 |
 |
To clarify terminology, topological FDR means apply FDR to topological features, being height or size. Cyril ----- Reply message ----- From: "H. Nebl" <[log in to unmask]> To: <[log in to unmask]> Subject: [SPM] FDR correction Date: Tue, Jul 14, 2015 16:17 Just to make sure, topological FDR doesn't mean it has to be applied on cluster level. The second paper by Chumbley and colleagues proposes a toplogical FDR on peak level. But indeed, as Shashwath stated, an initial voxel threshold is necessary as well. Accordingly, it is difficult to infer what the reviewer doesn't like. Does he prefer cluster thresholds in general or does he "just" dislike your voxel FDR because he thinks topological FDR is better (and possibly overseeing that there's a topological FDR on peak level as well)? > Does the same apply for cluster level FWE? Yes, you would also start with an uncorrected voxel threshold, the FWEc corresponds to the smallest, still sig. cluster at cluster FWE .05. If you want to go with another cluster FWE then you have to check the cluster p values yourself. Best Helmut