I computed the forward model with vecterview example in bst after I extract just the meg squid,
exery thing work fine,
I then worked with default anatomy and the extracted meg (every think looks right in the coregisteration)
I plotted the values for a (semi-)random channel (e.g., "80"):
plt.plot(gain[80,:])
plt.show
and looks like there are outliers in the forward solution, approximately around vertex #36390-36392..
What happen?
If you have computed the BEM OpenMEEG solution, it is sometimes unstable at some limit conditions (when the dipoles are too close to the inner layers).
Right-click on the forward model > View leadfield vectors.
Post the screen capture here. You should see your outliers here.
Post also a screen capture showing: the cortex surface without smoothing + the inner skull in red + the other layers you've been using.
Set the inner skull transparency to 0% and check if there is any point of the cortex that seems outside the inner skull.
Right-click on the cortex you've be estimating the leadfield for > Force inside skull.
Compute again the leadfield. Does it change anything?
Indeed, we can see one abnormal field in the right-temporal area.
Use the overlapping spheres method for MEG, instead of the OpenMEEG BEM which has this instabilities, this is our general recommendation.
