Hi Anderson,
Thanks for your patience. I have uploaded the input image (tfce_p_tstat1) into the specified folder. I have also uploaded the a priori mask we used in randomise, and we used this same mask with the fdr tool (unsure if this is correct). Finally, I uploaded the corrp tstat image in case you need it.
Thanks again!
Erin
------------
Hi Erin,
Could you upload the input image given to fdr? You should have just received an automated email with instructions.
Thanks.
All the best,
Anderson
On 27 February 2017 at 23:05, Erin W. <[log in to unmask]> wrote:
Hi Anderson,
I had a follow-up question in regards to this inquiry. I recently ran randomise with TFCE on some pilot data--we used a mask based on a priori predictions. My corrp image, thresholded at 0.95, has a significant result, but the voxel size is extremely small (size=2). I ran FDR on the uncorrected image using the same mask but the probability threshold is 0.
I ran the following command based on the FDR page and forums: fdr -i tfce_p_tstat.nii.gz --oneminusp -m /path/to/mask/mask.nii.gz -q 0.05 -v
Is there a reason this cluster would be significant using TFCE but not FDR? (Perhaps this is due to limiting the proportion of false positives? Or how the sig. clusters are computed in each calculation?) Or is my command incorrect?
Thank you for your time and input!
Best,
Erin
|