Hi,
1) If you run invwarp then you can get a warp from the standard space to the example_func space, which will do what you want.
2) If your melodic has run at 4mm resolution then you should have a reg/standard.nii.gz image in your ica results directory that will serve this purpose.
All the best,
Mark
> On 31 Oct 2014, at 15:48, Alle Meije Wink <[log in to unmask]> wrote:
>
> I have a number of fMRI time series pre-processed in the Melodic pipeline, including nonlinear registration with FNIRT and resampling at 4mm isotropic.
>
> For another analysis on the same time series, I need to warp atlas regions to the coordinate system of the fMRI input images.
>
> Atlas regions (I am using the non-lateralised Harvard-Oxford cortical atlas) can be saved to NifTI images on images loaded as 'open standard', but not on the 'standard.nii.gz' images in the MELODIC directories.
>
> So I have warps that were used for producing images sampled at 4mm and atlases sampled at (coarsest level) 2mm.
>
> For fMRI data processed by Melodic at the standard resolution, it seems that warping atlas regions to the original input space requires either
> a. re-running Melodic at a resolution of 2mm or finer
> (which not everyone has the computer memory to do)
> b. extracting the region at {.5,1,2} mm and then resample with 'flirt ... -appliyisoxfm 4' / 'fslmaths ... -subsamp2'
> (if I remember well the 1st works best)
>
> My questions are:
> 1. Is there a way to use the Melodic-produced warps 'example_func2standard_warp.nii.gz' to produce atlases in the original fMRI coordinate system?
> 2. Would it be handy to have a 'MNI152_T1_4mm_brain' in the 'data/standard' directory to extract 4mm atlas regions to use with Melodic-produced warps?
>
> Many thanks
> Alle Meije Wink
|