Changes to recon-all
This outlines the changes made to recon-all effective Friday January 27, 2006.
OLD RECON-ALL vs NEW RECON-ALL
the brain volume is now called brainmask.mgz, but is still used in exactly the same way. You will make all manual edits needed to the brainmask.mgz volume.
- recon-all-nmr will no longer exist.
the old flags (-stage1, -stage2, -stage3, -stage4a, -stage4b) will no longer be accepted by recon-all. Instead use the new set of flags (-autorecon1, -autorecon2, -autorecon3, etc.) See the ReconAllDevTable for a list of flags.
- the new recon-all will not accept COR format, only .mgz files. (this is only for NEW data being processed, see section below for re-running previously processed data)
- the new recon-all will accept DICOM files as input. Previously mri_convert was used to convert native scanner images into either COR or .mgz, now recon-all will accept anything that could have been used as an input to mri_convert. When you run recon-all from your SUBJECTS_DIR and use native scanner formats as your input it will first convert the files to $SUBJECTS_DIR/subject_name/mri/orig/001.mgz etc., from there it will average and motion correct your multiple acquisitions and continue with the cortical reconstruction stream. The inputs are specified using this command:
recon-all -i </full/path/to/invol_1> -i </full/path/to/invol_2> -subjid subject_name
where invol_1 is the first dicom file of your first acquisition and invol_2 is the first dicom file of your second acquisition.
previously the command mksubjdirs was needed to set up the appropriate directory structure for each subject in your SUBJECTS_DIR, but now recon-all will do this for you as well. By specifying the subject's id, subject_name on the command line, recon-all will create each directory as it's needed through the stream.
- the old recon-all required you to specify when you wanted to keep any manual edits you'd done (i.e., -keepwmedits, -usecontrolpoints, -notalairach if you'd manually changed yours). Now, by default, recon-all will keep all of your edits. This includes: talairach changes, control points, manually set cutting planes, brain volume edits, and wm volume edits. You now have to specify when you want these edits to be ignored.
- there are new outputs associated with the new recon-all, mostly brainmask.auto.mgz, aseg.auto.mgz, wm.auto.mgz. These are used to keep your edits separate from the manually corrected volumes. If you have made no manual edits, the ?.auto.mgz and the ?.mgz volumes will be the same. Make any edits, as normal, to the wm.mgz and brainmask.mgz volume and not to the wm.auto.mgz or brainmask.auto.mgz volume.
RE-RUNNING PREVIOUSLY PROCESSED DATA
From here on any data that has been previously processed, to any extent, by FreeSurfer will be referred to as a legacy data set.
any legacy data set that is currently in COR format will first be converted over the mgz format and then re-run using that.
- all manual edits/interventions will, by default, be kept (i.e., talairach interventions, wm edits, control points, etc.)
you must use the -legacy flag the first time legacy data sets are re-run. This directs recon-all to convert everything over properly, and to maintain your edits. After this initial re-run your data set is no longer a legacy data set and the -legacy flag is not needed. If you run a legacy data set without the -legacy flag, recon-all will exit immediately and warn you to use the -legacy flag. If you use the -legacy flag on non-legacy data, recon-all will exit immediately and warn you that this flag is not needed.
USEFUL INFORMATION ON THE WIKI
Please be patient as it may take some time to fully update all the information on the wiki. Here are some quick useful links to pages that have information reflecting the current changes to recon-all.
- the new recon-all is described in full detail here:
https://surfer.nmr.mgh.harvard.edu/fswiki/recon_2dall
- workflow pages outlining this new process have been updated and can be found here:
https://surfer.nmr.mgh.harvard.edu/fswiki/FreeSurferWorkFlows
- the table outlining the steps and underlying commands of the entire process can be found here:
https://surfer.nmr.mgh.harvard.edu/fswiki/ReconAllDevTable
- a table of other useful flags to use with the new recon-all can be found here: