Many MRI steps only work if the scans are roughly aligned with the MNI template. Rotations (pitch) in particular is difficult to compensate for.
Sep 2020 Onwards
Approach: all scans get manually aligned
Scans are manually converted to nifti in the raw data folder.
Each scan is linked to a “study”, and this study is added to one or more projects.
At the project level, a person can choose manually reorient the scan. Specifically, they determine the 6 rigid-body parameters that will move the raw scan into a alignment with the template.
Those parameters are saved in the database, and they are associated with the study.
Rationale: One scan = one alignment. If the study is used in multiple projects, the same parameters will be reused. Additionally, if a second preprocessing is done, that will also use the same parameters.
CSPM will apply the parameters after the files are copied from the raw folder to the project preprocessing folder.
Each scan is aligned separately. However, it’s easy to copy the same parameters between scans (useful if one subject has 10 scans with the same rotation, for example).
Rationale: problems arise if we assume that one subject has the same rigid-body shifts for all scans. This approach is more time-consuming, but simple and reliable.
How-to
Example of new GUI
Up to Sep 2020
T1
Gets manually aligned in project folder Legacy - Manually Aligning T1's to MNI
Failed experiment:
All other scans get T1 coreg parameters applied
Rationale was that person in scanner would have similar orientation for all scans (e.g, 30 degree pitch for T1, fMRI, DTI, etc.). This is actually not the case genreally, as various scans have various orientations, and it was only a conincidence when fMRI and T1 rotations were the same. This is now basically disabled in CSPM unless you go through several hoops.
Non-T1 scans:
Manually adjust in project rawimages nifti folders