Navigator-Free Dynamic Phase Correction for Echo-Planar Imaging Based Functional MRI

INTRODUCTION In dynamic echo-planar imaging (EPI), such as functional MRI, where a time series of EPI images are collected, there is additional modulation along echo readout due to changes in hardware states (e.g., temperature) that the standard reference scan based phase correction methods [1-3] cannot account for, thereby resulting in an increase of ghost level over time. To address this problem, non-phase-encoded navigator echoes can be collected at each temporal frame to calibrate the additional modulation between odd and even echoes [4]. The per-temporal-frame modulation is then corrected in image reconstruction in addition to the static modulation measured from the reference scan. However, the navigator-based method assumes that the additional modulation that the center echoes (corresponding to echoes covering the center of the k-space) experience is the same as that predicted by the first few navigator echoes. When this assumption is not true (e.g., when there is additional modulation building up across the echo train), the modulation of the center echoes would not be well corrected, thereby still leading to significant ghost drift. We propose a new, navigator-free method to use scan data itself for estimating the additional per-temporal-frame modulation. This method can estimate the modulation more faithfully than the navigator-based method, thus significantly reduced ghost drift.