You can position your opening statement here, either in Connexions Purple or reversedout'

1
Arterial blood volume changes measured using ASL
in humans. C.L. Hoad, S.T. Francis, P.A. Gowland
Sir Peter Mansfield Magnetic Resonance Centre,
School of Physics and Astronomy, University of
Nottingham, Nottingham, UK.
INTRODUCTION Original haemodynamic models of the
BOLD effect 1 assumed that on neuronal
activation blood volume changes were localised to
the venous compartment. It is now becoming
accepted that CBV changes also occur on the
arterial side with changes in flow being
accommodated on the venous side by increasing
velocity rather than simply ballooning of the
vessels 2. Currently no non-invasive, direct
measurements of arterial CBV changes have been
made in humans. This study uses the EPISTAR ASL
technique with manipulation of diffusion
weighting (DW) to measure resting cerebral blood
flow (CBF), arteriole cerebral blood volume
(aCBV) and transit times, and the changes induced
in these parameters during a motor task study.
METHODS Image Acquisition Experiments were
performed on 7 healthy volunteers using a 3.0 T
scanner. A 5-slice EPISTAR sequence was
implemented with TR 2.5 s and TE 35 ms five
transaxial slices (64 x 64 matrix, 4 x 3 x 6 mm3
resolution, 1.25 kHz switching frequency) were
acquired in approximately 300 ms, with the most
superior slice being acquired first. A 15 mm gap
was allowed between the tag and control slabs and
their adjacent imaging slices. Immediately prior
to the inversion pulse, a presaturation slab was
applied to the imaging slab to overcome the
effects of imperfect inversion profiles.
Multislice images were acquired with either no
diffusion weighting (NDW), retaining arterial
blood volume signal, or with diffusion weighting
(DW) using a bipolar gradient pulse with first
order gradient moment of 1.5 radm-1s to suppress
the signal from arteriole blood. Following the
activation study, T1 maps were acquired using a
non-selective inversion recovery sequence with 13
TIs (50-4000 ms), TR 10 s.
Paradigm A two-handed finger opposition task
(paced by a visual cue) was employed, which
consisted of 60 s ON and 60 s rest period. Each
experiment consisted of either a DW or NDW
acquisition at a single inversion time (TI) as
outlined in Table 1. Each state of activation or
rest consisted of 6 pairs of control-tag images
the number of cycles acquired for the activation
and rest conditions being given in Table 1. Where
both DW and NDW data was acquired for a given TI,
interleaved acquisition was employed, with TI s
being acquired in a random order. Analysis
Images were realigned using a 2D rigid body model
(to allow for signal recovery through the
multislice set due to presaturation effects).
Tag and control images were subtracted to yield
difference (?M) images at each TI, and T1 maps
generated for each multislice image. To identify
areas of activity, ?M (perfusion) images and
control (BOLD) TI files were concatenated into
early and late time points as shown in Table 2.
Control data were normalised to the baseline rest
level at each TI (to account for the saturation
recovery). All data was then spatially filtered
with a Gaussian of FWHM of 5 mm, temporally
smoothed by convolving with a Gaussian of 3 s
(FWHM), and high-pass filtered to 0.005 Hz.
Early and late, DW and NDW, perfusion data were
then correlated with a haemodynamic response
function (HRF) and statistical maps thresholded
at p lt 0.05 (corrected) to provide ROIs for
subsequent analysis. Model A compartmental
model to numerically monitor the magnetisation in
the non-exchanging arteriole compartment and
exchanging tissue compartment and calculate CBF
and aCBV ( of voxel containing arterial blood)
at rest and on activation was formed. Following
the initial transit time for the tag to reach the
imaging plane, Da, the exchange of tagged water
between the blood and tissue does not occur
immediately. Water molecules remain in the
arteriole compartment for a further exchange
time, tex. After a time Dc, tagged blood reaches
the exchange site in the capillary bed.
Initially the DW (?M) images (assumed to contain
negligible non-exchange arteriole component) were
fitted to the modified Bloch equation with
exchange where f indicates perfusion to the
tissue (CBF) To improve upon a plug flow
approximation and allow for a distribution of
arrival times of the tag, the input function of
the tag was modelled to be the convolution of the
tag duration in the capillary compartment,
tdurat, and a Poisson function with time to peak
(TTP) 200 ms, the initial arrival time occurring
at Dc. Parameters f , tdurat, and Dc were fitted
for both rest and activation conditions. To
measure aCBV, the NDW difference (?M) signals
were first adjusted for perfusion effects by
subtracting any contribution from perfusion (f)
changes using the parameter estimates calculated
above. The adjusted NDW difference (DM) signals
then arose solely from the arteriole compartment.
These adjusted ?M signals were then fitted to a
modified Bloch equation assuming no exchange,
where F indicates flow in the arteriole A
distribution of tag arrival and outflow times was
modelled by convolving the tag duration through
the arteriole, texc, with a poisson function of
TTP 200 ms, with an initial arrival time of Da
Parameters F, texc, and Da were fitted for both
rest and activation conditions, and aCBV
calculated from F texc. Statistics The changes
in the calculated parameters on activation were
compared using a Wilcoxon test in SPSS.
Table 3. Fitted parameters for arteriole and
tissue compartments.
Figure 1 ROIs defined from the early perfusion
(NDW) data with the corresponding arteriole and
tissue plots at rest and on activation.
RESULTS One subject showed no BOLD response.
Figure 1 shows the activated regions found on the
early perfusion (NDW) dataset for one subject,
with the corresponding arterial and capillary
component data on rest and activation. The fitted
parameters for a region of interest in the
primary motor area for all subjects are shown in
Table 3. A significant increase in aCBV is seen
on activation. No significant change in the other
measured parameters was found. This may arise due
to the ROIs being selected from the early (NDW)
data.
DISCUSSION AND CONCLUSION We have measured a
change in arterial cerebral blood volume on
neuronal activation, the arterial blood volume
being approximately 0.6 on rest, increasing to
0.8 on activation. Data such as these will be
invaluable in furthering our understanding of the
basis of the BOLD effect and also for studying
the effects of some pharmacological
interventions. Further work will investigate
areas activated in the late NDW, DW, and BOLD
data sets.
REFERENCES 1. R.B. Buxton, et al., MRM, 39,
855-864, 1998. 2. S-P. Lee, et al., MRM,
45, 791-800, 2001. 3. A. Sleigh, et al.,
Proc. Intl. Soc. Mag. Res. Med. 10, 1058, 2002.
4. Duong, Magn Res Med, 43, 393-402 2000.
5. H An and W Lin, Mag. Res. Med., 47, 958-966,
2002. ACKNOWLEDGEMENTS This study was funded by
the Medical Research Council, UK