I'm trying out numerous comparison correction methods on my fMRI dataset and I've encountered multiple papers which used threshold free cluster enhancement, which I'd like to incorporate, but I'm running into difficulty regarding

  1. actually implementing TFCE
  2. Understanding how to display the results on an MNI template, should I be able to even implement TFCE

One such paper which uses this method is given here, by Park et al. However, within this paper (and other similar ones) only the original TFCE methods paper is cited without specifying any tool-box or practical advice on how to actually implement this method in SPM.

Inferences on a multidimensional social hierarchy use a grid-like code

For example, in the above article we have:

For all analyses using a GLM (GLM1, GLM2, GLM3 and GLM4), group-level effects were tested with a one-sample t-test. First, we performed group-level inference on a priori task-independent EC62,63 and HC61 ROIs that were defined anatomically. Second, we performed exploratory whole-brain analyses. To correct for multiple comparisons, we used permutation-based TFCE68. The medial EC has been well known to contain grid cells and is thought to be their primary source7. A signature of these grid cells in EC has been found in recent human fMRI studies, and, thus, a central hypothesis of our study was that EC would show grid-like coding. Specifically, these recent studies show EC grid-like coding both for spatially navigating virtual environments24 and also for tracking the abstract task state defined by the continuously morphing neck and leg lengths of a bird stimulus14. The EC activity in these studies was modulated by the direction of the vector defined on a 2D space with hexagonal patterns. The EC is, therefore, the primary ROI to test our main hypothesis in the current study (note that all vectors are aligned to the EC grid orientation). In addition, the anatomical geometry of the EC is not ideal for cluster-based inference, such as TFCE. For those reasons, we used the anatomically defined ROI-based TFCE correction for statistical inference of the grid effect in the EC (PTFCE < 0.05 corrected in the ROI). We report activations (Figs. 3, 4 and 5) that are cluster-corrected at the threshold PTFCE < 0.05 over the whole brain for all brain regions, except for the EC, where we used a threshold of PTFCE < 0.05 corrected within anatomically defined ROI due to our strong a priori hypothesis of grid coding in EC.

Are there any references or first-hand-experience anyone could share regarding TFCE, its implementation, and how to go about displaying these results?




Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.