1. I have some noisy patient data and am doing spike regression to clean it. This leaves some subjects with low degrees of freedom. I am curious what subjects I need to throw out to do a simple block design? I know for resting state, 4 minutes of data is usually used as the cut off. Is it the same for block design?

2. What about if I do a PPI analysis which involves the interaction of a block design with connectivity. Would that also require at least 4 minutes of data?

My tr is 2 seconds.