Abstract
Purpose/Introduction: Microscopic tagging can be used to enhance
the spatial resolution of fast MRI sequences (1) without the sensitivity
to motion-related phase fluctuations, which plagues conventional
multi-shot scanning schemes. It is a direct analogue of the structured
illumination method of super-resolution (SR) optics (2,3) and can be
seen more generally as a combination of standard Fourier encoding
with a phaseless encoding basis (4). This novel acquisition strategy
may produce unusual artefacts due to experimental imperfections.
The purpose of this study was to establish their catalogue based on
simulated signals and to test possible correction methods (5).
Subjects and Methods: Phaseless SR encoding was simulated by
numeric modulation of the high-resolution Shepp-Logan phantom
with different shifts, followed by 2DFT and clipping to the low-resolution k-space range. The modulation pattern was modified to
mimic experimental imperfections such as a B0 field offset (pattern
shift), flip-angle deviations (pattern amplitude), incomplete relaxation
(non-sinusoidal modulation) and random signal amplitude variation
between encoding steps. The low-resolution magnitude images
reconstructed in each step were further processed to obtain SR images
by algebraic separation of k-space bands (4) without any correction,
and with a correction based on the B0 and B1 maps (5).
Results: A ringing effect appears at locations of static B0 inhomo-
geneity, which causes an offset of the periodic tagging and gives a
phase error of k-space side bands. Periodic stripes occur as a result of
signal amplitude fluctuation, which gives rise to extra sinusoidal
modulated images superimposed on the ideal image. Low/high-pass
filtering and ringing appear when the tagging flip angle deviate from
the nominal value of 45°, which results in false amplitude of recon-
structed k-space side bands and perturbs high frequency components
of the image. A slight ringing effect occurs when the tagging mod-
ulation is distorted by incomplete relaxation.
Discussion/Conclusion: The method of phaseless encoding may be
sensitive to static B0 inhomogeneity due to deformations of periodic
tagging. This problem can be minimized by keeping the tagging
sequence short. It can also be completely corrected during recon-
struction by including the B0 map in the band combination step. The
reconstruction can also correct for the flip angle deviations caused by
inhomogeneous B1. Further, so far uncorrected sources of artefacts
are fluctuations of signal amplitude (the fluctuations of phase are
inherently ignored by the phaseless encoding) and, to lesser extent,
pattern distortions caused by incomplete relaxation between the tag-
ging sequences.
