Journal of Neuroscience Methods 227 (2014) 83–89 Contents lists available at ScienceDirect Journal of Neuroscience Methods jo ur nal ho me p age: www.elsevier.com/locate/jneumeth Basic Neuroscience Reducing respiratory effect in motion correction for EPI images with sequential slice acquisition order Hu Cheng a,b,∗ , Aina Puce a,b a Imaging Research Facility, Indiana University, Bloomington, IN, United States b Department of Psychological and Brain Sciences, Indiana University, Bloomington, IN, United States h i g h l i g h t s • We investigated the effect of respiration on motion correction of EPI images. • Respiration introduces additional noise after motion correction for long TR. • We proposed a new segmented motion correction for sequential acquisition order. • The method works best for superior slices with performance comparable to RETROICOR. a r t i c l e i n f o Article history: Received 24 September 2013 Received in revised form 10 February 2014 Accepted 11 February 2014 Keywords: Motion correction Respiratory noise fMRI EPI Sequential acquisition a b s t r a c t Motion correction is critical for data analysis of fMRI time series. Most motion correction algorithms treat the head as a rigid body. Respiration of the subject, however, can alter the static magnetic field in the head and result in motion-like slice shifts for echo planar imaging (EPI). The delay of acquisition between slices causes a phase difference in respiration so that the shifts vary with slice positions. To characterize the effect of respiration on motion correction, we acquired fast sampled fMRI data using multi-band EPI and then simulated different acquisition schemes. Our results indicated that respiration introduces additional noise after motion correction. The signal variation between volumes after motion correction increases when the effective TR increases from 675 ms to 2025 ms. This problem can be corrected if slices are acquired sequentially. For EPI with a sequential acquisition scheme, we propose to divide the image volumes into several segments so that slices within each segment are acquired close in time and then perform motion correction on these segments separately. We demonstrated that the temporal signal-to-noise ratio (TSNR) was increased when the motion correction was performed on the segments separately rather than on the whole image. This enhancement of TSNR was not evenly distributed across the segments and was not observed for interleaved acquisition. The level of increase was higher for superior slices. On superior slices the percentage of TSNR gain was comparable to that using image based retrospective correction for respiratory noise. Our results suggest that separate motion correction on segments is highly recommended for sequential acquisition schemes, at least for slices distal to the chest. © 2014 Elsevier B.V. All rights reserved. 1. Introduction A typical functional MRI (fMRI) dataset consists of many image volumes acquired sequentially in time. The outcome of the fMRI data analysis is greatly influenced by original image quality and data preprocessing. The data quality of fMRI is usually character- ized by the temporal signal-to-noise-ratio (TSNR) before statistical analysis. There are many sources of noise in fMRI time series (Lund ∗ Corresponding author at: Department of Psychological and Brain Sciences, Indi- ana University, Bloomington, IN, United States. E-mail address: hucheng@indiana.edu (H. Cheng). et al., 2006). Besides the thermal noise and hardware related signal fluctuations, physiological noise originating from cardiac pulsation, respiratory motion and neurovascular circulation contributes sig- nificant components to the noise spectrum. The physiological noise in fMRI data has been studied and characterized in great detail with many sophisticated methods available to reduce physiolog- ical noise in acquired fMRI data (Cheng and Li, 2010; Glover et al., 2000; Hu et al., 1995; Kruger and Glover, 2001; Lund et al., 2006). As a physiological noise source, respiration can have a number of effects on fMRI time series. The characteristic frequency of respi- ratory noise is around 0.2–0.4 Hz, higher than Nyquist frequency of typical fMRI data sampling. The origin of respiratory noise is http://dx.doi.org/10.1016/j.jneumeth.2014.02.007 0165-0270/© 2014 Elsevier B.V. All rights reserved.