Hi Nick,
Thanks so much for the helpful reply.
Actually, the ICA program that I use actually will do the normalization and detrending steps. Based on what I have been told bandpassing should not be used because the ICA program benefits from all the frequency information. I did try bandpassing and did a motion regression prior to ICA, and got some strange results. Additionally, the alogrithm that estimates the amount of components in the data got completely thrown off.
Perhaps as a next step I will give 3dtcat a try on the blur files and then run it through the additional preprocessing steps run in GIFT.
Thanks so much,
Emily