[Mne_analysis] BEM coregistration

[Christian O'Reilly]

        External Email - Use Caution        

Hmm... it looks like I was not running MNE will all the last commits. The
image I shown is probably what you would get on a pip install of the last
official release. On the current master, the co-registration IS good

[image: image.png]

The BEM surfaces were generated with the original MRI, so checking them
against a volume converted with --conform shows that they are not in
co-registration anymore (which is accurate) :

[image: image.png]

So, part of the issue was already kinda solved... I say kinda, because in
the current version, the displayed co-registration is faithful to the
underlying data. However, the orientation is good only for volume in the
"conform" space (i.e., these are supposed to be in coronal view... which is
not the case for the first (not conform) figure).

Sure, I'll create an issue on GitHub for that. Thanks for the feedback. For
now, I'll re-rerun with --conform and adjust my BEM surfaces accordingly.

Best,

Christian



Le mer. 15 janv. 2020 à 14:31, Eric Larson <larson.eric.d at gmail.com> a
écrit :

>         External Email - Use Caution
>
> I am using the mne.viz.plot_bem function and I get some co-registration
>> issues between the volume and the surfaces:
>>
>
> This sort of problem has occurred before, and I think it had to do with
> handling volume information / orientation properly.
>
> which do not produce the T1.mgz file that is expected by the
>> mne.viz.plot_bem function (as a side note, it might be nice to have an
>> optional mri arguments to mne.viz.plot_bem to explicitly pass the location
>> of the MRI file to use), so I generated it from the original mprage.nii.gz
>> volume expected as input by infant_recon_all:
>>
>> mri_convert $SUBJECT_DIR/mprage.nii.gz $SUBJECT_DIR/mri/T1.mgz
>>
>
> We might assume that the T1.mgz is in some standard orientation. Can you
> see if adding `--conform` fixes things? I think plot_bem assumes a specific
> slice orientation.
>
>
> The different surfaces co-register well in freeview:
>>
>
> Internally freeview probably does this conforming, we probably do not. I
> guess we could in theory use nibabel's resampling to get it in the correct
> orientation.
>
> I agree it would also make sense to:
>
> 1. Be able to specify the MRI image to use
> 2. Ensure the MRI is conformed to the correct orientation / reslice it if
> necessary
>
> Can you open an issue on GitHub and we can go from there?
>
> Eric
>
> _______________________________________________
> Mne_analysis mailing list
> Mne_analysis at nmr.mgh.harvard.edu
> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20200115/799bde07/attachment-0001.html 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 298415 bytes
Desc: not available
Url : http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20200115/799bde07/attachment-0002.png 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 165539 bytes
Desc: not available
Url : http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20200115/799bde07/attachment-0003.png