Hi,
What version of FSL are you running? If your original input data has
NaN's in it you can use
fslmaths <input_filename> -nan <input_filename>
to overwrite the file without NaN's
Regards
Matthew
> Dear Bethany,
>
> NaN's creeping in from the edge after smoothing suggests that non-brain
> voxels have the value of NaN. Have you checked the data going into the
> 1st
> level? Are you sure there are no NaN's in it? You could do
> fslmaths 4d_functionals -Tmean functional_mean
> and then check the 3D mean image for NaN's with fslview.
>
> Otherwise, I'm not sure what could be causing the trouble.
>
> -Tom
>
>
> On Wed, Jul 2, 2008 at 10:06 PM, Bethany Weber <[log in to unmask]>
> wrote:
>
>> Hi --
>>
>> I'm working on analysis: we're collecting 4-6 runs per subject,
>> event-related design, collapsing across runs at the second level and
>> across
>> subjects at the third level.
>>
>> I've been having a problem with NaNs appearing in my data starting after
>> my
>> second-level, and going on to produce errors at the third level. The
>> number
>> of NaNs in the second-level cope1.nii.gz files is always the same for
>> any
>> given subject, both for all the copes in my current analysis and even
>> across
>> analyses. The NaNs always occur around the edge of the brain.
>>
>> We think the problem may be related to something happening during
>> smoothing.
>> In the second level log file we're getting lines that look like this:
>>
>>
>> Post-stats
>>
>> /usr/local/fsl/bin/fslmaths stats/zstat1 -mas mask thresh_zstat1
>>
>> echo 309240 > thresh_zstat1.vol
>> zstat1: DLH=nan VOLUME=309240 RESELS=nan
>>
>> /usr/local/fsl/bin/cluster -i thresh_zstat1 -c stats/cope1 -t 2.3 -p
>> 0.05
>> -d
>> nan --volume=309240 --othresh=thresh_zstat1 -o cluster_mask_zstat1
>> --connectivity=26 --mm --olmax=lmax_zstat1_std.txt >
>> cluster_zstat1_std.txt
>>
>>
>> We're not sure where those NaNs are coming from since these same numbers
>> look normal for all runs in the first-level log files, e.g.:
>>
>> echo 26833 > thresh_zstat1.vol
>> zstat1: DLH=0.641048 VOLUME=26833 RESELS=7.21775
>>
>>
>> Does anyone have any ideas what might be happening?
>>
>> -Bethany Weber
>>
>>
>
>
> --
> ____________________________________________
> Thomas Nichols, PhD
> Director, Modelling & Genetics
> GlaxoSmithKline Clinical Imaging Centre
>
> Senior Research Fellow
> Oxford University FMRIB Centre
>
|