Hi Mark,
Thanks for responding. I checked activation maps in functional space, it's hard to tell whether activation is on grey matter or empty space, but I think I did see activation on gyri. I also checked head movement data, the maximum translation was 0.8mm, while our voxel size was 2.2mm, so I don't think head movement could cause those large errors (i.e. activation completely in empty space). I am attaching and functional image (example_func.nii.gz), activation maps (runX_tstat.nii.gz) and motion correction results (.png files) in this link (https://www.dropbox.com/sh/072e51o0m0bm09h/AABYkzV1I8ltFTq04U6SABZNa?dl=0), in case I missed anything.
As for registration, unfortunately we don't have a fieldmap. For now I am just using BBR in GUI, are there other options using command line that can potentially improve registration without a fieldmap?
On you point on motion correction, I think it is possible that MC is not optimal, but I didn't fully understand what you mean by a single-band reference image. I just used MCFLIRT in FEAT gui, and I think in that case the reference image is the middle volume, which was also acquired by multi-band sequence. Are there ways to improve motion correction? Should I consider running MELODIC ICA and denoise based on components revealed by the ICA analysis?
Thank you very much for your help,
Yuqi
|