Dear SPM experts,
I re-submit this mail b/c I’m still in
trouble.
I’m very new
SPMer and statistical approach, so making effort to catch up and understand the
procedure.
For now, I’m trying to analyze multi-subject NIRS data based on FDR.
I found that we can get the threshold
for activation from ‘spm_uc_FDR.m’.
In the ‘spm_uc_FDR.m’, it converts the t-values (which I’m using)
to p-values, sort p-values,
find the p-value’s index which decides the activation, and get the threshold from
reconverting the p-value to t-value.
Am I
understanding well?
If it’s right, here’s some question.
1) In the spm_uc_FDR.m, there’s some code, ‘Ps =
(1-spm_Tcdf(Ts,df(2))).^n;’
I’m not sure what the conjunction number ‘n’ means.
For now,
I’m doing work with 5
subjects’ data (they did same task),
in this case, how do
I determine the conjunction number?
2) and, we need
degree of freedom for converting between t-values and p-values.
As far
as I know, degree of freedom is the number of subject (minus one) for
multi-subject analysis.
So, for
my case, I thought it has range from 0~4. Is that right?
Actually, I could not
see any activation from that. That’s why I’m confused.
In each individual
analysis, I can see the activation. And those df values are around 90~92
Any suggestion or advice would be very
appreciate for me.
Thank you.
Jin