Would you mind spending some time to reply me about using GIFT to do group-ICA?

Dear everyones,

Recently, I use GIFT to conduct group Independent component analysis(ICA) on 5 sessions of one subject ,which were preprocessed by DPARSF.  However, just 3 ICs were found, and they seems to be noises.

Besides, using the same processing steps on 4sessions of  the other one subject , we could obtain 12 ICs, of which there were exists functional networks, such as auditory. 

I think that this results are not relate to the process method.  So, what causes this differents???    I am looking forward to your replay if you have some idea about it.  Thanks.

Chengzhen Guo.