Dear Richard.
I cannot be 100% certain, but it looks a little like a problem I have seen before. It has to do with choosing a diffusion weighted volume in which there has been no intra-volume movement as a “shape reference”, i.e. a volume where it trusts the shape not to have been distorted by movement. In doing so it uses a set of heuristics, and one of those is that there can’t be any outliers in that volume, since those would be an indication of movement.
I have seen cases where it cannot find a single volume without “outliers”, and that has then been cause not by movement but by slight timing issues in the sequence which has caused one slice (or group of slices) to have a slightly lower intensity than expected. You can typically tell by looking at the .eddy_outlier_n_stdev_maps if you have this problem or not.
If you do, you should talk to your physicists about a long term solution. Short term, you can increase the --ol_nstd parameter to maybe 5 or 6 to make it more likely that you see at least one volume without outliers.
Jesper
> On 31 Jan 2019, at 12:11, Richard Rees <[log in to unmask]> wrote:
>
> Dear all,
>
> I am processing multi shell DWI data, using FSL tools. I originally used topup, bet and then 'eddy' (5.0.10). I have run this on several data sets and (although slow on my laptop) it worked. However, I want to use the 5.0.11 version to gain the intravolume motion correction.
>
> Using the same dataset, I have been unable to make eddy_cuda or eddy_openmp work: I'm using the patch version of eddy_cuda8.0, using the university cluster, including GPU for the cuda version.
>
> Here is my script:
>
> eddy_cuda8.0 --imain=$DATA/01/test/01_combined.nii.gz --mask=$DATA/01/test/01_mask_mask.nii --acqp=$DATA/acqparamstopup.txt --index=$DATA/index.txt --bvecs=$DATA/combined.bvec --bvals=$DATA/combined.bval --data_is_shelled --topup=$DATA/01/test/01_topupout --out=$DATA/01/test/testeddy --mporder=6 --slspec=$DATA/slspec.txt --s2v_niter=5 --s2v_lambda=1 --s2v_interp=trilinear
>
> I have also tried with eddy_openmp, and tried both with and without the final flags (from --mporder to the end).
>
> With each trial I get the same error:
>
> Running sm.ApplyDWILocationReference
> eddy: msg=ECScanManager::GetGlobal2DWIIndexMapping: Global index not dwi
> terminate called after throwing an instance of 'EDDY::EddyException'
> what(): eddy: msg=ECScanManager::GetGlobal2DWIIndexMapping: Global index not dwi
> Aborted
>
> I have been unable to find any other posts regarding this error, and given that the deprecated version of eddy was able to run (with exactly the same inputs) I'm at a loss with this.
>
> Comments greatly appreciated.
>
> Richard Rees
>
> ########################################################################
>
> To unsubscribe from the FSL list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
########################################################################
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
|