Hello SPMmers,
I have been studying the notation and the levels of significance between
set-level, cluster-level and peak level values. Thomas Nichols' definitions
are helpful, but that is for SPM99. I wonder if someone can provide clear
definitions for SPM8. Thomas Nichols' definitions here:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind01&L=SPM&P=R53954&X=0F7060015A3831710E&Y=narayanan.32%40osu.edu
My understanding is that cluster-level refers to the probability of whether
the more widespread broad activation occurs by chance. Peak-level refers to
the probability whether the activations are more localized, whether this
occurs by chance. Is this correct?
Secondly, is the 'T' field in peak-level activation referring to the
threshold? That would explain why there are multiple rows for peak-level for
every cluster-level. My interpretation is that there are multiple peaks
within each cluster. Then what about the troughs or locations where there is
an inverted peak? Finally, what does the number underneath 'T' field
actually stand for?
Also, what is the difference between 'p_FWE-corr' and 'p_FDR-corr' (in both
peak and cluster-level stats)? Also, what does the Z (in peak-level) and k_E
(in cluster-level) stand for? If someone could provide the definitions, I'd
appreciate it.
Thanks in advance.
- Ananth Narayanan
|