 |
 |
Dear Feng, the AFNI group fixed the bug related to the reduced size searched for clusters. In our preprint you can find a longer explanation (see below). Bob Cox is still working on improving the 3dClustSim function. If you have further questions about 3dClustSim, I propose that you ask them at the AFNI message board, https://afni.nimh.nih.gov/afni/community/board/ "Secondly, a 15 year old bug was found in 3dClustSim while testing the three software packages (the bug was fixed by the AFNI group as of May 2015 1 , during preparation of this manuscript). The effect of the bug was an underestimation of how likely it is to find a cluster of a certain size (in other words, the p-values reported by 3dClustSim were too low). The main idea behind the 3dClustSim function is to generate Gaussian noise with unit variance, and then smooth it using a Gaussian lowpass filter with a size corresponding to the estimated group smoothness. This procedure is repeated a large number of times, to obtain an estimate of how common different cluster sizes are for Gaussian noise. The smoothed noise is rescaled back to unit variance, and 3dClustSim performs the rescaling by first estimating the variance of the smoothed noise. Due to edge effects caused by the smoothing operation the boundary of the volume is attenuated, which has two effects. First, the estimated variance used for standardization will be biased down, increasing the variance of the simulated images. Second, the attenuation will reduce the chance that clusters will ever occur near the boundary, effectively reducing the search volume and under estimating the severity of the multiple testing problem." Regards, Anders >Dear SPM experts, > >There is a blog post explaining the results of the paper “”Cluster failure: Why fMRI inferences for spatial extent have inflated false-positive rates > >In this blog, the authors state “AFNI problem identified. The results presented in this manuscript include the use of a pre May 2015 version of AFNI, specifically the 3dClustSim function used to >implement the parametric FWE control. One of the discoveries made during this project was the smoothness estimate used in this older version of 3dClustSim had a flaw that >increased the FWE. This was fixed by the AFNI developers in versions after May 2015. Although the new version reduces FWE, it is still inflated above the target of 5%; the p=0.01 >and p=0.001 cluster defining thresholds’ FWE with 3dClustSim changed from 31.0% to 27.1% and 11.5% to 8.6%, respectively. ” > >In Eklund’s paper, there is a subsection about this issue. > >Why Does AFNIs Monte Carlo Approach, Unreliant on RFT, Not Perform Better? As can be observed in SI Appendix, Figs. S2, S4, S8, and S10, AFNI’s FWE rates are excessive even for a >CDT of P = 0.001. There are two main factors that explain these results. > >First, AFNI estimates the spatial group smoothness differently compared with SPM and FSL. AFNI averages smoothness estimates from the first-level analysis, whereas SPM and FSL estimate >the group smoothness using the group residuals from the general linear model (32). The group smoothness used by 3dClustSim may for this reason be too low (compared with SPM and FSL; SI >Appendix, Fig. S15). > >Second, a 15-year-old bug was found in 3dClustSim while testing the three software packages (the bug was fixed by the AFNI group as of May 2015, during preparation of this >manuscript). The bug essentially reduced the size of the image searched for clusters, underestimating the severity of the multiplicity correction and overestimating significance >(i.e., 3dClustSim FWE P values were too low). > >My question is that which bug was fixed by AFNI group on May 2015? The smoothness estimate or reduced the size of the image searched for clusters? In PNAS paper, it wrote that the bug is >reduced the size of the image searched for clusters. However, in the blog “KEEP CALM AND SCAN ON”, the bug is smoothness estimate problem. > >Best, >Feng