ETAC: Equitable Thresholding and Clustering in AFNI

A little over 1 year ago, a short article in PNAS (http://www.pnas.org/content/113/28/7900.full) put forward claims that the most widely used cluster thresholding technique in FMRI was terribly inaccurate with highly inflated false positive rates, and the paper initially called into "question the validity of some 40,000 fMRI studies" (this specific number was later retracted). These claims were widely commented on in the popular press, with further exaggeration and misinterpretation. This paper contained a mixture of real concerns and over-strong statements about the implications of their analyses. In response, I have spent much of the last year investigating this issue and developing new methods in AFNI to deal with the real problems that were raised. I will present an overview of the cluster thresholding methods used in AFNI, from the "classic" days to the newest method: ETAC, which has the potential to detect both large and small clusters, removes arbitrariness in choice of thresholding parameters, and maintains the correct false positive rate.