Munchen: Correction of L/R DiscrepancyPosted By: Maarten Mennes - Apr 6, 2010
Tool/Resource: 1000 Functional Connectomes Project
Together with the PIs of the Munchen dataset we managed to trace the left/right orientation discrepancy between the MPRAGE and EPI in the Munchen dataset to a L/R flip of the EPI during conversion of SPM analyze to nifti. All subjects were affected by this issue.
As of today, APRIL 6 2010, we replaced the Munchen dataset with the correct one.
If you are using this dataset, your best option is to download the dataset again and process them from scratch. Alternatively, you could choose to L/R flip the EPI scan (rest.nii.gz or your specific end product) using the following commands. Remind yourself to rerun registration to the MPRAGE and standard space afterwards.
3drefit –orient LPI rest.nii.gz
3dresample –orient RPI –prefix temp.nii.gz –inset rest.nii.gz
mv temp.nii.gz rest.nii.gz
fslswapdim rest.nii.gz -x y z rest.nii.gz
But again, if you have any doubt about the correctness of your Munchen data, please download and reprocess them again.
By correcting the Munchen dataset we have now resolved all L/R discrepancy issues as brought to our attention by users of the data. We hope to avoid such issues in the future by asking new contributors (and we have some in the works!) to verify the L/R orientation of their data prior to release.
However, these L/R issues have indicated the need for a discussion on how to quality control a multitude of datasets that is not necessarily collected by the person analyzing them. We hope to start such a discussion over the course of the next weeks on the forums. As always, any input is valued!