Dear Group,
I collected an axial flair that consisted of interleaved odd and even slices, 2 mm thick.
Unfortunately, the subject moved between odd and even sets and we now see
horrible stepping in the front of the sagittal image where the movement was most pronounced
(and where it is easiest to see the misalignment of axial slices)
I thought to realign the odds and evens, but my technique (shown below) does not give impressive results.
Can anyone suggest a better approach, or is this series a loss?
Thankyou,
Dianne
================
1) First, I separate the odd and even dicom images into separate folders and use MRIconvert to create each nifti volume
(the even volume and the odd volume)
2) FSLview thinks that the slices in the odd and even volumes are twice as thick as the original acquisition (4 mm instead of 2 mm). I suppose there is nothing to do about this.
3) Use flirt to realign odd and even flair series damaged by movement (align odds to evens):
/usr/local/fsl/bin/flirt -in /Exps/Data/dti_erp/e3393/3393_4_odd.nii -ref /Exps/Data/dti_erp/e3393/3393_4_even.nii -out 3393_odd_reg -omat 3393_odd_reg.mat -bins 256 -cost corratio -searchrx -90 90 -searchry -90 90 -searchrz -90 90 -dof 6 -interp trilinear
dpat% ls
3393_4_even.nii 3393_odd_reg.mat 3393_odd_reg.nii.gz
4) Now, recombine the odd and even volumes:
dpat% avwinterleave 3393_4_even.nii 3393_odd_reg.nii.gz newflair
Unsurprisingly, the recombined image (newflair) still has 4 mm thick slices, so it is a very tall head.
5) Correct the slice thickness by copying the geometry of my original flair (terrible alignment but correct slice thickness) to my newly realigned and re-interleaved flair.
dpat% avwcpgeom 3393_4.nii newflair.nii.gz
6) I compare newflair.nii.gz to the original (3393_4.nii) and I see a little change, but it is not consistently better
looking. e.g., now the coronal looks rougher, the bad sagittal slice looks smoother (poorer resolution), though perhaps the slices are better aligned now?? In both cases, the axials look okay, since those individual slices were fine.
Is there any way to save data like this?
Thankyou for your time,
Dianne
--
Dianne Patterson, Ph.D.
[log in to unmask]
ERP Lab
University of Arizona
621-3256 (Office)
