Back to the main page.
Bug 662 - ft_megplanar should be made consistent with ft_neighbourselection
| Status | CLOSED DUPLICATE |
| Reported | 2011-05-12 16:45:00 +0200 |
| Modified | 2011-07-29 10:34:12 +0200 |
| Product: | FieldTrip |
| Component: | core |
| Version: | unspecified |
| Hardware: | PC |
| Operating System: | Mac OS |
| Importance: | P1 normal |
| Assigned to: | Jörn M. Horschig |
| URL: | |
| Tags: | |
| Depends on: | |
| Blocks: | |
| See also: |
Robert Oostenveld - 2011-05-12 16:45:55 +0200
... instead of using its own implementation inside megplanar_sincos. At the moment it does for i=1:Ngrad for j=(i+1):Ngrad distance(i,j) = norm(pnt(i,:)-pnt(j,:)); distance(j,i) = distance(i,j); end end fprintf('minimum distance between gradiometers is %6.2f %s\n', min(distance(find(distance~=0))), grad.unit); fprintf('maximum distance between gradiometers is %6.2f %s\n', max(distance(find(distance~=0))), grad.unit); % select the channels that are neighbours, channel is not a neighbour of itself neighbsel = distance<cfg.neighbourdist; for i=1:Ngrad neighbsel(i,i) = 0; end fprintf('average number of neighbours is %f\n', sum(neighbsel(:))./size(neighbsel,1)); wher the default cfg.neighbourdist is 4 or 40. This fails for yokogawa64, and might be suboptimal for other systems as well. The optimal neighbours for megplanar are probably the same as for clustering. Note that the decision might be that the user should be forced to call ft_neighbourselection manually instead of doing it automatically. I guess this needs some thinking, followed by some discussion.
Jörn M. Horschig - 2011-05-13 10:26:45 +0200
I suppose using a neighbour template might be most appropriate here (in case one exists) - I'll start working on this next week