Beauchamp:CreateCortSurfMod

From OpenWetWare

(Difference between revisions)
Jump to: navigation, search
Line 1: Line 1:
-
== Automation ==
+
==Creating the surface==
 +
To create the surface, run the script
 +
  /Volumes/data1/scripts/@recon
 +
This script should be run in the home directory of the experiment, e.g.
 +
  cd /Volumes/data1/UT/DB
 +
  /Volumes/data1/scripts/@recon
 +
The scripts assumes that all of the steps performed by the script file @prep_dir have been performed, see
 +
[[Beauchamp:PrepCortSurfModels|Preparation for Creating Cortical Surface Models]]
-
All of the steps below are contained in the scripts file
+
FreeSurfer will take about 6 - 8 hours to complete. After it is finished, run the final script:
 +
 
 +
 
 +
For web pages describing the previous method, please see
 +
[[Beauchamp:CreateCortSurfModOLD|OLD version of Creating Cortical Surface Models]]
 +
 
 +
 
 +
== Automation ==
 +
All of the steps below are contained in the scripts file; information about them is included here for educational purposes.
/Volumes/data9/surface/scripts/@recon
/Volumes/data9/surface/scripts/@recon
To run this command type,
To run this command type,
-
  /Volumes/data9/surfaces/scripts/@recon ec subjname
 
Where ec is the two-letter experiment code and subjname is the subject's name in the format lastname_firstname.
Where ec is the two-letter experiment code and subjname is the subject's name in the format lastname_firstname.
The scripts assumes that all of the steps performed the script file @prep_dir have been performed (see Preparation for Creating Cortical Surface Models page)
The scripts assumes that all of the steps performed the script file @prep_dir have been performed (see Preparation for Creating Cortical Surface Models page)
Line 11: Line 25:
   /Volumes/data9/surfaces/scripts/@recon BD doe_jane
   /Volumes/data9/surfaces/scripts/@recon BD doe_jane
-
For web pages describing the previous method, please see
 
-
[[Beauchamp:CreateCortSurfModOLD|OLD version of Creating Cortical Surface Models]]
 

Revision as of 10:57, 15 September 2011

Contents

Creating the surface

To create the surface, run the script

 /Volumes/data1/scripts/@recon 

This script should be run in the home directory of the experiment, e.g.

 cd /Volumes/data1/UT/DB
 /Volumes/data1/scripts/@recon 

The scripts assumes that all of the steps performed by the script file @prep_dir have been performed, see Preparation for Creating Cortical Surface Models

FreeSurfer will take about 6 - 8 hours to complete. After it is finished, run the final script:


For web pages describing the previous method, please see OLD version of Creating Cortical Surface Models


Automation

All of the steps below are contained in the scripts file; information about them is included here for educational purposes. /Volumes/data9/surface/scripts/@recon

To run this command type, Where ec is the two-letter experiment code and subjname is the subject's name in the format lastname_firstname. The scripts assumes that all of the steps performed the script file @prep_dir have been performed (see Preparation for Creating Cortical Surface Models page) e.g.

 /Volumes/data9/surfaces/scripts/@recon BD doe_jane


Running FreeSurfer

The cortical surface reconstruction program recon-all has many options for processing the files, which are listed on the wiki: https://surfer.nmr.mgh.harvard.edu/fswiki/recon_2dall The easiest option is to run the entire file stream with the following command:

recon-all –all –s subjID

Often this will result in errors because FreeSurfer has permissions issues when copying and appending files. Processor usage can be checked using the program top in another window. If recon-all has been running one subprogram for an inordinate amount of time you can kill it using ctrl-c from the window where you are running recon-all. Reset the permissions so everyone can read and write files from the subject’s directory:

sudo chmod –R 777 $SUBJECTS_DIR/subjID

After resetting the permissions you should check where the program crashed by viewing recon-all-status.log in the subject’s scripts folder. Check the FreeSurfer wiki on how to restart the file stream from where the program crashed. For example, if the program during CA Normalize the program can be restarted using the following line:

recon-all –nogcareg –autorecon2 –autorecon3 –s subjID

Since the all the Macs in the lab are multiprocessor machines, the program can delegate each hemisphere to a single processor. To do this recon-all first needs to preprocess the anatomical data and then identify the white matter:

recon-all -autorecon1 -s subjID
recon-all -autorecon2 -s subjID -notessellate -nosmooth1 -noinflate1 -noqsphere -nofix -nofinalsurfs -nosmooth2 -noinflate2 –nocortribbon

After the white matter has been defined, recon-all can process the two hemispheres separately using separate log and status files:

recon-all -s subjID -hemi lh -log $SUBJECTS_DIR/subjID/scripts/recon-all_lh.log -status $SUBJECTS_DIR/subjID/scripts/recon-all-status_lh.log -tessellate -smooth1 -inflate1 -qsphere -fix -finalsurfs -smooth2 -inflate2 -cortribbon -autorecon3 &; recon-all -s subjID -hemi rh -log $SUBJECTS_DIR/subjID/scripts/recon-all_rh.log -status $SUBJECTS_DIR/subjID/scripts/recon-all-status_rh.log -tessellate -smooth1 -inflate1 -qsphere -fix -finalsurfs -smooth2 -inflate2 -cortribbon -autorecon3

Alternatively you can use the script in the surfaces folder, which runs recon-all in the manner described above:

/surfaces/@recon-all_sep_hemi –subj subjID

Be forewarned that use of this script on the iMacs will hamper system performance when the two hemispheres are simultaneously processed.

Using FreeSurfer on Another Mac To Speed Up Surface Creation

Some Macs have multiple processors and faster CPUs, making them better for surface creation. The fastest Mac in the lab is currently Mike Disano's MacPro. To create a surface on this Mac, you must first log in:

 ssh mbeauchamp@139.52.6.5

/data9 should already be accessible. Check this with

 ls ~/mount/data9/surfaces

Only if this gives an error message instead of a list of files, then you must mount the drive:

 cd mount
 mkdir data9
 mount_afp -i afp://mbeauchamp@129.106.236.180/data9 data9
 [enter password]

Next, run the @recon script:

 ~/mount/data9/surfaces/scripts/@recon_remote EI maili_lorena

It is easier to do the @prep_dir and @finish scripts on your local machine since they run quickly.


Making a Pial Envelope

It can also be useful to have a pial "envelope" that holds the whole brain. For instance, electrodes can be mapped to this surface with the guarantee that they will be visible in any surface model and not buried in cortex. The command to make a pial envelope is

 set subjid = pasalar_siavash
 set SUBJECTS_DIR = /Volumes/data9/surfaces/{$subjid}/
 setenv SUBJECTS_DIR /Volumes/data9/surfaces/{$subjid}/
 set ec = SP
 recon-all -s $ec -localGI

This command requires that Matlab be in your path; you will receive an error if it is not. You may also not be able to run multiple copies of Matlab at once, so you should quit all open Matlabs before executing. This command also computes a local gyrification index, which may be decreased in diseases or disorders and could be interesting to look at.

Sample Subjects To Look At

Condensed Notes

/surfaces/autism/ASD/EB_010806/EB_010806notes.rtf


When errors were encountered during autorecon-all

/surfaces/autism/ASD/DB_073006/DB_073006notes.rtf


Pretty complete Notes

/surfaces/hicks_kali/BVnotes.rtf

covers merging in afni, running recon-all, creating SS500, creating sulc.asc files, aligning to experimental data

Personal tools