Hi-
I think that you would probably have to add a lot more details about your data (how many subjects? what age group? any particular concerns within it, like being likely to move?), processing choices (what kind of alignment to MNI space did you use? How did you verify it? It is task, rest or naturalistic scanning---I'm just guessing it is FMRI? did you use afni_proc.py, or something else to process? if afni_proc.py, what was your command?), quality control description (did you use afni_proc.py's QC HTML?), group analysis method (what was it?), and more to be able to suggest ideas.
There are lots of places that FMRI processing can go awry, and hence why we focus a lot on the need for understanding+verifying each block of data analysis. It isn't possible to guess a potential source of woe from just final images, but please feel free to add more description or concerns that you might have about the data, and we can mention things to check/verify or ways to try to address them.
--pt