1
Multimodal Neuroimaging of Corticobasal Syndrome McMillan et al
e-Appendix
Cortical Thickness Imaging Analysis
T1 image preprocessing was performed using a previously reported pipeline implemented in Advanced Normalization Tools (ANTs)1. We first perform N4 bias correction of all images to minimize image inhomogeneity effects that could otherwise yield artificial intensity variation and bias segmentation priors 1,2. We then perform brain extraction using a combination of template-based and segmentation strategies. This involves registering a dilated template brain to each individual subject brain that can then be used to guide brain segmentation from the full MRI volume. We then perform Atropos six-tissue class (cortex, deep grey, brainstem, cerebellum, white matter, and CSF/other) segmentation using an optimized combination of prior knowledge from N4 bias-correction and template-based priors to guide the segmentation process 2,3. We then calculate voxelwise measurements of cortical thickness4. Finally, to make group-level inferences we use a diffeomorphic and symmetric registration algorithm to warp each cortical thickness map to a custom template. The template used in this study, The Penn Neurodegenerative Disease Template, was comprised of 115 controls and 93 neurodegenerative disease patients (PD, AD, amyotrophic lateral sclerosis, and frontotemporal degeneration) who are demographically-comparable to the imaging series in the current study. The resulting cortical thickness images were then downsampled to 2mm isotropic voxels and smoothed using a 4mm FWHM Gaussian kernel.
Diffusion Tensor Imaging Acquisition and Analysis
Diffusion-weighted images were preprocessed using ANTS software. Briefly, the unweighted (b=0) images are first extracted and averaged. All DW images (including the individual b=0 volumes) are then aligned to the average b=0 using ANTs1,3. An affine transform is applied to capture eddy distortion in the DW images as well as motion. Diffusion tensors are computed using a weighted linear least squares algorithm in Camino1,5. The corrected average b=0 image is aligned to the subject's T1 image from the same scanning session, first rigidly to correct for motion, then using a deformable diffeomorphic transformation with mutual information to correct for inter-modality distortion. The diffusion to T1 warp is composed with the T1 to template warp (from the cortical thickness pipeline), producing a mapping from DWI space to the population T1 template in a single interpolation. Tensors are resampled into the template space using log-Euclidean interpolation5,6 and reoriented to preserve the anatomical alignment of white matter tracts6,7. We report fractional anisotropy (FA) that was computed on the tensor image in the group analysis template space. The resulting FA images were smoothed using a 4mm FWHM isotropic Gaussian kernel.
1.Tustison NJ, Cook PA, Klein A, et al. Large-scale evaluation of ANTs and FreeSurfer cortical thickness measurements. NeuroImage. 2014 Oct 1;99:166–179.
2.Tustison NJ, Avants BB, Cook PA, et al. N4ITK: improved N3 bias correction. IEEE Trans Med Imaging. 2010 Jun;29(6):1310–1320.
3.Avants BB, Tustison NJ, Wu J, Cook PA, Gee JC. An open source multivariate framework for n-tissue segmentation with evaluation on public data. Neuroinformatics. 2011 Dec;9(4):381–400.
4.Das SR, Avants BB, Grossman M, Gee JC. Registration based cortical thickness measurement. NeuroImage. 2009 Apr 15;45(3):867–879.
5.Cook P, Bai Y, Nedjati-Gilani S. Camino: Open-source diffusion-MRI reconstruction and processing. International Society for Magnetic Resonance Imaging in Medicine. 2006;:2759.
6.Arsigny V, Fillard P, Pennec X, Ayache N. Log-Euclidean metrics for fast and simple calculus on diffusion tensors. Magn Reson Med. 2006 Aug;56(2):411–421.
7.Alexander DC, Pierpaoli C, Basser PJ, Gee JC. Spatial transformations of diffusion tensor magnetic resonance images. IEEE Trans Med Imaging. 2001 Nov;20(11):1131–1139.
Table e-1. Regions of reduced grey matter density (GM) and reduced fractional anisotropy (FA) in white matter (WM) for CBS-nonAD and CBS-AD relative to healthy controls.
Neuroanatomic Region (Brodmann Area)
L/R
Peak Voxel
MNI Coordinates
X Y Z
Peak Voxel
P-Value
Cluster Size (voxels)
Grey Matter Cortical Thickness: CBS-nonAD < Eld
Inferior parietal (40)
R
60
-36
46
0.001
477
Inferior parietal (40)
L
-50
-44
54
0.001
121
Primary motor (4)
L
-44
-2
50
0.002
140
Supplementary motor (6)
R
40
0
50
0.001
150
Grey Matter Cortical Thickness: CBS-AD < Eld
Inferior parietal (40)
R
56
-34
52
0.001
753
Primary motor (4)
L
-52
-2
50
0.001
535
Ventral striatum
L
-22
-6
-14
0.001
474
Ventral striatum
R
28
-2
-24
0.001
152
Dorsomedial prefrontal (9)
L
-38
44
26
0.001
130
Fusiform (37), middle temporal (21)
L
-46
-48
-14
0.001
7366
Fusiform (37)
L
-32
-60
-16
0.001
654
Fusiform gyrus (37)
R
28
-54
-16
0.001
187
Middle temporal gyrus (21)
R
62
-56
-6
0.001
382
Posterior cingulate (31)
--
-8
-54
28
0.001
683
Extrastriate (19)
R
30
-72
28
0.001
152
White Matter Fractional Anisotropy: CBS-nonAD < Eld
Superior longitudinal fasciculus
L
-53
1
24
0.001
525
Superior longitudinal fasciculus
L
-45
-20
38
0.001
3201
Superior longitudinal fasciculus
R
32
-26
23
0.001
150
Corpus callosum
--
-8
-8
28
0.001
39101
Corpus callosum
--
-7
14
59
0.001
352
Corpus callosum
--
2
24
-3
0.001
507
Corpus callosum
L
-27
10
41
0.001
251
Corpus callosum
L
-12
58
1
0.001
277
Corpus callosum
R
27
-64
26
0.001
1318
Corpus callosum
R
9
29
-17
0.001
442
Corpus callosum
R
16
37
19
0.001
126
Corticospinal tract
L
-8
-27
-43
0.002
111
Cingulum
L
-21
-40
-4
0.001
1014
Middle cerebellar peduncle
R
26
-37
-33
0.001
830
Inferior frontal-occipital fasciculus
L
-37
-55
22
0.001
124
Inferior frontal-occipital fasciculus
L
-45
-32
2
0.001
440
Inferior frontal-occipital fasciculus
L
-33
-14
5
0.001
715
Inferior frontal-occipital fasciculus
L
-40
33
5
0.001
424
Inferior frontal-occipital fasciculus
R
25
5
-13
0.001
203
Inferior frontal-occipital fasciculus
R
37
38
5
0.001
136
Inferior frontal-occipital fasciculus
R
27
42
-8
0.001
786
Inferior longitudinal fasciculus
L
-46
-12
-13
0.002
142
Inferior longitudinal fasciculus
L
-44
-7
-38
0.002
103
Inferior longitudinal fasciculus
R
28
-45
-15
0.001
273
Inferior longitudinal fasciculus
R
48
-9
-7
0.001
4007
Anterior thalamic radiation
L
-2
-7
13
0.001
338
Anterior thalamic radiation
L
-33
11
15
0.001
326
Anterior thalamic radiation
R
4
-7
9
0.001
11338
White Matter Fractional Anisotropy: CBS-AD < Eld
Cingulum
L
-25
-30
-7
0.006
104
Cingulum
R
28
-30
-6
0.003
148
Cingulum
R
28
-23
-21
0.001
349
Corpus callosum
L
-34
-75
22
0.001
145
Corpus callosum
L
-20
-64
30
0.001
632
Corpus callosum
L
-20
-54
16
0.002
113
Corpus callosum
L
-21
-45
35
0.001
1065
Corpus callosum
R
28
-63
28
0.001
128
Globus pallidus / ventral striatum
R
20
-3
-5
0.001
167
Inferior longitudinal fasciculus
L
-45
-8
-33
0.001
561
Inferior longitudinal fasciculus
R
49
-7
-7
0.001
151
Superior longitudinal fasciculus
L
-44
-40
29
0.001
130
Figure e-1. Multimodal neuroimaging of corticobasal syndrome (CBS) patients with evidence of Alzheimer’s disease (CBS-AD) or not (CBS-nonAD) relative to healthy seniors: left images illustrate reduced cortical thickness for CBS-AD relative to controls (red) and CBS-nonAD relative to controls (blue); right images illustrate reduced white matter fractional anisotropy for CBS-AD relative to controls and CBS-nonAD relative to controls (yellow regions) overlayed on red/green/blue (RGB) images illustrating average diffusion direction (red=left/right; green=anterior/posterior; blue=superior/inferior).