Dear FSL experts,
I have a theoretical question regarding the operation of dual regression. If I were to input a volume from ICA into dual regression, why should I input the whole image and not a thresholded image, for example at z>4?
If I was doing a network analysis, wouldn't using an unthresholded image "wash out" the signal from network? Does dual regression weight the voxel z-values of the input ICA volume?
Thank you in advance,
Isaac
|