Thanks a lot for pointing that out! I am not sure, whether this is, what you meant, but looking at the sources I found, that fsl is apparently normalizing the noise variance by the variance within the IC map, if I am not mistaken.
// var = e*var_on_e
// e is the estimate of the variance of the timeseries, sigma^2
var = (r.Column(1).t()*r.Column(1)*c.t()*inv_xx*c).AsScalar()/sizeTS;
Cheers
|