I would like perform ISC analysis on whole-brain level as well as for a list of a priori defined ROIs. In addition to synchronicity overall, I am also interested in the effects of group and other covariates as well as their interaction. Thus far, I have used 3dISC for whole brain analysis specifying a grey matter mask, thresholded the results using p = 0.001 and cluster-size corrected the resulting map based on the output of 3dClustSim using ACF estimate from the errts.* files as input. Is this an appropriate way of dealing with the multiple testing problem within ISC analysis on whole brain? Likewise, I was wondering whether I can proceed in a similar manner for the ROI analysis specifying the corresponding ROI masks in the 3dISC command?
I also came across your recent publication in NeuroImage on applying a BML framework to the ISC framework. Is this a more appropriate framework for the ROI analysis? Has this been implemented yet as an AFNI program?
I have used 3dISC for whole brain analysis specifying a grey matter mask, thresholded the results using p = 0.001 and
cluster-size corrected the resulting map based on the output of 3dClustSim using ACF estimate from the errts.* files
as input. Is this an appropriate way of dealing with the multiple testing problem within ISC analysis on whole brain?
I was wondering whether I can proceed in a similar manner for the ROI analysis specifying the corresponding ROI masks
in the 3dISC command? I also came across your recent publication in NeuroImage on applying a BML framework to the ISC
framework. Is this a more appropriate framework for the ROI analysis? Has this been implemented yet as an AFNI program?
How many ROIs do you have? With one ROI, you could directly use 3dISC. For multiple ROIs, the Bayesian multilevel modeling framework would be a good approach.
Could you please clarify what you mean when asking how the results look to me? I have read the paper you linked and liked the suggestion of reporting a highlighted response map with little to moderate thresholding.
In total, I have anatomically defined 4 ROIs, all subcortical. Hence, I was wondering whether there is either any “small volume correction” available within the 3dISC framework or whether I should use the BML framework for ISC analysis instead. I also really liked the whole-brain results you presented in NeuroImage publication (Untangling the effects part 3) and was considering to report a similar analysis using a whole-brain parcellation in addition to my whole-brain 3dISC results.
In our 3dISC analyses, we are interested in the effects of a between-subject variable (our experimental manipulation), a behavioural covariate (ratings obtained during naturalistic viewing) as well as their interaction. Specifying an initial p-value of 0.001 cluster-extend thresholded (k=32) to achieve alpha = 0.05, I find effects of group and the covariate, but no interaction. However, there is not really much research out there on that interaction effect we are interested in, especially not in the context of naturalistic stimuli, so it is very difficult to tell whether not finding results is expected or whether this is due to the harsh penalty of multiple testing.
The ROIs we are interested in are the Nucleus accumbens and caudate nucleus (they are contiguous), but also the anterior hippocampus and midbrain (not contiguous with each other or the other two ROIs). The problem is also that some of the ROIs are quite small, for instance the midbrain ROI only contains 60 voxels.
Specifying an initial p-value of 0.001 cluster-extend thresholded (k=32) to achieve alpha = 0.05,
I find effects of group and the covariate, but no interaction.
The problem is also that some of the ROIs are quite small, for instance the midbrain ROI only contains 60 voxels.
If you have 6 or ROIs, it might be possible to adopt the Bayesian multilevel approach as described in the NeuroImage paper you mentioned earlier.
The
National Institute of Mental Health (NIMH) is part of the National Institutes of
Health (NIH), a component of the U.S. Department of Health and Human
Services.