[Eeglablist] interpolating electrode using topoplot

[Natasa Kovacevic]

I run into a problem when trying to interpolate values for one electrode
using other electrodes based on the following the logic:

eeg = original data set where time series for electrode Q needs to be
replaced by interpolated values

tmp_eeg = pop_select(eeg,'nochannel',Q); % remove bad electrode

use topoplot on tmp_eeg with 'noplot' option to get interpolated value
for coordinates corresponding to Q

insert interpolated values back into eeg 

- it reports 'out of bound' error for many values of Q. I believe that
there is a bug in topoplot function when using 'noplot' option and it
has to do with mapping of electrode coordinates (theta and radius) into
the rectangular grid of interpolated values. 

I believe that the following portion of topoplot.m  (starting at line
~1260)


%%%%%%%%%% Return interpolated value at designated scalp location %%%%%%
%%%%
  %
  if exist('chanrad')   % optional first argument to 'noplot' 
      chantheta = (chantheta/360)*2*pi;
      chancoords =
round(ceil(GRID_SCALE/2)+GRID_SCALE/2*2*chanrad*[cos(-chantheta),...

-sin(-chantheta)]);
      if chancoords(1)<1 ...
         | chancoords(1) > GRID_SCALE ...
            | chancoords(2)<1 ...
               | chancoords(2)>GRID_SCALE
          error('designated ''noplot'' channel out of bounds')
      else
        chanval = Zi(chancoords(1),chancoords(2));
        grid = Zi;
        Zi = chanval;  % return interpolated value instead of Zi
      end
  end
%%%%%%%%%%%%%%%%%%%%%%%%



should be replaced by :




%%%%%%%%%% Return interpolated value at designated scalp location %%%%%%
%%%%
  %
  if exist('chanrad')   % optional first argument to 'noplot' 
      chantheta = (theta(i)/180)*pi; 
      chanrad = rad(i);  
      [chanx chany] = pol2cart(chantheta,chanrad); 
      chanx    = chanx * squeezefac;   
      chany    = chany * squeezefac;       
      chancoords = round(GRID_SCALE/2 + GRID_SCALE/2*[chanx chany]);
      
      if chancoords(1)<1 ...
         | chancoords(1) > GRID_SCALE ...
            | chancoords(2)<1 ...
               | chancoords(2)>GRID_SCALE
          error('designated ''noplot'' channel out of bounds')
      else
        chanval = Zi(chancoords(1),chancoords(2));
        grid = Zi;
        Zi = chanval;  % return interpolated value instead of Zi
      end
  end
%%%%%%%%%%%%%%%%%%%%%%%%


Can anyone confirm that this is correct. Thanks, 
Natasa


-- 
Natasa Kovacevic, PhD
Rotman Research Institute
Baycrest Centre
Toronto
(416) 858-8095
nkovacev at rotman-baycrest.on.ca