Magnetic Resonance Imaging (MRI) is realized by the properties of a spinning proton (a small magnetic moment, or

1
Introduction Magnetic Resonance Imaging
(MRI) is realized by the properties of a spinning
proton (a small magnetic moment, or spin). MRI
correlates signal measurements with spatially
encoded frequency components by applying a
gradient magnetic field Gx. Signals with
spatially varying frequencies are produced
according to ?x?Gx, where ? is the gyromagnetic
ratio. Fourier transforms are used to invert
information from signal space to position space,
after which an image can be reconstructed.
Functional MRI (fMRI) is a relatively new
technique used to study neural activity in the
brain and spinal cord. When a region of the
brain is activated, it causes small signal
intensity changes (0.5-3). An fMRI experiment
requires consistency in the scanner trials may
last for days or weeks. In reality, scanners
drift over time (even between image
acquisitions), and hardware imperfections may
vary site-to-site. Even when usual performance
indicators (such as the signal-to-noise ratio)
are consistent, there may be significant image
intensity fluctuations across a series. These
fluctuations have the potential to degrade the
contrast-to-noise ratio, which is a measure of
the usefulness of an experiment.
Figure 1. Log-log plot of data from actual
stability experiment. Deviation from the
theoretical best is observed, as well as a
plateau in the relative fluctuation.
Acknowledgements This project would not
have been possible without the guidance, advice,
and generosity of Prof. Robert Brown. In
addition, I owe many thanks to Minhua Zhu for her
insight and assistance with finding and analyzing
data. I am incredibly grateful for the time both
have shared with me. Thanks also goes to
Michael Steckner and Toshiba Medical Research
Institute of Cleveland, Ohio, who allowed me the
use of data from stability experiments. I
enjoyed having the opportunity to visit and
observe experiments.
Methods The goal of this project is to
compare data from stability experiments that
(supposedly) induced correlated noise to
simulations that model potential sources of
instability. Two potential sources (that are
easy to model) are translational and oscillatory
motion. A simulation program was written (in
MatLab), that generates a phantom and assumes it
is moving. The Weisskoff analysis was then
performed for both types of motion. Next, a
spatially varying magnetic field was added for
each type of motion.
Conclusions We have been able to model
translational and oscillatory motion, as well as
varying magnetic fields. We have studied their
effects on simulated phantom data and compared
that to actual data from stability experiments.
It is evident that motion in the phantom plays a
role in experiment stability, but it is more
likely that the shape of the magnetic field plays
a greater role. This is supported by
experimental data.