Dear FSL experts,
I'm trying to compare DMN between healthy controls and patients. I want to
analyze my fMRI data in resting state as Greicius et al. 2007.
Data acquisition(resting state without any kind of task)
Philips 1.5T MR scanner, TR 2411, TE 40,
Flip angle 80, FOV 64*64, Matrix 64*64, Voxel size 3.0*3.0*3.6(gap=0),
35slices
200volumes
Preprocessing(SPM5)
realign: register to first image
no slice timing correction
normalize to EPI template
source image : one of EPI image
source weighting image: default setting
image to write : all of one subject's EPI data
template: EPI nii.(in SPM5)
smooth FWHM 6mm
4D>3D
component pick up(the goodness-of-fit)
one-sample t-test
two-sample t-test
I want to have any kinds of suggestions about following things.
1.3D>4D
I change my preprocessed 3D data into 4D with MRIcro. In MRIcro, this
reconstructed data is recognized as 200 volumes. But when I put this into
FSL MELODIC, it is recognized as only 1 volume.
I think the way to change dimension is wrong. How can I change my data into
4D appropriately?
2.the goodness-of-fit strategy
I realize this step consists of two steps. One is frequency filter step and
another is selecting component step with DMN mask.
I wonder how can I do this step. Is there a tool suitable for this step?
3.one-sample t-test
I wonder this step is needed to check whether each subject's 'best-fit' DMN
components distribute in gaussian distribution or not.
If this step is needed to check such kind of thing, do I have to do
following steps?
(i)GICA in each group
(ii)one-sample t-test between each subject's DMN thresh_zstat.nii(assigned
by the ICA of each subject) and each group's DMN thresh_zstat.nii(assigned
by the GICA of each group)
(iii)finally, if there are significant difference between one subject's
'best-fit' DMN component and each group's DMN component, do I habe to
discard the data?
Thank you in advance for your help. Any suggestion will be really appreciate.
Sincerely,
Yuki
|