Hi,
just by curiosity, when its a good idea to use the Max of the PCA function instead the mean function for scout amplitude extraction?
is it wrong to do for between group study?
Thanks
1 Like
I think using the maximum is always a bad idea. It is a very noisy statistic.
Regarding the PCA vs. the mean: it's a matter of personal taste or beliefs.
The mean is more intuitive, I have personally never used much this PCA option.
Maybe @Sylvain can give you better reasons to use the PCA?
1 Like
My two cents on this is that there isn't really a good way to sum up signals over an extended region where the cortex geometry varies. Having different source orientations changes the signal a lot purely from the physics of the system (silent "radial" sources vs. "tangential" sources for example).
But I think PCA might be more appropriate when one wants to keep the sign of the signal, for doing phase-based connectivity measures for example. It's still rather arbitrary, but gives something somewhat understandable (the component with the largest variance), whereas averaging is not really well defined if the polarity can't be matched between voxels.
1 Like