I've just created the following docs for mri_mcsim doug

mri_mcsim --o top-output-dir --base csdbase --surface subjectname hemi --nreps nrepetitions --fwhm FWHM <FWHM2 FWHM3 ...> --fwhm-max FWHMMax : sim with fwhm=1:FWHMMax (default 30)

--avgvtxarea : report cluster area based on average vtx area --seed randomseed : default is to choose based on ToD --label labelfile : default is ?h.cortex.label --mask maskfile : instead of label --no-label : do not use a label to mask --no-save-mask : do not save mask to output (good for mult jobs) --surfname surfname : default is white

--log LogFile --done DoneFile : will create DoneFile when finished --stop stopfile : default is ourdir/mri_mcsim.stop --save savefile : default is ourdir/mri_mcsim.save --save-iter : save output after each iteration --sd SUBJECTS_DIR --debug turn on debugging --checkopts don't run anything, just check options and exit --help print out information on how to use this program --version print out version and exit

$Id: mri_mcsim.c,v 1.26 2014/04/11 15:31:56 greve Exp $

This program computes tables for performing multiple comparisons on the surface based on a monte carlo simulation using white gaussian noise smoothed on the surface. mri_mcsim was used to generate the tables found in $FREESURFER_HOME/average/mult-comp-cor and used by mri_glmfit-sim with the --cache option. The tables in mult-comp-cor are for fsaverage and fsaverage_sym for a search space over the whole cortex. mri_mcsim can be used to create new tables for a new subject or for fsaverage/fsaverage_sum with a smaller search space.

Example 1: Create tables for a new subject for whole hemisphere

mri_mcsim --o /path/to/mult-comp-cor/newsubject/lh/cortex --base mc-z --save-iter --surf newsubject lh --nreps 10000

This may take hours (or even days) to run; see below for parallelizing. When running mri_glmfit-sim, add --cache-dir /path/to/mult-comp-cor

Example 2: Create tables for superior temporal gyrus for fsaverage

First, create the label by running

mri_annotation2label --subject fsaverage --hemi lh --outdir labeldir

mri_mcsim --o /path/to/mult-comp-cor/fsaverage/lh/superiortemporal --base mc-z --save-iter --surf fsaverage lh --nreps 10000 --label labeldir/lh.superiortemporal.label

When running mri_glmfit, make sure to use --label labeldir/lh.superiortemporal.label When running mri_glmfit-sim, add --cache-dir /path/to/mult-comp-cor --cache-label superiortemporal

Example 3: running simulations in parallel (two jobs, 5000 iterations each for a total of 10000)

mri_mcsim --o /path/to/mult-comp-cor/fsaverage/lh/superiortemporal --base mc-z.j001 --save-iter --surf fsaverage lh --nreps 5000 --label labeldir/lh.superiortemporal.label

mri_mcsim --o /path/to/mult-comp-cor/fsaverage/lh/superiortemporal --base mc-z.j002 --save-iter --surf fsaverage lh --nreps 500 --label labeldir/lh.superiortemporal.label

When those jobs are done, merge the results into a single table with

mri_surfcluster --csd /path/to/mult-comp-cor/fsaverage/lh/superiortemporal/fwhm10/abs/th20/mc-z.j001.csd

--csd /path/to/mult-comp-cor/fsaverage/lh/superiortemporal/fwhm10/abs/th20/mc-z.j002.csd

--csd-out /path/to/mult-comp-cor/fsaverage/lh/superiortemporal/fwhm10/abs/th20/mc-z.csd

--csdpdf /path/to/mult-comp-cor/fsaverage/lh/superiortemporal/fwhm10/abs/th20/mc-z.cdf --csdpdf-only

Repeat the above command for each FWHM, sign (pos, neg, abs) and threshold

On 04/11/2014 11:08 AM, Douglas N Greve wrote:

That should have been --nreps 10000 (not 100:)

On 04/11/2014 11:04 AM, Douglas N Greve wrote:

...

On 04/11/2014 02:52 AM, Gautam Tammewar wrote:

...

Hi all,

I'm trying to run Monte Carlo simulations on cortical thickness data I got using mri_glmfit, but I'm running into some errors.

1. I used the command

mri_glmfit-sim --glmdir lh.thickness.fwhm20.diagnosis_scanner.dods.glmdir/ --cache 2.0 pos

The error I receive says "cannot find /usr/local/freesurfer_x86_64-5.1.0/average/mult-comp-cor/fsaverage/lh/cortex/fwhm40/pos/th20/mc-z.csd"

Even though you've smoothed by 20, the actual FWHM in the data is 40, which is pretty high, and the tables do not go that high. This may indicate that there is something wrong with your data

...

I'm wondering why the script is looking in the directory "fwhm40" when the data is smoothed using FWHM 20. The directories available under fsaverage/lh/cortex go from fwhm01 to fwhm30.

2. Also, for another test I'm running, I created my own average

subjects (instead of using fsaverage). Is there a way to generate the files that are created under /average/mult-comp-cor for this average subject so that I don't have to use the "--sim" flag to run simulations?

Yes, you can use mri_mcsim, something like

mri_mcsim --o /path/to/mult-comp-cor/yoursubject/lh/cortex --base mc-z --save-iter --surf yoursubject lh --nreps 100

This may take several hours or days to run

When running mri_glmfit-sim, add --cache-dir /path/to/mult-comp-cor

doug

...

Thanks, GT

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer