Hi Yizhwan,
Is there a way to register the time 1 5TT image to the time 2 DWI image so I can run tractography?
You don’t want to register the 5TT image, as the 5TT image is inappropriate to be driving registration. You instead want to register the time 1 T1-weighted image to the time 2 DWI, and then use the computed rigid-body transformation to transform the 5TT image into alignment with the time 2 DWI (or the other way around). See terminology disambiguation here.
Although the transformation works very well (first image), the result of 5ttgen fsl is very strange- see attached (bottom image):
It’s hard to tell exactly what the complaint is. My first guess is that you’ve provided what looks like a skull-stripped image, but failed to include the -premasked option, which means that FSL bet will be executed on the data, and probably remove a lot of brain. The tissue decomposition also looks slightly unusual. But I’m not sure how much of this is image quality. Have you perhaps not just applied a rigid-body transformation to the T1-weighted image, but also resampled the T1-weighted image data onto the voxel grid of the DWI? I don’t think it’s of that low a resolution, but the T1-weighted image nevertheless looks somehow … artificial.
Rob