. Resting state data required further pre-processing; motion regression, regression of WM and CSF signal, data de-trending and band-pass filtering (>0.1Hz) prior to seed based connectivity analysis. After TE dependent analysis (which removes non-exponential decay of the ME signal) and ICA, a de-noised, optimally combined time series using all four echoes was obtained³. Two 6mm³ spherical seeds, identified from peak significance in the SE_TASK group analysis, were used to produce functional connectivity maps for both SE_RS and ME_RS.

Both ME-EPI and single echo data from the motor task revealed significant areas (p_voxel<0.001, p_cluster < 0.05) of activation in bilateral primary motor cortices, supplementary motor area (SMA), thalamus and cerebellum. However (as clearly shown in Fig. 1), ME-EPI derived activation was better localised and constrained in comparison to the single echo data.

One-sample t-test of the seed based rs-fMRI, using a seed in the SMA (obtained from the SE_TASK activation map) revealed significant (p_voxel < 0.001, p_cluster < 0.05) connectivity with the bilateral primary motor areas, visual cortex and superior cerebellum, using both methodologies. But significant connectivity with the insula cortex and the brainstem were only identified with the ME-EPI analysis (Seen from Fig. 2).

This is (to our knowledge), one of the first within-subject, within-session evaluations of conventional single echo fMRI acquisition against a combined ME-EPI sequence. We employed ME-EPI with TE-dependent and ICA analysis, to remove non-BOLD signal components in both task and resting state fMRI. We demonstrate that in both instances ME-EPI produces more spatially defined regions of activation vs single echo, and better represents the underlying functional anatomy from which the signal originates. We recommend the use of ME-EPI over single echo protocols for both task evoked and resting state functional studies.