Hello SPM Developers & Experts,

My question is how does one come up with a rationale for setting one's extent cluster thresholds in SPM (e.g., using K=10 vs K=5 or even K=0).

I know Chumbley et al., 2010, showed a scatter plot depicting the FDR of voxels vs clusters, where the elbow at which agreement was highest was around K=10 (according to the red line in attachment); however, I couldn't find any text to help address my current question. Of course, this study was addressing a much more important issue than my question. To this end, perhaps there is better resource (reference) you can call to my attention as well, which will be even better! :)

Furthermore, does one's FWHM settings have anything to do with setting this K threshold?

Any feedback will be greatly appreciated!

Thanks in advance,

Will Goodman