A theoretical model to assess the effect of respiratory motion on chemical
shift imaging (CSI) data of an extended sample is described. Data from the
model are shown to agree with those acquired experimentally From phantoms.
In CSI, Fourier bleed and ghost artefacts are of particular significance wh
en the tissue of interest is adjacent to other tissue with much higher conc
entrations of metabolites, such as chest wall muscle adjacent to breast tum
our tissue. We have evaluated contamination due to Fourier bleed and ghosti
ng due to subject motion separately. The results demonstrate that the amoun
t of signal contamination due to motion is relatively independent of the ra
tio of sample width to voxel width and of the period of motion, but has a s
trong dependence on the number of phase-encoding steps and the amplitude of
the motion.