Dear Daniel,
I suspect that what you see is a kind of “avalanche” of zeros as you progress through the analysis.
1. Eddy uses spline interpolation, so some voxels will have negative values after correction.
2. dtifit will interpret the negative values as NaN and any voxel that has a negative number in any volume will be set to zero.
3. flirt/fnirt will interpret zeros as NaN and when interpolating will set any voxels that falls within a 2x2x2 neighbourhood of a zero to zero.
4. When calculating the mean a zero in any subject will be seen as NaN (missing data) so the mean for that voxel will be zero.
I can’t remember for absolute certain how each of the tools treat zeros, but I _think_ that is the explanation.
I suggest you get rid of the negative values in the eddy corrected images using fslmaths. You can either use -abs to set those (very small) negative numbers to positive, or you can chose some very small positive number to replace them by.
Jesper
> On 3 Apr 2019, at 14:24, Daniel Callow <[log in to unmask]> wrote:
>
> Thank you for the reply Jesper,
>
> The holes are 0's when I look in fsleyes. I have disabled the -ero function in fsl based on previous discussions to see if that improved the issue but it still persists. I can't submit all of the screenshots I took due to the file size limit. But in general, I don't see many holes or black stripes in the eddy corrected data or a huge amount in the skull stripped FA images. The issue seems to be after the flirt or fnirt that is run. The main issue seems to be that my resulting FA_mean has giant holes in it. I have attached 1 of the FA images post warping and the mean_FA image as this seems to be the point where these holes occur or at least are accentuated.
>
> ########################################################################
>
> To unsubscribe from the FSL list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
> <Screen Shot 2019-04-03 at 9.23.02 AM.png><mean.png>
########################################################################
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
|