Dear Alvaro and Claus,
SPM indeed displays 0.05 critical threshold for the various correction
methods, at the bottom of the results section. If you wanted another
alpha-level, see spm_getSPM.m line 696 for, e.g., peak-FDR:
[up, Pp] = spm_uc_peakFDR(0.05,df,STAT,R,n,Zum,XYZum,u);
change 0.05 to 0.1 to get in variable 'up' the 0.1 critical heigth
threshold for peak-FDR.
Best regards,
Guillaume.
Claus Lamm wrote:
> Alvaro,
>
> if you go through the results procedure once and then click on the button
> "whole brain",
> you will get a results table in the Graphics window.
> There, in the lower left corner, you should see four values:
> FWEp, FDRp, FWEc, FDRc
>
> These are the t-values or cluster extent values
> for obtaining maps thresholded at P <= 0.05 using the respective
> correction method (FWEp = family-wise error, height FWEc = cluster extent;
> FDRp/c for FDR-correction using height and extent)
>
> Just enter these values when going through the results procedure again.
>
> As to how to get p<0.1, I am not sure, you'd probably have to look that up
> in the code,
> not sure where though - maybe someone else can comment on that?
>
> Hope this helps,
>
> Claus
>
>
> -----Original Message-----
> From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On
> Behalf Of Alvaro Muro
> Sent: Mittwoch, 17. März 2010 12:36
> To: [log in to unmask]
> Subject: [SPM] FDR p-threshold in SPM8
>
> Hi all,
>
> I'm trying to use FDR correction in SPM8 as explained in
>
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=SPM;pxeBbw;20090406192205%2B0
> 100
>
> However, I don't have a clear understanding of the equivalence between the T
> and p threshold values. I would be very grateful if somebody could point
> which value should I enter after clicking 'none' if I want to use a p<0.05
> FDR threshold. And for a p<0.1 ?
>
> Thank you very much for your help and guidance!
>
> Best wishes,
>
> Álvaro
>
>
--
Guillaume Flandin, PhD
Wellcome Trust Centre for Neuroimaging
University College London
12 Queen Square
London WC1N 3BG
|