Hi,
About Biswal...what you can do is a cross-correlation between all your IC and MetaICA_4mm.nii.gz template....to see which of your components have anatomical sense (no noise, no white matter, no motion...).
You will need dual_regression script and the template. Ask in the mailing list and do not forget to mention him in your paper.
About the command line: fslcc melodic_IC.nii.gz metaICA_4mm.nii.gz which wil give you a first column with all your components, a second column with de numbers of your IC components that match with metaICA_4mm.nii.gz template and the value of correlation. From my little experience, I recommend you take value from 0.45 to high.
This will sure you that you are identifiying correctly your IC of interest.
Hope this helps and good luck!
Cheers,