Conference PaperPDF Available

Apparent Fiber Density: A novel method to detect axonal degeneration in patients with MS

Authors:

Abstract

Axonal degeneration is a key pathological driver of disability in MS. Diffusion-weighted MRI can non-invasively detect microstructural changes in white matter that are associated with axonal loss. A novel diffusion-weighted MRI measure, "apparent fiber density" (AFD) can be obtained from the fiber orientation distributions (FODs) computed by spherical deconvolution techniques. Using this approach it is possible to estimate differences in both Fiber Density (FD) and Fiber Cross-section (FC), for each fiber element (termed 'fixel') in each voxel. We sought to determine whether FD and/or FC differences exist specifically in the visual pathways in patients with a history of acute optic neuritis compared to healthy controls.
Apparent Fibre Density: A novel method to detect axonal degeneration in
patients with MS
Sanuji Gajamange1, David Raffelt2, Thijs Dhollander2, Elaine Lui3, Annie Shelton4, Owen White5, Trevor Kilpatrick1,2, Alan Connelly2, Joanne Fielding4, Scott Kolbe1
1Department of Anatomy and Neuroscience, University of Melbourne, Australia
2Florey Institute of Neuroscience and Mental Health, Melbourne, Australia
3Department of Radiology, Royal Melbourne Hospital, Melbourne, Australia
4School of Psychological Science and Institute for Cognitive and Clinical Neuroscience , Monash University, Australia
5Department of Neurology, Royal Melbourne Hospital, Melbourne, Australia
Multiple sclerosis is the most common neurological disorder affecting young adults. It is now believed that axonal degeneration is the principle cause of progressive disability. Treatments aiming to reduce or reverse axonal
loss require sensitive and specific markers of axonal loss.
Diffusion-weighted MRI can non-invasively detect microstructural changes in white matter (WM). A novel diffusion-weighted MRI measure, apparent fibre density can be obtained from the fibre orientation distributions
(FODs) computed by spherical deconvolution techniques (Raffelt 2012). Using this approach it is possible to estimate differences in both fibre Density (FD) and fibre Cross-section (FC), for each fibre element (termed ‘fixel’) in
each voxel.
We aimed to: (1) determine the sensitivity of FD compared to the standard Diffusion Tensor Imaging (DTI) approach, (2) determine whether fixel-specific differences exist
specifically in the visual pathway in patients with a history of optic neuritis, and (3) assess the pathological specificity of FD by comparing to electrophysiological measures of
axonal loss.
Introduction
MRI acquisition
Diffusion weighted scans were acquired for 17
patients with historical optic neuritis (disease
duration: 4.48 ± 0.61 years) and 14 healthy controls
(TR/TE = 7800/112ms; voxel size = 2.5x2.5x2.5mm3,
b=3000s/mm2, 7 non-diffusion and 60 directions
encoded scans). Structural T1 images (TR/TE =
1900/2.63ms; FA = 9°; voxel size = 0.8x0.8x0.8mm3)
were acquired for all subjects, along with double
inversion recovery (TR/TE = 7400/324ms; FA = 120°;
voxel size = 0.55x0.55x1.1mm3sagittal acquisition)
scans for patients.
Methods
Figure 1. A disease-induced change to the number of axons may
manifest as 1) a change to within-voxel fibre density 2) a
macroscopic difference in a fibre bundle’s cross-section 3) a
combination of both fibre density and bundle cross-sectional area.
Image processing
Implications of lesions in tractography
Effect size comparison between fibre density and fractional anisotropy
Conclusions
Characterisation of fibre density in white matter tissue
Figure 6. Regions of significant reductions in FD, FC, and FDC in MS patients compared to controls (FWER corrected p<0.05).
Fixel-specific differences between patients and controls
Functional and structural correlates of fixel-specific measures
Population average FOD template Fixel mask Population average tractogram
FOD fixel = fibre element
Healthy control
MS Patient
A. Severe lesion
Healthy control
MS Patient
B. Less severe lesion
Figure 7. Coefficients for Pearson correlation analyses between fixel-specific measures (FD and FC), and structural and functional
measures of optic nerve and brain injury. FD, being a marker of axonal degeneration, correlated significantly with white matter volume,
but did not correlate with cortical or subcortical grey matter volume. Values in red indicate p<0.05.
fibre density
fibre density
and
cross-section
fibre
cross-section
0
2
1
FA
FD
In this study we have shown that the pattern of FD varied in different tissue types,
and also demonstrated its increased sensitivity compared to DTI. Regions of
reduced FD and FC were detected in patients with early MS compared to healthy
controls that correlated with WM but not GM atrophy. Fixel based analysis of FD
and FC is a promising new tool for measuring axonal degeneration in MS.
Figure 4. A. Density plot of the Cohen’s d between patients and controls within WM
demonstrate that FD effect size is greater than FA effect size B. Visualisation of
Cohen’s d for controls > patients reveal that the effect detected by FD covers a wider
extent of WM than FA.
Figure 3. An assessment of FD and DTI metrics within T2 lesion, normal appearing WM of patients, and normal WM of controls showed
that the pattern of FD and FA followed the degree of WM pathology, while MD was not able to differentiate between NAWM and
control WM. ** = p<0.001, * = p<0.05
Figure 2. fibre orientation distributions
(FODs) were computed for each voxel.
A population-average FOD template
was generated from a subset of
subjects. The fibre elements (“fixel”) for
each fixel were derived from the FOD
template. For statistical comparison
between groups, template-based
tractography was performed to identify
structurally connected fixels (CFE).
Figure 5. Visualisation of FODs and tractography within lesions of an MS patient and a healthy control. A. Severe lesion. The fibre
orientational information and amplitude of FODs are markedly affected, leading to the tracks not being able to traverse the lesion. B.
Less severe lesion. The FOD amplitude is affected, but not reduced to zero, enabling tractography to be achieved with reduced fibre
tract density.
A. B.
Le s i
on v o
l
OCT
(a f
fect e
d ey e )
VEP a
m p
(af f
ecte d e
ye)
VEP l
at
(
af f e c
te d e
ye)
O C
T
( u n a f f e c t e d e y e )
V E P a m p
(u n a f
fe c t e
d e y e )
VEP l a t
( u
naffe c
ted e
ye)
Ven t
ri c l e
vo l
Co r t G M v o l
W M v o l
FD -0.359 -0.057 0.285 0.355 0.309 0.071 -0.230 -0.306 0.054 0.080 0.698
FC -0.224 -0.101 0.028 0.303 -0.033 0.229 -0.404 -0.208 -0.167 0.085 0.438
Statistical analysis
For fixel-specific measures, a population-average FOD template was generated from subject specific FOD
images obtained by a novel method that accounts for non-white matter tissues (Dhollander 2016). Each
subjects FOD image was then registered to the template, allowing for whole-brain fixel-based comparison
between patients and controls. The statistical analysis was performed using connectivity fixel enhancement
(CFE) (5000 permutations) to identify regions of reduced FD and FC (see figure 1 for schematic illustration)
(Raffelt 2015). For voxel-specific measures, DTI was performed to compute Fractional Anisotropy (FA) and
Mean Diffusivity (MD). All results were family wise error corrected for multiple comparisons.
References
Raffelt, D., et al, NeuroImage, (2012). 59(4): p. 3976-3994
Raffelt, D., et al, Neuroimage, (2015). 117: p. 40-55
Dhollander, T., Connelly, A., Proc. ISMRM (2016). 24: p. 3010
Reduced Fibre
Density & bundle
Cross-section (FDC)
Reduced
Fibre-bundle
Cross-section (FC)
Reduced
Fibre Density (FD)
b)
a) c)
Postgraduate scholarship supported by Multiple Sclerosis Research Australia
Funding
ResearchGate has not been able to resolve any citations for this publication.
  • D Raffelt
Raffelt, D., et al, NeuroImage, (2012). 59(4): p. 3976-3994
  • D Raffelt
Raffelt, D., et al, Neuroimage, (2015). 117: p. 40-55