Hi Anatasia,
According to the original TFCE paper,
TFCE incorporates the cluster information across a range of possible cluster-forming thresholds,
which makes it difficult,
especially on the TBSS skeleton, to compare the results with clusters formed after thresholding the T-stat map.
Could you please upload the data in order for us to take a look at it?
Cheers
Hello,
I am working on TBSS/VBM comparison and was wondering whether there are any
cases where TFCE correction could fail. Examining our raw t-stat images for
both TBSS and VBM analysis, we note t-values of 5 and above distributed in
clusters that do support our hypothesis. However, once TFCE is performed, no
group differences survive the correction. Is there a reason why this might
be happening?
Sincerely,
Anastasia