search

CPernet / spmup

SPM tools to get the best of mass univariate analyses
GNU General Public License v3.0
40 stars 10 forks source link

spmup_auto_reorient does not like defaced anat files #5

Open Remi-Gau opened 6 years ago

Remi-Gau commented 6 years ago

Am starting testing spmup on a BIDS data set of mine and it seems that the spmup_auto_reorient is not too happy with defaced structurals (see attached check-reg: top left (before reorient) - top right (after reorient)). I was also getting funky results on some EPI that had a restricted field of view.

I don't really have a "fix" in mind but I could make a pull request to add a manual_coregistration function that has been circulating for a while in my lab but has not found a home in a github repo yet.

autoreorient_defacing

CPernet commented 6 years ago

ooh that it bad -- works ok with mri_deface and others because there are still some face .. funny the affine transform founds a minimum with a tilt! got to think what to do here ?? note I usually use it on T1 and apply on all EPI (which typically are roughly in register)

Remi-Gau commented 6 years ago

I guess if we had some skullstrip routine, this could help...

Will test spmup_auto_reorient.m on some other examples from the bids-examples repo that were defaced differently from mine. As soon as I am done with testing/improving the spmup_BIDS.m.

Will also test it on some MP2RAGE images I have: their weird noisy background messes up a lot of other softwares so it would be good to check if they are a problem here even before defacing.

Remi-Gau commented 6 years ago

OK checked on a few other data sets including MP2RAGE.

2 Subjects (1 and 2) from the 7t_trt BIDS data set: spmup_auto_reorient.m does not like MP2RAGE auto_reorient_anat_sub-01-02_bids_7trt

1 subject with MP2RAGE from one of my data set with and without defacing: ditto auto_reorient_anat_sub-02_bids_avt

2 subjects (1 and 2) from the ds009 BIDS data set with a normal T1w but with a different defacing than the one I used: this looks more acceptable though would require checking on other subjects. auto_reorient_anat_sub-01-02_bids_ds009

Remi-Gau commented 6 years ago

I will try to play around with the separation parameters when doing the affine registration: maybe this will help.

CPernet commented 1 year ago

is spmup autorient still a pblem? we could try Chris function https://github.com/hMRI-group/hMRI-toolbox/blob/master/hmri_autoreorient.m

Remi-Gau commented 1 year ago

will try to run it on a series of bids dataset to get an idea.

some vanilla T1w and some more exotic MP2RAGE to check.

CPernet commented 1 year ago

let's try to use https://github.com/lrq3000/auto_acpc_reorient