Dear FSL experts,
We performed a whole-brain diffusion MRI study to compare changes between two cognitively healthy groups with different risk for Alzheimer’s disease, APOE E4 (high risk) carriers and non carriers (low risk) (352 subjects in total). We have run TBSS analysis (FSL 5.0.5) to test group differences in FA, MD, MO, L1 and RD. We have found significant results with MO at the superior longitudinal fasciculus (SLF). Acording to [Douaud_2011], we should also see co-localized MO and FA differences, but we haven’t got the expected results, even looking at the uncorrected p image.
As the obtained results are located in a crossing fibre region, we also conducted an analysis using crossing-fibre measures [jbabdi_2010], but we didn’t get any result in SLF, neither with f1 nor with f2.
Both [Douaud_2011, jbabdi_2010] report the limitations of TBSS skeleton in crossing fibre regions, so I was wondering if we should perform voxel-wise, as opposed to “skeletonised”, MO, FA, f1 and f2 in these regions as Doaud et al., did for MO and FA in [Douaud_2011]. That is, take the MO, FA, f1 and f2 maps registered to the standard space and convolute them with a Gaussian kernel (sigma=1mm).
The steps I think I should follow to do this are:
tbss_1_preproc *.nii.gz
tbss_2_reg -n
tbss_3_postreg -S
# The following steps are common for FA and MO. I don’t know how to do it for f1 and f2 because tbss_x only generates “all_F1_x_skeletonised.nii.gz” and “all_F2_x_skeletonised.nii.gz”
fslmaths all_FA.nii.gz -kernel gauss 1 -fmean all_FA_smooth.nii.gz
randomise -i all_FA_smooth.nii.gz -o outputfile -m mean_FA_mask -d design.mat -t design.con -n 5000 --T2
Are these steps correct? How could I do it for f1 and f2? Would it improve the previously obtained results and explore the expected co-localization of MO and FA?, Is it possible that this co-localization would not be found as we are dealing with HEALTHY subjects who are classified only on a risk basis? Is the MO finding valuable on its own?
As for the initial TBSS analysis, we followed the TBSS user guide steps to preprocess all indices and do the statistical inference, as follows:
For FA:
tbss_1_preproc *.nii.gz
tbss_2_reg -n
tbss_3_postreg -S
tbss_4_prestats 0.2
randomise -i all_FA_skeletonised.nii.gz -o outputfile -m mean_FA_skeleton_mask -d design.mat -t design.con -n 5000 --T2
Using non-FA images (for MD, MO, L1 and RD):
tbss_1_preproc -f 1000000 *.nii.gz
tbss_non_FA MD
randomise -i all_MD_skeletonised.nii.gz -o outputfile -m mean_FA_skeleton_mask -d design.mat -t design.con -n 5000 --T2
Using crossing-fibre measures:
tbss_x F1 F2 D1 D2
randomise -i all_F1_x_skeletonised.nii.gz -o outputfile -m mean_FA_skeleton_mask -d design.mat -t design.con -n 5000 --T2
randomise -i all_F2_x_skeletonised.nii.gz -o outputfile -m mean_FA_skeleton_mask -d design.mat -t design.con -n 5000 –T2
MRI ACQUISITION:
Data were collected on a 3 T in combination with a 32-channel head coil, voxel size: 1.7 mm isotropic, 64 diffusion directions with b-value 1000 s/mm2, and 1 image with no diffusion weighting.
REFERENCES:
[Douaud_2011] Douaud, G., Jbabdi, S., Behrens, T. E., Menke, R. A., Gass, A., Monsch, A. U., ... & Smith, S. (2011). DTI measures in crossing-fibre areas: Increased diffusion anisotropy reveals early white matter alteration in MCI and mild Alzheimer's disease. Neuroimage, 55(3), 880-890.
[jbabdi_2010] Jbabdi, S., Behrens, T. E., & Smith, S. M. (2010). Crossing fibres in tract-based spatial statistics. Neuroimage, 49(1), 249-256.
Thank you and best regards,
Maite García-Sebastian.
|