Display on Cortex (Time Frequency Map)

balukrishnan · January 21, 2022, 10:11pm

I had a question regarding the interpolation techniques used for displaying TF maps on the cortical surface. How does brainstorm interpolate the TF maps to be displayed on the cortical surface? Are there any relevant literature that can help me understand it better?

Nichrishayes · January 21, 2022, 10:24pm

This will largely depend on which source imaging method you used, and whether you used constrained vs unconstrained methods. This tutorial on source estimation is helpful. For example, if you have 3 orientations unconstrained, it will perform the vectorial sum operation.

balukrishnan · January 22, 2022, 12:33am

I forgot to mention that I am working on intracranial EEG and TF plots are directly estimated from the intracranial EEG. I suspect it to some sort of interpolation technique.

Francois · January 22, 2022, 7:56am

When EEG/SEEG/ECOG data are displayed on a surface (head, gray matter, white matter, skull...), the values at the surface vertices are computed from the recordings using a 3D nearest-neighbor interpolation using Shepard's weighting.
This is done in the same way for temporal (EEG recordings), spectral (PSD) and time-frequency data.

The relevant functions:

https://en.wikipedia.org/wiki/Inverse_distance_weighting

github.com

brainstorm-tools/brainstorm3/blob/master/toolbox/gui/panel_surface.m#L1865

    
      
                      disp('BST> All values are null. Please check your input file.');

                  end

              end

              % Update surface definition

              setappdata(hFig, 'Surface', TessInfo);

              % Update colormap

              UpdateSurfaceColormap(hFig, iSurfaces);

          end

          

          

          %% ===== COMPUTE SCALP INTERPOLATION =====

          function TessInfo = ComputeScalpInterpolation(iDS, iFig, TessInfo)

              global GlobalData;

              % If surface is not displayed: exit

              if isempty(TessInfo.hPatch) || ~ishandle(TessInfo.hPatch)

                  return;

              end

              % Get vertices of surface

              Vertices = get(TessInfo.hPatch, 'Vertices');

              % Get selected channels indices and location

              selChan = GlobalData.DataSet(iDS).Figure(iFig).SelectedChannels;

github.com

brainstorm-tools/brainstorm3/blob/master/toolbox/math/bst_shepards.m

function Wmat = bst_shepards(destLoc, srcLoc, nbNeighbors, excludeParam, expDistance)
% BST_SHEPARDS: 3D nearest-neighbor interpolation using Shepard's weighting.
%
% USAGE:  Wmat = bst_shepards(destLoc, srcLoc, nbNeighbors=8, excludeParam=0, expDistance=2)
%
% INPUT:
%    - srcLoc       : Nx3 array of original locations, or tesselation structure (Faces,Vertices,VertConn)
%    - destLoc      : NNx3 array of locations onto original data will be interpolated, or tesselation structure (Faces,Vertices,VertConn)
%    - nbNeighbors  : Number of nearest neighbors to be considered in the interpolation (default is 8)
%    - excludeParam : If > 0, the source points that are two far away from the destination surface are ignored.
%                     Excluded points #i that have: (minDist(i) > mean(minDist) + excludeParam * std(minDist))
%                     where minDist represents the minimal distance between each source point and the destination surface
%                     If < 0, exclude the vertices that are further from the absolute distance excludeParam  (in millimeters)
%    - expDistance  : Distance exponent (if higher, influence of a value decreases faster)
%    
% OUTPUT:
%    - Wmat : Interpolation matrix

% @=============================================================================
% This function is part of the Brainstorm software:

This file has been truncated. show original