Closed kousu closed 1 year ago
Here's the distribution of how we've used this field:
u108545@joplin:~/data-multi-subject$ ls */*/*acq-*.nii.gz | sed 's/.*\(acq-.*\)_.*/\1/' | sort | uniq -c
22 acq-b0
259 acq-MToff
259 acq-MTon
254 acq-T1w
So, I know what I want to do with MToff
and MTon
. Do we need the other two?
We can get rid of acq-T1w
by renaming those files to match how most files in this dataset tagged T1w
do it, putting it into the final label: _T1w.nii.gz
:
And acq-b0
is in a funny situation; a bunch of the existing subjects encode that piece of information in the subject name itself, which isn't a solution:
Same issue, different dataset: https://github.com/neuropoly/data-management/issues/147
We can get rid of acq-T1w by renaming those files to match how most files in this dataset tagged T1w do it, putting it into the final label: _T1w.nii.gz
no because this T1w file is intended for MTS computation (hence the "_MTS" intended for suffix)
And acq-b0 is in a funny situation; a bunch of the existing subjects encode that piece of information in the subject name itself, which isn't a solution:
aouch-- that's quite ugly indeed. Probably a mistake
We can get rid of acq-T1w by renaming those files to match how most files in this dataset tagged T1w do it, putting it into the final label: _T1w.nii.gz
no because this T1w file is intended for MTS computation (hence the "_MTS" intended for suffix)
I don't have the background to understand this sentence. I'm sorry. But I can tell there's some conflict here because BIDS defines both _MTS
and _T1w
as suffixes:
and I don't think BIDS allows more than one suffix.
I can disambiguate: the filename spec says that the only filenames with _MTS
in them must also have _mt-
in them. It's not optional:
sub-<label>[_ses-<label>][_acq-<label>][_ce-<label>][_rec-<label>][_run-<index>][_echo-<index>]_flip-<index>_mt-<on|off>[_part-<mag|phase|real|imag>]_MTS.json
sub-<label>[_ses-<label>][_acq-<label>][_ce-<label>][_rec-<label>][_run-<index>][_echo-<index>]_flip-<index>_mt-<on|off>[_part-<mag|phase|real|imag>]_MTS.nii[.gz]
so e.g. .sub-nottwil04_acq-T1w_MTS.nii.gz
does not fit the template. Only the (currently: acq-MTon
) mt-on
and (currently: acq-MToff
) mt-off
files can have _MTS
suffixes. That third file has to go somewhere else.
u108545@joplin:~/data-multi-subject$ ls -l ./sub-nottwil04/anat/*acq-*.nii.gz
-rw-r--r-- 2 u108545 domain users 1511577 Nov 15 19:06 ./sub-nottwil04/anat/sub-nottwil04_acq-MToff_MTS.nii.gz
-rw-r--r-- 2 u108545 domain users 1458987 Nov 15 19:06 ./sub-nottwil04/anat/sub-nottwil04_acq-MTon_MTS.nii.gz
-rw-r--r-- 2 u108545 domain users 1109815 Nov 15 19:06 ./sub-nottwil04/anat/sub-nottwil04_acq-T1w_MTS.nii.gz
Swapping the metadata around, e.g. to sub-nottwil04_acq-MTS_T1w.nii.gz
, at least fits the template since acq
is a relatively freeform field. But I'm not sure that makes any more sense, 'cuz I don't really have the MRI background to judge.
And acq-b0 is in a funny situation; a bunch of the existing subjects encode that piece of information in the subject name itself, which isn't a solution:
aouch-- that's quite ugly indeed. Probably a mistake
Nevermind about the subject IDs; some site IDs that we picked ("brnoUhb", "cmrrb", "oxfordFmrib") all happen to end in 'b'.
But as for acq-b0
itself, let's follow up here: https://github.com/spine-generic/data-multi-subject/issues/104
I can disambiguate: the filename spec says that the only filenames with _MTS in them must also have _mt- in them. It's not optional:
indeed, so i think it would look like this:
sub-nottwil04_flip-1_mt-on_MTS.nii.gz
sub-nottwil04_flip-1_mt-off_MTS.nii.gz
sub-nottwil04_flip-2_mt-off_MTS.nii.gz
@kousu the "flip-2_mt-off" is in fact a T1w scan (because of the longer flip angle), hence the reason we called it T1w in the first place.
@agahkarakuzu can confirm
Indeed @jcohenadad that's how MTS entities are organized.
https://github.com/bids-standard/bids-examples/tree/master/qmri_mtsat
Oh great, thanks for clearing that up. So then this is the migration I'm going to do:
for subject in $subjects; do
(cd $subject/anat/
git mv $subject_acq-MTon_MTS.nii.gz $subject_flip-1_mt-on_MTS.nii.gz
git mv $subject_acq-MToff_MTS.nii.gz $subject_flip-1_mt-off_MTS.nii.gz
git mv $subject_acq-T1w_MTS.nii.gz $subject_flip-2_mt-off_MTS.nii.gz
)
done
I'll update #105 to match.
@kousu I think that this issue and associated PR has been resolved by https://github.com/spine-generic/data-multi-subject/pull/135, right?
As noted by @valosekj, this is already done.
We've misused the
acq
entity to store theMTState
field. We should be using themt
entitityAll these files are affected: