Single-shot diffusion-weighted echo-planar imaging (EPI) is typically used
for most clinical diffusion studies due to its low sensitivity to patient m
otion. Although the Nyquist ghost artifact in EPI can be substantially redu
ced, there is frequently a residual ghost with low signal intensity. As rep
orted in this study, this residual ghost can produce severe artifacts when
maps of the apparent diffusion coefficient (ADC) are calculated from single
-shot echo-planar images. The artifacts presented in this paper appear as r
egions of apparently low ADC which simulate regions of reduced diffusion, b
ut are in fact generated by b-value dependent Nyquist ghosts of the orbits.
Data acquired in vivo were used to demonstrate that these artifacts can be
avoided by including standard methods of spatial presaturation or fluid-su
ppression in the diffusion-weighted EPI protocol. In addition, phantom stud
ies were used to illustrate how phase and amplitude variations in the ghost
generate the artifacts and theoretical expressions, derived elsewhere, wer
e used to provide a detailed understanding of the artifacts observed in viv
o. The level of Nyquist ghost reported for the current generation of commer
cial scanners suggests that this is a general phenomenon which should be a
consideration in all EPI-based diffusion studies. Magn Reson Med 42:385-392
, 1999. (C) 1999 Wiley-Liss, Inc.