Download A Brief Introduction to Functional MRI

FUNCTIONAL MAGNETIC RESONANCE IMAGING
A Brief Introduction
to Functional MRI
History and Today’s Developments
© DIGITAL STOCK
BY JAMES J. PEKAR
n this issue, IEEE Engineering in Medicine and Biology
Magazine focuses on modern methods for the analysis of
data from functional magnetic resonance imaging (fMRI)
studies. Accordingly, the guest editors have seen fit to
begin with a brief article on the mechanisms and methods
behind fMRI. In this context, it is worth noting that magnetic
resonance imaging (MRI), in the first place, rests on a series
of unlikely accomplishments: Who might have thought, at
the turn of the last century, that the spin of subatomic particles could be detected in bulk matter using radio frequency
(RF) energy, or that RF energy, with wavelengths of tens of
centimeters, could be used to form exquisite images with
submillimeter spatial resolution?
In 1924, Wolfgang Pauli suggested, on theoretical
grounds, that particles could have intrinsic angular momentum, or spin. In the 1920s, Otto Stern and I.I. Rabi showed
that beams of such particles, traveling in a vacuum through
strong magnetic fields, would absorb RF energy in a narrow
range of frequencies. Those beam experiments were the first
measurements of spin resonance.
In 1946 Felix Bloch and Edward Mills Purcell showed that
such spin resonance could be detected among nuclear spins
in bulk matter; for this they shared the Nobel Prize in
Physics in 1952. This phenomenon, dubbed nuclear magnetic resonance (NMR), turned out to be immensely valuable
for chemistry because nuclei resonate with exquisite sensitivity to local electromagnetic fields and consequently serve as
lucid reporters on their microenvironments.
Applications of NMR to chemistry grew rapidly, as did
applications to living systems. In the late 1950s, scientists
began applying NMR to isolated cells and excised tissues;
by the late 1960s, NMR data were being acquired from
intact animals [1]. In the early 1970s, scientists noted that
such tissue NMR signals from water’s hydrogen nuclei
change in disease. A seminal contribution was made in
1973 when Paul Lauterbur [2] showed how NMR signals
could be used to form an image, using a spatial gradient of
the static magnetic field to yoke frequency to location, by
causing the resonance frequency of nuclei to vary linearly
with their spatial location. It is worth noting that these
images are not diffraction limited in that, unlike, for example, optical or electron microscopy, we can use MRI to
I
24 IEEE ENGINEERING IN MEDICINE AND BIOLOGY MAGAZINE
resolve features much smaller than the wavelength of the
RF energy used. In 2004, Lauterbur shared the Nobel Prize
in Physiology or Medicine with Peter Mansfield [3], who
laid the foundation for snapshot MRI (image formation
from data acquired in a fraction of a second). Richard Ernst
introduced another central innovation [4] when, in 1975, he
first demonstrated a Fourier rather than projection-based
approach to magnetic resonance (MR) image encoding; for
this (and other) work, he was awarded the Nobel Prize in
Chemistry in 1991.
Development of commercial MR scanners took off in
the 1980s, and by 1990, clinical MRI was of primary
importance for brain tumors, stroke, and multiple sclerosis. In 1990, Ogawa [5] showed that these MRI water signals can be sensitized to cerebral oxygenation, using
deoxyhemoglobin as an endogenous susceptibility contrast agent. Using gradient-echo imaging, a form of MRI
image encoding sensitive to local inhomogeneity of the
static magnetic field, Ogawa demonstrated (in an animal
model) that the appearance of the brain’s blood vessels
changed with blood oxygenation. Within two years, his
and two other groups had published papers using this
blood-oxygenation-level-dependent (BOLD) contrast MRI
to detect brain activation in humans [6]–[8], and, today,
an explosion of studies use this so-called fMRI technique
to map human brain function.
Of course, BOLD fMRI does not measure brain function
directly. Rather, BOLD fMRI brain activation studies are
the latest in a line of approaches, dating from the 19th century, which use brain perfusion as a proxy for brain function. Without making direct measurements of brain
function (i.e., without measuring computations performed
in neuronal cell bodies, action potentials traveling along
axons, or neurotransmitter trafficking at synaptic junctions),
these approaches take advantage of the phenomenon that
increases in neuronal activity are accompanied by local
increases in perfusion. Generally, these approaches map
changes in perfusion (or its concomitants) to shed light on
regional changes in brain activity. Specifically, BOLD
fMRI sensitizes MRI acquisitions to the local decreases in
deoxyhemoglobin due to reactive hyperemia [9] accompanying neuronal activation: Following an increase in
0739-5175/06/$20.00©2006IEEE
MARCH/APRIL 2006
neuronal activity, local blood flow increases (the mechanisms responsible for this neurovascular coupling are still
being explored). The increase in perfusion, in excess of that
needed to support the increased oxygen consumption due to
neuronal activation, results in a local decrease in the concentration of deoxyhemoglobin. As deoxyhemoglobin is
paramagnetic, a reduction in its concentration results in an
increase in the homogeneity of the static magnetic field,
which yields an increase in the gradient-echo MRI signal.
Given that BOLD fMRI does not measure brain activity
directly but, rather, relies on neurovascular coupling to
encode information about brain function into detectible
hemodynamic signals, it may be useful to look at how most
fMRI studies are designed and performed.
For BOLD fMRI, image data are typically acquired slicewise using single-shot echo planar imaging (EPI), the snapshot imaging method proposed by Mansfield. A
frequency-selective RF pulse is applied in the presence of a
static magnetic field gradient to selectively excite nuclear
spins in a virtual slice; the slice-select gradient is then turned
X [A] AFNI 2.56b: ./20050603human
X
Colr
Swap
off, and the signals from these spins are encoded along the
dimensions of the slice using rapidly switched magnetic field
gradients. Within approximately 50 ms, a dataset is acquired,
which, when Fourier transformed, will yield an image of the
slice in question. This is rapidly repeated for all the slices
(typically around 30–40) in the brain, such that a complete
multislice volume is built up within a time of repetition (TR),
after which the process is repeated. For a typical TR of 2 s, if
200 volumes are acquired, then we have a volume movie
consisting of 200 volumes of the brain (each consisting of
30–40 slices) acquired over 400 s. During these 400 s, a
neurobehavioral paradigm is played out in which the
research participant is exposed to sensory stimuli or asked to
perform some set of mental and motor tasks or some combination of them. So we have a situation where 400 s of temporally structured brain activity (e.g., watching flashing lights
every other 30 s, tapping one’s fingers every other 20 s, reading words, or solving math problems) are accompanied by
the acquisition of a brain volume movie with 2 s temporal
resolution (see Figure 1).
[A] AFNI 2.56b: ./20050603human2_4_1.hdr+orig and 20050603human2_4_
557
[+102]
Norm
c
b
r
g
i
9
z
pan
crop
30
Disp
Sav1.ppm
X
[A]
Mont
Done
Rec
2.50
AFNI 2.56b: ./20050603human
475
42
Disp
Sav1.ppm
Mont
Done
Rec 897
ent = 3.65
X: 42
Y: 38
Z: 30
Grid: 20
# 0:119
Index=0 Value=505 at 0
Scale: 1 pix/datum
Mean: 506.1583
Base: Separate
Sigma: 15.29321
FIM
Op
Fig. 1. Sample fMRI data (from an alternating hand-squeezing paradigm). Over a period of 240 s, a participant engaged in a
paradigm consisting of alternating 6 s of rest, 6 s of left hand squeezing, 6 s of rest, and 6 s of right hand squeezing. Instructions
were provided visually. Data were acquired at 3.0 T using single-shot echo planar imaging of axial slices. Note the relatively
poor spatial resolution and contrast (compared to anatomical MRI). The right-hand figure graphs single-voxel raw data from
a 5×5 grid of voxels in one axial slice. Note how the paradigm-related signal changes can be easily seen by the naked eye,
even though no preprocessing has been performed on these data. The red curve shows the ideal time course (convolution
of paradigm timing with assumed hemodynamic impulse response function) used to calculate the activation map shown
overlaid in color. Data acquired at the F.M. Kirby Research Center for Functional Brain Imaging
(http://mri.kennedykrieger.org); the figure is a screen shot from the AFNI program (http://afni.nimh.nih.gov).
IEEE ENGINEERING IN MEDICINE AND BIOLOGY MAGAZINE
MARCH/APRIL 2006
25
The future appears to promise a
more integrative approach to
functional brain imaging.
This brain movie has the interesting property that any single image from it contains no information about brain function. Instead, information about brain function is encoded in
the variance of image intensity over time. How then are such
data typically analyzed to yield the now-familiar brain activation maps? The conventional approach to fMRI data
analysis consists of three steps: preprocessing, regression,
and inference [10].
In preprocessing, the raw images are subjected to spatial
registration to correct for small head motions, temporally
interpolated to compensate for the fact that different slices
are acquired at different times, spatially smoothed to enhance
signal to noise, and often spatially normalized or transformed
into a common stereotactic space to facilitate group analyses
and neuroanatomical labeling.
In regression, the paradigm timing is convolved with an
estimate of the hemodynamic impulse response function.
The resultant hemodynamically lagged and blurred version
of the paradigm timing is used as a regressor of interest,
forming a general linear model, which is then fit to the data,
allocating temporal variance in the (preprocessed) data
among such regressors.
In inference, the spatial map of regression coefficients resulting from the regression step is thresholded for significance,
allowing formation of a map showing the suprathreshold
regions as hot spots, often overlaid in color on a (higher resolution) anatomical MR image.
This standard inferential univariate fMRI data analysis
approach can be seen as an excellent machine for testing
prior temporal hypotheses. However, it cannot discover
unanticipated structure in the data, i.e., it cannot detect brain
activations with timings that were unanticipated by the
investigator. In the last decade, investigators have developed
a variety of data-driven or exploratory techniques that can
discover brain activity not anticipated in advance. It may be
useful to view such approaches in the context of the finding
[11] that in the resting state, synchronous fluctuations (or
covariance) of voxel time courses were found throughout the
bilateral motor cortex.
Recent years have seen significant progress in paradigm
design for fMRI as well as the development of other methods
for assessing the functional anatomy of the human brain,
such as diffusion tensor imaging [12] for mapping white
matter fiber tracts. The future appears to promise a more
integrative approach to functional brain imaging, in which
data from multiple modalities are entered into comprehensive analyses of brain function and connectivity.
James J. Pekar is an associate professor of radiology at the
Johns Hopkins University School of Medicine and serves as
manager and research coordinator of the F.M. Kirby
26 IEEE ENGINEERING IN MEDICINE AND BIOLOGY MAGAZINE
Research Center for Functional Brain Imaging, at Kennedy
Krieger Institute. He holds a B.S. in physics from the
Massachusetts Institute of Technology and a Ph.D. in
Biophysics from the University of Pennsylvania. After completing a postdoctoral fellowship in neuroimaging at the
National Institutes of Health (NIH), he was a senior staff
fellow in the Laboratory of Diagnostic Radiology Research
at the NIH. He subsequently became an assistant professor
in the Department of Neurology and a member of the
Institute for Cognitive and Computational Sciences at
Georgetown University, before joining Johns Hopkins. His
research focuses on the development of advanced methods
for the acquisition and analysis of magnetic resonance data
reporting on brain function; he has authored over 40 articles
on brain imaging technology.
Address for Correspondence: James J. Pekar, Ph.D., F.M.
Kirby Research Center for Functional Brain Imaging,
Kennedy Krieger Institute and The Russell H. Morgan
Department of Radiology and Radiological Science, Johns
Hopkins University, 707 North Broadway, Baltimore, MD
21205 USA. Phone: +1 443 923 9510. Fax: +1 443 923
9505. E-mail: [email protected].
References
[1] J.A. Jackson and W.H. Langham. “Whole-body NMR spectrometer,” Rev. Sci.
Instrum., vol. 39, no. 4, pp. 510–513, 1968.
[2] P.C. Lauterbur, “Image formation employing induced local interactions:
Examples employing nuclear magnetic resonance,” Nature, vol. 242, no. 5394, pp.
190–191, 1973.
[3] P. Mansfield and P.K. Grannell, “Diffraction and microscopy in solids and liquids by NMR,” Phys. Rev. B, vol. 12, no. 9, pp. 3618–3634, 1975.
[4] A. Kumar, A.D. Welti, and R.R. Ernst, “NMR Fourier zeugmatography,”
J. Magn. Reson., vol. 18, pp. 69–83, 1975.
[5] S. Ogawa, T.M. Lee, A.R. Kay, and D.W. Tank, “Brain magnetic resonance
imaging with contrast dependent on blood oxygenation,” Proc. Nat. Acad. Sci.
U.S.A., vol. 87, no. 24, pp. 9868–9872, 1990.
[6] S. Ogawa, D.W. Tank, R. Menon, J.M. Ellermann, S.G. Kim, H. Merkle, and
K. Ugurbil, “Intrinsic signal changes accompanying sensory stimulation:
Functional brain mapping with magnetic resonance imaging,” Proc. Nat. Acad.
Sci. U.S.A., vol. 89, no. 13, pp. 5951–5955, 1992.
[7] K.K. Kwong, J.W. Belliveau, D.A. Chesler, I.E. Goldberg, R.M. Weisskoff,
B.P. Poncelet, D.N. Kennedy, B.E. Hoppel, M.S. Cohen, R. Turner, H.-M. Cheng,
T.J. Brady, B.R. Rosen, “Dynamic magnetic resonance imaging of human brain
activity during primary sensory stimulation,” Proc. Nat. Acad. Sci. U.S.A., vol. 89,
no. 12, pp. 5675–5679, 1992.
[8] P.A. Bandettini, E.C. Wong, R.S. Tikofsky, R.S. Hinks, and J.S. Hyde, “Time
course EPI of human brain function during task activation,” Magn. Reson. Med.,
vol. 25, no. 2, pp. 390–397, 1992.
[9] C.S. Roy and C.S. Sherrington, “On the regulation of the blood-supply of the
brain,” J. Physiol., vol. 11, pp. 85–108, 1890.
[10] S.A. Huettel, A.W. Song, and G. McCarthy, Functional Magnetic Resonance
Imaging. Sunderland, MA: Sinauer Associates, 2004.
[11] B. Biswal, F.Z. Yetkin, V.M. Haughton, and J.S. Hyde, “Functional connectivity in the motor cortex of resting human brain using echo-planar MRI,” Magn.
Reson. Med., vol. 34, no. 2, pp. 537–41, 1995.
[12] S. Wakana, J. Jiang, L.M. Nagae-Poetscher, PCM van Zijl, and S. Mori,
“Fiber tract-based atlas of human white matter anatomy,” Radiol., vol. 230,
no. 1, pp. 77–87, 2004.
MARCH/APRIL 2006