AFNI and Clustering: False Positive Rates Redux

Abstract

In response to reports of inflated false positive rate (FPR) in FMRI group analysis tools, a series of replications, investigations, and software modifications were made to address this issue. While these investigations continue, significant progress has been made to adapt AFNI to fix such problems. Two separate lines of changes have been made. First, a longtailed model for the spatial correlation of the FMRI noise characterized by autocorrelation function (ACF) was developed and implemented into the 3dClustSim tool for determining the clustersize threshold to use for a given voxelwise threshold. Second, the 3dttest++ program was modified to do randomization of the voxelwise t tests and then to feed those randomized t statistic maps into 3dClustSim directly for clustersize threshold determination-without any spatial model for the ACF. These approaches were tested with the Beijing subset of the FCON1000 data collection. The first approach shows markedly improved (reduced) FPR, but in many cases is still above the nominal 5%. The second approach shows FPRs clustered tightly about 5% across all pervoxel p value thresholds ≤ 0.01. If t tests from a univariate GLM are adequate for the group analysis in question, the second approach is what the AFNI group currently recommends for thresholding. If more complex pervoxel statistical analyses are required (where permutation/randomization is impracticable), then our current recommendation is to use the new ACF modeling approach coupled with a pervoxel p threshold of 0.001 or below. Simulations were also repeated with the now infamously "buggy" version of 3dClustSim: the effect of the bug on FPRs was minimal (of order a few percent).

Temporal gradient during famous face naming is associated with lower cerebral blood flow and gray matter volume in aging

“…ASL image processing was performed with Analysis of Functional NeuroImages (AFNI, afni.nimh.nih.gov) (Cox, 1996), FMRIB Software Library (FSL, Oxford, United Kingdom), and locally created Matlab scripts. Field map correction was applied to the ASL time series prior to co-registration to the middle time point to minimize the effects of participant motion.…”

“…Significance was determined by applying cluster-size correction derived from randomization of voxel-wise t-tests (via AFNI’s Clustsim option in 3dttest++) and then feeding those randomized t-statistic maps into Monte-Carlo simulations directly for cluster-size threshold determination (via AFNI’s 3dClustSim) to guard against false positives on data initially thresholded at a value of p <0.01 (uncorrected). The Clustsim option added to the 3dttest++ command approach was developed in response to reports of inflated false positive rate (FPR) in fMRI group analysis tools (Eklund, Nichols, & Knutsson, 2016) and is recommended for use when t-tests from a univariate general linear model (GLM) are adequate for the group analysis in question (Cox, Chen, Glen, Reynolds, & Taylor, 2017; Cox, Reynolds, & Taylor, 2016). Based on these simulations, it was determined that a minimum cluster volume of 270 μL (10 contiguous voxels) was required to correct for multiple comparisons at p <0.01 corresponding to a voxel-level threshold of p <0.01.…”

Hierarchical organization of the functional brain identified using floating aggregation of functional signals

“…The cortical meshes were extracted from sMRI scans and aligned using Freesurfer [11], and downsampled uniformly on the sphere at 3-mm resolution using Freesurfer. All the fMRI data were preprocessed using AFNI [12], including slice timing, head movement correction, bandpass filtering, and regressing out nuisance covariates. The preprocessed fMRI data were then projected with Freesurfer onto the cortical meshes.…”

Evidence accumulation detected in BOLD signal using slow perceptual decision making

“…This ACF was fit to a mixture model of Gaussian plus mono-exponential. We then used the group mean ACF parameters to estimate the probability of false positives (using the most recent version of AFNI's 3dClustSim; Cox, Reynolds, & Taylor, 2016). We set a voxelwise threshold of p < 0.001 and a corrected threshold of p < 0.01 with bi-sided and face/edge nearest neighbor clustering parameters.…”