spm_vb_laplacian

 Create spatial Laplacian matrix for a slice of data 
 FORMAT [L] = spm_vb_laplacian (vxyz)

 vxyz      list of neighbours of voxels to be analysed
 L         laplacian matrix

 %W% Will Penny and Nelson Trujillo-Barreto %E%

0001 function [L] = spm_vb_laplacian (vxyz)
0002 % Create spatial Laplacian matrix for a slice of data
0003 % FORMAT [L] = spm_vb_laplacian (vxyz)
0004 %
0005 % vxyz      list of neighbours of voxels to be analysed
0006 % L         laplacian matrix
0007 %
0008 % %W% Will Penny and Nelson Trujillo-Barreto %E%
0009 
0010 voxels=size(vxyz,1);
0011 L=4*speye(voxels);
0012 
0013 for j=1:voxels,
0014     
0015     L(j,nonzeros(vxyz(j,:)))=-1;
0016     
0017     % Ensure normalisation is correct for edges/corners
0018     % keeping this line makes L rank deficient
0019     %L(j,j)=length(nonzeros(vxyz(j,:)));
0020     
0021     % With above line commented out, this produces the
0022     % Laplacian used in LORETA, with biased boundary conditions
0023     % - see discussion in section 2 of paper VB2
0024 end;