The situation is complicated. First, you should run with --cwp 1 to get
a list of all clusters. it maybe that a cluster is just barely out of
threshold and so does not show up in the list.
When you change the threshold you, you change the probability curve and
obviously the size of the cluster; this can have unpredictable effects.
Eg, at 1.3, it may be MUCH more likely to get a false positive because
of the lower threshold.
On 3/14/2020 5:25 PM, Purcell, Juliann wrote:
External Email - Use Caution
At all other smoothing and threshold combinations, the simulations
indicate that there are no significant clusters on the lateral surface
of the left hemisphere, but there is a large cluster on the medial
surface (lingual gyrus). For 10mm+th.13, however, the simulation
indicates there is a large cluster on the lateral surface
(supramarginal gyrus), but no significant cluster on the medial
surface. All cluster.summary files have been attached to this email
(smoothing at 10mm, thresholds 1.3, 2.0, 2.3, and 3.0) in case they
are helpful.
Having a significant cluster at a lower threshold (i.e., 1.3), but not
at a more stringent threshold makes sense. But I’m not sure why the
lingual gyrus cluster isn’t present at 1.3 when it is present at more
stringent thresholds. Any guidance or suggestions are appreciated!
Best,
Juliann Purcell, MSc
Doctoral Candidate
Medical/Clinical Psychology Doctoral Program
University of Alabama at Birmingham
Department of Psychology
julia...@uab.edu <mailto:julia...@uab.edu>
On Mar 13, 2020, at 5:49 PM, Douglas N. Greve <dgr...@mgh.harvard.edu
<mailto:dgr...@mgh.harvard.edu>> wrote:
Can you say more about how the 10mm, th 1.3 is different than the others?
On 3/11/2020 2:48 PM, Purcell, Juliann wrote:
External Email - Use Caution
Hi FreeSurfer Experts,
I am running permutation simulations with several smoothing levels
and cluster forming thresholds to evaluate the best parameters for
our data, since there does not seem to be a consensus in the
literature. However, I’m getting an unexpected result at only one
smoothing/threshold level.
I have run permutations (mri_glmfit-sim) at 0mm, 5mm, 10mm using the
following thresholds: 1.3, 2.0, 2.3, and 3.0. Results of
permutation at 10mm + threshold 1.3 were not as expected and did not
line up with results at the other smoothing and threshold levels.
I’ve re-checked that the command is correctly referencing the data
and I’ve compared the th1.3 log file with the log files at the other
thresholds. I could not find any differences that would cause the
output of the permutation simulations to be so different between
th1.3 and th2.0. There are no errors or failures in the permutation
log files. I’ve searched the message board and didn’t see any posts
that related to this issue.
Although it’s possible that these are true results, I want to make
sure there isn’t a mistake on my end. Any help would be greatly
appreciated!
Below is information about my FreeSurfer version, platform, and the
log files from the permutation simulations:
1)FreeSurfer version:
freesurfer-Linux-centos6_x86_64-stable-pub-v6.0.0-2beb96c
2)Platform: Linux
3)uname -a: Linux c0135 3.10.0-1062.9.1.el7.x86_64 #1 SMP Mon Dec 2
08:31:54 EST 2019 x86_64 x86_64 x86_64 GNU/Linux
4)Perm sim log files: See attached logs for 10mm smoothing at
threshold 1.3 and threshold 2.0
Best,
Juliann Purcell, MSc
Doctoral Candidate
Medical/Clinical Psychology Doctoral Program
University of Alabama at Birmingham
Department of Psychology
julia...@uab.edu <mailto:julia...@uab.edu>
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu <mailto:Freesurfer@nmr.mgh.harvard.edu>
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu <mailto:Freesurfer@nmr.mgh.harvard.edu>
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
