[Eeglablist] A list of cortex fipoles removed

Robert Oostenveld roberto at smi.auc.dk
Wed Apr 13 03:09:31 PDT 2005

Dear Chihiro,

On 11 Apr 2005, at 23:57, Chihiro Sugai wrote:
> Is there anyway to make a list of cortex dipoles removed or included 
> when running autofit components (Fit multiple ICA components; 
> pop_multifit())? [Not as a part of the summary figure]

No, that is not possible. It would not be very difficult to implement 
yousself, but I doubt whether it would be very usefull. Let me explain.

The autofit components routine will first scan the whole brain volume. 
That is done on a regular 3D grid of dipoles. Subsequently, the best 
grid-location is selected and subsequently a non-linear fit is 
performed with that grid-location as starting point. The result of the 
non-linear fit in general will not ly on the original regular grid, but 
somewhere between the grid points.

If you would use an irregular cortical grid for the initial scan, the 
result of the non-linear fit still in general would not ly on the 
cortical grid, but somewhere between the (irregularly spaced) grid 

Assuming that the dipole ends up in the global minimum (which is what 
you want), the result will not be different for the recular 3D grid and 
the cortical grid. The amount of time to find that global minimum may 
vary, since that depends on how far away the initial search location is 
from the final location. also the chance of ending in a local minimum 
is greater if the initial location is further away from the final 

You could skip the non-linear fit and only perform the dipole scan over 
the cortical surface, which would give you an optimal grid location 
(and hence a location on your cortical sheet). That is also not 
implemented, but it could be implemented in the dipfit_batch function 
(which you can find in the dipofit plugin directory). Around line 165 
the code reads
   % create the regular grid with dipole positions
   [X, Y, Z] = meshgrid(xgrid, ygrid, zgrid);
   grid.pos = [X(:), Y(:), Z(:)];
You have to replace that part of the code with something else. The 
carthesian grid locations in grid.pos should correspond with your 
cortical sheet.

best regards,
Robert Oostenveld

Robert Oostenveld, PhD
F.C. Donders Centre for Cognitive Neuroimaging
Radboud University Nijmegen
phone: +31-24-3619695
N.B. Starting from 1 September 2004, the University of Nijmegen has 
changed its name to Radboud University Nijmegen. All web- and 
email-addresses ending in ".kun.nl" should therefore be changed into 
".ru.nl". Please update your address book and links.

More information about the eeglablist mailing list