Hi Doug,
Thanks for that! with regards to your last point
Unfortunately, it does not help much. Much was made of the bug that Eklund, et al, found in the alphasim program (the AFNI mc simulator), but even after fixing it, the volume-based analyses still had very high false positive rates. This is because the problem is that the smoothness in the data is not Gaussian, so any method that assumes Gaussianity will be inaccurate.
From what I read, the MC simulations on the surface uses this Gaussian
assumption. Do you have any ideas (or something in the works) on how to handle these issues on a single subject level? - since the permutation test won't be applicable.
Cheers,
*Dr Kevin Aquino* Research fellow, Sir Peter Mansfield Magnetic Resonance Center, The University of Nottingham.
Honorary Research Fellow School of Physics, Faculty of Science, University of Sydney
*E* Kevin.aquino@nottingham.ac.uk, aquino@physics.usyd.edu.au | *W* https://kevinaquino.github.io http://www.physics.usyd.edu.au/~aquino/
----------------------------------------------
The brain is a wonderful organ. It starts working the moment you get up and does not stop until you get into the office. - Robert Frost
CRICOS 00026A This email plus any attachments to it are confidential. Any unauthorised use is strictly prohibited. If you receive this email in error, please delete it and any attachments.
Please think of our environment and only print this e-mail if necessary.
On Wed, May 17, 2017 at 1:18 AM, Douglas Greve greve@nmr.mgh.harvard.edu wrote:
On 5/15/17 10:34 PM, Kevin Aquino wrote:
Hi all,
First of all, Freesurfer V6 is working like a dream, my 0.7 mm segmentations are running really well and in comparisons to hi-resrecon in 5.3 and early beta, I'm having to do fewer manual corrections!
Now for my questions,
- I'm running mri_mcsim in order to correct for multiple comparisons via
the FS-FAST stream. I'm wondering how many iterations are advised, and how can one check for convergence in an automatic fashion.
I've run the simulations with 1000 and 10,000 iterations on a 1mm segmentation with the FWHM simulations at 8mm. (i.e. using mri_mcsim --o . --base mc-z --save-iter --surf subject lh/rh --nreps 10000 --fwhm 8) and I can't see many differences between the two when correcting for multiple comparisons (i.e. using cluster-sess -analysis myanalysis -thresh 3 -cwp .05 -s SESSION -sign pos).
For the tables hat we distribute I ran it to 10,000, but there will probably not be much difference with 1000. If you look in the cluster summary file, it will actually give the 95% confidence intervals on the cluster p-values. If the worst is ok, then you don't really need to run it more.
- I'm trying to find some references that detail the simulations and form
the corrections, does anyone have advice which list I can read/start off with, as well as some key papers that use it (esp on a single subject level).
You can check out "Smoothing and cluster thresholding for cortical surface-based group analysis of fMRI data" by Don Hagler
I really like this approach and It does look to circumvent (I think...) a lot of the problems of cluster-wise corrections described with Eklund et al. (Cluster failure paper).
Unfortunately, it does not help much. Much was made of the bug that Eklund, et al, found in the alphasim program (the AFNI mc simulator), but even after fixing it, the volume-based analyses still had very high false positive rates. This is because the problem is that the smoothness in the data is not Gaussian, so any method that assumes Gaussianity will be inaccurate. doug
Cheers,
*Dr Kevin Aquino* Research fellow, Sir Peter Mansfield Magnetic Resonance Center, The University of Nottingham.
Honorary Research Fellow School of Physics, Faculty of Science, University of Sydney
*E* Kevin.aquino@nottingham.ac.uk, aquino@physics.usyd.edu.au | *W* *MailScanner has detected a possible fraud attempt from "www.physics.usyd.edu.au" claiming to be* https://kevinaquino.github.io http://www.physics.usyd.edu.au/%7Eaquino/
The brain is a wonderful organ. It starts working the moment you get up and does not stop until you get into the office.
Robert Frost
CRICOS 00026A This email plus any attachments to it are confidential. Any unauthorised use is strictly prohibited. If you receive this email in error, please delete it and any attachments.
Please think of our environment and only print this e-mail if necessary.
Freesurfer mailing listFreesurfer@nmr.mgh.harvard.eduhttps://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
The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.