Dear FSL experts,
I am struggling with generating p-values based upon the flame results. I know I have to use the GRF options (-d,--dlh and --volume and -p,--pthresh) to output p-values. However, when I do I encounter two problems.
1. When calculating the smoothness estimate using smoothest the DLH result is "nan"
I calculate the smoothest using the degrees of freedom, MNI152_T1_2mm_brain_mask (the same mask as used for the flame analysis) and the res4d image (output from flame as well). Only when I apply spatial smoothing of at least 3mm I do get a DLH 0.0126473. I assume the unsmoothed data is nearing a DLH of 0 and therefore does output "nan". However, when I spatially smooth with 3mm or more the significance of the voxels decreases (based on the z-values right now, since I don't get the p-values).
2. When I want to perform cluster-wise interference using cluster using a smoothing input of 0 it doesn't give the p-value output. When I use the spatial smoothed z-stat image and corresponding DLH I get the following error: "calculateExtrema: mask is empty"
Which results in the same output as without DLH, so I guess something went wrong with the calculation of the p-values again.
This is the command I use for cluster:
cluster --in=zstat1 -c cope1 -d ${smoothness} --volume=$(cat thresh_zstat1.vol) --connectivity=26 -t 2.3 -p 0.01 --minclustersize=20 -o cluster_mask_zstat1 --othresh=thresh_zstat1 --olmax=lmax_zstat1_tal.txt > cluster_zstat1_std.txt
So my question is how do I calculate the P-value when the smoothness is really low and how do I solve the empty mask error?
I am aware of the non-parametric randomise command with Threshold-Free Cluster Enhancement, but I also want to perform statistics using the traditional parametric method using flame.
Thanks in advance.
Best regards,
Jordi Kleinloog
PhD candidate Maastricht University
########################################################################
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
|