Ferran Prados1,2, Bhavana S Solanky2, Patricia Alves Da Mota2, Manuel Jorge Cardoso1, Wallace J Brownlee2, Frank Riemer3, David H Miller2, Xavier Golay4, Sebastien Ourselin1, and Claudia Angela Michela Gandini Wheeler-Kingshott2,5
1Translational Imaging Group, Medical Physics and Biomedical Engineering, University College London, London, United Kingdom, 2NMR Research Unit, Queen Square MS Centre, Department of Neuroinflammation, UCL Institute of Neurology, University College London, London, United Kingdom, 3Department of Radiology, University of Cambridge, Cambridge, United Kingdom, 4Brain Repair & Rehabilitation, UCL Institute of Neurology, University College London, London, United Kingdom, 5Brain Connectivity Center, C. Mondino National Neurological Institute, Pavia, Italy
Synopsis
Non-invasive measurement of in vivo total sodium concentration (TSC) has been possible due to advances in sodium MRI at clinical field strengths. Changes in white and grey matter concentrations have been reported in a number of different diseases like Multiple Sclerosis. However, the presence of regional differences in normal healthy brain TSC has not been yet investigated. Here we use Geodesic Information Flow technique for computing per subject brain parcellations to allow differentiation of these areas and subsequently characterization of regional TSC in healthy controls. Introduction
Non-invasive
measurement of in vivo total sodium concentration (TSC) has been
possible due to advances in sodium MRI at clinical field strengths.
Changes in white matter (WM) and grey matter (GM) concentrations have been reported in
a number of different diseases like, Multiple Sclerosis (MS). However, the presence of regional
differences in normal healthy brain TSC has not been yet
investigated. Here we use Geodesic Information Flow
1
(GIF) technique for computing per subject brain parcellations to
allow differentiation of these areas and subsequently
characterization of regional TSC in healthy controls.
Methods
Sodium
Protocol: 10
healthy controls were scanned on a 3T Philips Achieva and underwent
two protocols in the same session using: a 32-channel
head coil and a fixed
tuned sodium volume coil (Rapid Biomedical). The first protocol was
composed of PD/T2 (1x1x3 mm3)
and 3DT1-weighted (1x1x1 mm3)
MRI. The second 23Na-MRI protocol was a 3D-Cones UTE sequence2
(3x3x3 mm3) (see Figure 1) and a 1H
PD/T2w (1x1x3 mm3)
sequence. Two agar phantoms with 40 and 80 mM NaCl were placed either
side of the head for quantification.
Processing:
Probabilistic brain tissue segmentation was performed on the 3DT1
images using GIF (see Figure 2). The segmentation masks were registered
to the TSC maps. To preserve mask integrity, masks were resampled
to sodium space using a point-spread function3
with a linear interpolation. The resample step was performed using
the concatenation of three affine transformations from the
registrations of: 3DT1 to the pseudo-T1 images (PDw-T2w images4),
the T2w to the 1H
PD/T2w images, and finally 1H
PD-T2w image to the TSC map.
Partial
volume correction:
With the aim to remove
the CSF sodium contribution in the brain tissues, we applied a
voxel-by-voxel modified partition-based subject-specific correction5.
The average TSC for the voxels with 100% probability of being CSF
were computed. Then the final tissue sodium concentration was
expressed as
$$$C_{tissue}=\frac{(C_{voxel}-C_{CSF}*VF_{CSF})}{VF_{tissue}}$$$ where C represents TSC and VF
is the volume fraction provided by GIF. As in Paling et al., only
voxels with at least 20% tissue VF were included to produce a mask
threshold to 0.2 for each area. In order to reject unrealistically
high TSC values, a threshold was set to 2 standard deviations above
the mean of the TSC in voxels with at least 95% of tissue probability
(TSCmean_95%).
This was used as an additional threshold so that only voxels within
the 0.2 threshold mask, which also had TSC values within two standard
deviations of the TSCmean_95%,
were included.
Selected
brain areas: We
selected and compared the following brain areas from both the left
and right hemispheres: Frontal
Lobe, Occiptal Lobe, Sensorimotor Cortex,
Hippocampus, Limbic
System, Thalamus,
Cerebral White Matter,
Brainstem, Cerebellum
White Matter, Cerebellum Grey Matter and Cerebellar Vermis. Following this
the Sensorimotor Cortex
area was split into two groups depending on whether the subjects were
left or right handed.
Results
and Discussion
Table
1 shows the sodium concentration per brain tissue type, and Table 2
per brain area. Most of the brain regions have similar concentrations
in both hemispheres and differences are not significant. Only the
Occipital
Lobe
(visual areas) has significant differences between hemispheres. The
Sensorimotor
Cortex
for left handed subjects has significant differences from the TSC of
right-handed subjects, but this could be due to the lower sample
size. All left handed subjects have less sodium concentration in the
right hemisphere and vice versa with right handed subjects.
Conclusions
We
have presented the first regional sodium concentration study in human
healthy brain. Future work will analyse a larger cohort and will
explore the correlation with clinical scales for different diseases.
Acknowledgements
NIHR
BRC UCLH/UCL High Impact Initiative, EPSRC
(EP/H046410/1,EP/J020990/1,EP/K005278), MRC (MR/J01107X/1), UK MS
Society and Brain Research Trust.References
1)
Cardoso, IEEE-TMI, 2015; 2) Riemer, MAGMA, 2014; 3) Cardoso,
MICCAI 2015; 4) Hickman, MS Journal, 2002; 5) Paling, Brain, 2014