"Intensity too high" when using mne_watershed_bem

✉️ Mailing List Archive (read-only)
1

Hello,

mne_watershed_bem is exiting with the following error:

w=White Matter =Intensity too high (>240)...valid input ?

I'm running mne_watershed_bem on an average brain I created from 23 other
participants using Freesurfer's make_average_subject. I've been able to do
mne_watershed_bem without error on all the other subjects, but I get this
error when using the average breain. Does anyone have any guesses what
could be causing the error?

Thank you,
Steve Politzer-Ahles

2

See reply in text

Hello,

mne_watershed_bem is exiting with the following error:

w=White Matter =Intensity too high (>240)...valid input ?

This error is due to the MRI data having unusually large values. It actually comes from FreeSurfer's mri_watershed program (called by mne_watershed_bem). You would need to ask the FreeSurfer people about how the volumes are generated by the make average subject and any recommendations about using mri_watershed with them.

The analysis you describe does beg another question: Why do you want to generate a BEM for the average subject?

D

3

Hi Dan,

Thanks a lot for your reply. I will send a message to the FreeSurfer list
to see what they say.

Regarding your second question, I have two subjects without MRIs (one was
unable to be scanned, and the other one moved during the scan) and was
hoping to just analyze their data on an average brain rather than throw
them out entirely.

Best,
Steve

4

hi Steve,

Regarding your second question, I have two subjects without MRIs (one was
unable to be scanned, and the other one moved during the scan) and was
hoping to just analyze their data on an average brain rather than throw them
out entirely.

the cortical constraint might not be adapted in this case and you
might have problems to have a good coregistration. I tend to say that
using either one of your subject or the average brain is pretty much
the same here. I would personally use one of my subject where the
coregistration ends up being not too bad and then use a free
orientation inverse.

Alex

5

In that case, you may be better off (obviously throwing them out may
apply, but given you don't want to do that) using an individual who best
fits the digitized points. The average surfaces have very different
shape from individual surfaces and do not necessarily make sense to use
for source reconstruction i.e. the orientations in an average brain will
be very different from those in the individuals who compose it.

D

Stephen Politzer-Ahles wrote: