Closed jcohenadad closed 2 months ago
No other reason other than I had not thought of that/haven't done it before.
While working on using TFL labels only, I noticed that for some sites, eg NTNU, the qform-based registration is bad:
EDIT 20240627_124845: False alarm: the three scans are properly registered (meaning: the phantom did not move within the session):
Which means that the single-label approach is valid and we just need to fix the label for the NTNU-TFL (related to #81)
https://github.com/spinal-cord-7t/coil-qc-code/blob/da016b61566d5e8f4a00d700e82a921bed167e25/data_processing-phantom.ipynb#L316-L344
Assuming the phantom was not moved between the acquisition of the TFL, DREAM and SNR, there is no need to redefine the labels, given that using the qform we can put the label from the TFL space into the space of the DREAM and SNR space.
We should simplify the code and use only one set of pre-defined labels from one single contrast, unless there was another reason for creating labels for the DREAM and SNR images @evaalonsoortiz ?
Related to #81 #82