Dear Gaoyan

[log in to unmask]" type="cite">

Dear SPM users:

I have a question about the topological FDR in SPM8.

The reviewers asked me not to change the variable defaults.stats.topoFDR and set it to 1. But there is no buttons to apply this. I want to konw how to get the result of topological FDR in SPM8?

default parameters are located in spm_defaults.m
line 68 defaults.stats.topoFDR should already be set to 1

[log in to unmask]" type="cite">

I have seen in the forum an example: First choose “none” in spm, use a T=3 to obtain the FWEc and the FDR p/c and then use the values when going through the results procedure again. Is it right?

in the result table, you will see both FWE and FDR ; you want to keep clusters with p value FDR corrected < say 0.05 ; keep the initial uncorrected p value at 0.001

[log in to unmask]" type="cite">

Another question: What is the relation between peak FDR, cluster FDR and topological FDR?

one use to have voxel FDR, ie correct the expected false discovery rate across voxels --> do p = 0.001 uncorrected and then remove the proportion of false positives among them
--> this procedure is no longer available, the main weakness being that is doesn't account for the topology

topological FDR applies FDR on clusters (thus account for the fact that data/results have a spatial structure) --> do p= 0.001 to get clusters then from the cluster distribution do the FDR correction --> on peak = get the peak values across clusters and correct --> on extent = get the distribution of size and correct

see Chumbley et al - there are 2 neuroimage papers 44(2):62-70 and 49(4):3057-3064

Cyril