Hello,
Projectors are applied to the data when we epoch data from a continuous file (import data to database).
Why do we need to apply the projector too on the head model (brainstorm3/toolbox/process/functions/process_inverse.m at master · brainstorm-tools/brainstorm3 · GitHub) and noise covariance ?
Actually, how does brainstorm knows which projector to apply once the data are imported in the database (since its loading them from the channel matrix)
% Rebuild projector in the expanded form (I-UUt)
Proj = process_ssp2('BuildProjector', ChannelMat.Projector, [1 2]);
% Apply projectors
if ~isempty(Proj)
% Get all sensors for which the gain matrix was successfully computed
iGainSensors = find(sum(isnan(HeadModel.Gain), 2) == 0);
% Apply projectors to gain matrix
HeadModel.Gain(iGainSensors,:) = Proj(iGainSensors,iGainSensors) * HeadModel.Gain(iGainSensors,:);
% Apply SSPs on both sides of the noise covariance matrix
NoiseCovMat.NoiseCov = Proj * NoiseCovMat.NoiseCov * Proj';
end
end
We can imagine a case where I have a continuous file A. I import the file into the database (no projector selected), then apply the projector to A, and reimport the data; they will be in the same folder as the previous one. And there is no way to know which file needs the projector applied when doing the inverse problem.
Case 1:
Then the channel file is:
But then if i import event again withoiut applying SSP:
They end up in with the same channel file and source analysis will think SSP got applied to both files:
Regards
Edouard