I have used fsl’s MELODIC in the past and I remember it telling you what the variance explained of each returned component was. But now that I have taken a deeper dive into ICA I see that ICA cannot tell you variance, after all it is maximizing non-gaussianity of the components.
I could not find this feature explained in the documentation of fsl’s ICA algorithm: https://www.fmrib.ox.ac.uk/datasets/techrep/tr04cb1/tr04cb1.pdf