I have a question about mrcat: if I try to concatenate two DW datafiles that have different qform matrices (such that they are both aligned with the same whole-head T1-weighted dataset), does mrcat take into account those matrices and reslice one of the data sets? or is one of the two qforms ignored?
Is it better to first reslice one of the 2 datasets so that both have the same qforms before concatenation?
mrcat concatenates the data without taking the different header transformations into consideration, it uses the first image’s transformation for the output. So you’ll need to regrid both images onto the same reference grid (voxel and transformation) before concatenation with
mrtransform a.mif -template reference.mif a_regridded.mif.
reference.mif can be either image or the unbiased average (
mraverageheader a.mif b.mif reference.mif) if you want to avoid different degrees of interpolation between the images. To avoid double interpolation, make sure that the two DW images are transformed without regridding (the equivalent of
mrtransform -linear trafo.txt without
-template) and specify the target resolution in
For concatenating multiple DWI series accounting for differential intensity scaling, have a look at dwicat in the
Very comprehensive answer, thanks! I will try your suggestions.