External Email - Use Caution

Yes, that looks ok to me.

Regarding code: this is just a draft, and kind of slow/inefficient, but it should work:

addpath $FREESURFER_HOME/matlab BRAINSTEM=MRIread(‘brainstemSsLabels.v10.FSvoxelSpace.mgz’); ASEG=MRIread(‘aseg.mgz’);; maskToFill=ASEG.vol==16;

llist=unique(ASEG.vol(~maskToFill)); dmap=zeros([size(ASEG.vol) length(llist)]); for l=1:length(llist) dmap(:,:,:,l)=bwdist(ASEG.vol==llist(l)); end [~,idx]=min(dmap,[],4); filling=llist(idx); ASEG.vol(maskToFill)=filling(maskToFill); ASEG.vol(BRAINSTEM.vol>0)=BRAINSTEM.vol(BRAINSTEM.vol>0);

MRIwrite(BRAINSTEM,’modifiedASEG.mgz’);

-- Juan Eugenio Iglesias ERC Senior Research Fellow Centre for Medical Image Computing (CMIC) University College London http://www.jeiglesias.com http://cmictig.cs.ucl.ac.uk/

From: Mehul Sampat mpsampat@gmail.com Date: Thursday, 21 June 2018 at 16:47 To: "Iglesias Gonzalez, Eugenio" e.iglesias@ucl.ac.uk Cc: Freesurfer Mailing List Freesurfer@nmr.mgh.harvard.edu Subject: Re: Merging aparc+aseg.mgz and brainstemSsLabels.v10.FSvoxelSpace.mgz

Hi Eugenio, Thank you for your reply. The solution to blend the two segmentation will work for me. If you could share the Matlab code that would be great.

Also one other related question i have is related to edits and brainstem-structures.

1. The first time i run recon-all, i use: "recon-all -all -s bert -brainstem-structures. 2. Then suppose, I add control points or edit brain-mask; i re-run recon-all with: 3. recon-all -s bert -autorecon2-cp -autorecon3

* when i do this i think i should also add the brainstem-structures flag again. thus my command will be:

1. recon-all -s bert -autorecon2-cp -autorecon3 -brainstem-structures. Does step-4 look correct to you ? Thanks Mehul

On Thu, Jun 21, 2018 at 2:44 AM, Iglesias Gonzalez, Eugenio <e.iglesias@ucl.ac.ukmailto:e.iglesias@ucl.ac.uk> wrote: Dear Mehul, Yes, the labels from the Brainstem module overlap with the definition of ventral-dc in aseg.mgz, since they are based on atlas built with delinations made with different criteria. A trick you can use to blend the two segmentations into one is: 1. Create a binary mask for the whole brainstem using brainstemSsLabels.v10.FSvoxelSpace.mgz. 2. In aseg.mgz (or aparc+aseg.mgz), replace the label of every voxel in the mask by that of the nearest voxel outside the mask. 3. Paint the brainstem from brainstemSsLabels.v10.FSvoxelSpace.mgz over the mask in the output of step 2. Does this make any sense? I can provide you with some Matlab code if you with. Cheers, /E

-- Juan Eugenio Iglesias ERC Senior Research Fellow Centre for Medical Image Computing (CMIC) University College London http://www.jeiglesias.com http://cmictig.cs.ucl.ac.uk/

From: Mehul Sampat <mpsampat@gmail.commailto:mpsampat@gmail.com> Date: Thursday, 21 June 2018 at 04:10 To: Freesurfer Mailing List <Freesurfer@nmr.mgh.harvard.edumailto:Freesurfer@nmr.mgh.harvard.edu>, "Iglesias Gonzalez, Eugenio" <e.iglesias@ucl.ac.ukmailto:e.iglesias@ucl.ac.uk> Subject: Merging aparc+aseg.mgz and brainstemSsLabels.v10.FSvoxelSpace.mgz

Hi Eugenio and Bruce,

We used freesurfer-6.0.0 to segment the brain-stem into different different structures.

I see we get two mgz files after this process: aparc+aseg.mgz and brainstemSsLabels.v10.FSvoxelSpace.mgz

Do you have any recommendations to combine these two masks into a single mask ?

I see pons in brainstemSsLabels.v10.FSvoxelSpace.mgz overlaps with the brain-stem in aparc+aseg.mgz but the midbrain in brainstemSsLabels.v10.FSvoxelSpace.mgz overlaps with brain-stem and the ventral-dc in aparc+aseg.mgz.

Any recommendations are appreciated. Thanks Mehul