Re: [Freesurfer] Cortical depth at each cortical voxel

Hi Panos,

Please see my responses interleaved.

Best, Iman

From: freesurfer-bounces@nmr.mgh.harvard.edu freesurfer-bounces@nmr.mgh.harvard.edu On Behalf Of Fotiadis, Panagiotis Sent: Monday, April 25, 2022 10:48 PM To: Freesurfer support list freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] Cortical depth at each cortical voxel

Hi Iman,

Apologies for the follow up! I had a couple questions concerning the software:

1. Concerning the MLI_thickness.m function: I used a binary mask of the cortical ribbon (= I) as its first input and L (where L = makeLineSegments(round(.2*size(I,1)) ) as the second input (similarly to the suggestion in EXAMPLE.m file). I tried different radii as well for the makeLineSegments function and the thickness results are highly linear to each other. Do those inputs seem reasonable in this case (i.e., for real 3D MRI scans), in your opinion?

In makeLineSegments, you may want to use a radius that is a few voxels larger than the maximum thickness that you expect (voxels being the unit of the measured thickness).

1. Concerning the cortical depth calculation for each voxel: Unless I'm mistaken, the (cortical depth = Thickness/2 - SkeletonDistance) equation would give me the cortical depth of each voxel from the closest surface layer, right? So for instance, if my voxel of interest were "above" the skeleton (i.e., the middle layer of the cortical ribbon), then the above equation would give out its distance from the surface layer that is also above the skeleton, whereas if my voxel were "below" the skeleton, then the aforementioned equation would give its distance from the surface layer that is also below the skeleton, right?

That is correct.

Would there be, however, a way to compute the cortical depth of each cortical voxel from the outer surface layer of the cortical ribbon? I believe that would be distance t1(v) mentioned in the equations and text in p. 3199 of your paper, but I unfortunately wasn't able to pinpoint that variable in the scripts. Is that variable hard-coded in the scripts, or would you recommend that I follow a different approach to calculate that (for instance set up a conditional argument where I use cortical depth = Thickness/2 - SkeletonDistance if the voxel is above the skeleton layer, and cortical depth = Thickness/2 + SkeletonDistance if the voxel is below the skeleton layer)?

t1(v) is the oneSidedThickness_partial variable in the MLI_thickness function. The algorithm, however, works pretty locally and cannot really distinguish between the inner and outer surfaces of the cortical ribbon. So, In order to use t1(v) for your purpose (or your other suggested approach), you would need a label for each voxel that says which surface the voxel is closer to. If you have masks for the subcortical areas (e.g. the white-matter mask) and the beyond-pial area (~(GM | WM)), you could create such a label by comparing the distance transform maps created from the two masks (by Matlab's bwdist function). In fact, the distance transform from the latter mask might just give you the cortical depth that you are looking for.

Thank you again very much for your time and help!

Best, Panos

From: freesurfer-bounces@nmr.mgh.harvard.edumailto:freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mgh.harvard.edumailto:freesurfer-bounces@nmr.mgh.harvard.edu> on behalf of Aganj, Iman <IAGANJ@mgh.harvard.edumailto:IAGANJ@mgh.harvard.edu> Date: Thursday, April 21, 2022 at 1:02 PM To: Freesurfer support list <freesurfer@nmr.mgh.harvard.edumailto:freesurfer@nmr.mgh.harvard.edu> Subject: [External] Re: [Freesurfer] Cortical depth at each cortical voxel Of course! Please let me know if you have any questions about the toolbox.

Best, Iman

From: freesurfer-bounces@nmr.mgh.harvard.edumailto:freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mgh.harvard.edumailto:freesurfer-bounces@nmr.mgh.harvard.edu> On Behalf Of Fotiadis, Panagiotis Sent: Thursday, April 21, 2022 12:32 PM To: Freesurfer support list <freesurfer@nmr.mgh.harvard.edumailto:freesurfer@nmr.mgh.harvard.edu> Subject: Re: [Freesurfer] Cortical depth at each cortical voxel

Hi Iman,

That is extremely helpful, thank you very much (also the paper is a great read)!

Best, Panos

From: freesurfer-bounces@nmr.mgh.harvard.edumailto:freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mgh.harvard.edumailto:freesurfer-bounces@nmr.mgh.harvard.edu> on behalf of Aganj, Iman <IAGANJ@mgh.harvard.edumailto:IAGANJ@mgh.harvard.edu> Date: Thursday, April 21, 2022 at 12:06 PM To: Freesurfer support list <freesurfer@nmr.mgh.harvard.edumailto:freesurfer@nmr.mgh.harvard.edu> Subject: [External] Re: [Freesurfer] Cortical depth at each cortical voxel Hi Panos,

You can do such a volumetric analysis using this toolbox:

Code: https://www.nitrc.org/projects/thickness Paper: https://onlinelibrary.wiley.com/doi/epdf/10.1002/hbm.20740

You can apply the MLI_thickness function to the gray matter (soft or binary) mask, which returns the layer thickness at each voxel (Thickness) and the distance of each voxel to the center of the layer (SkeletonDistance). Cortical depth would then be SkeletonDistance at the given voxel subtracted from half of Thickness (at the voxel, or preferably on the skeleton for better accuracy). A tutorial is included in the EXAMPLE file.

Best, Iman

From: freesurfer-bounces@nmr.mgh.harvard.edumailto:freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mgh.harvard.edumailto:freesurfer-bounces@nmr.mgh.harvard.edu> On Behalf Of Fotiadis, Panagiotis Sent: Thursday, April 21, 2022 11:48 AM To: freesurfer@nmr.mgh.harvard.edumailto:freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] Cortical depth at each cortical voxel

Hi FS experts,

I was wondering whether there was a way to assign each cortical voxel within the T1w (volumetric) scan an index corresponding to its "cortical depth." I have essentially taken the cortical ribbon .mgz file, assigned each cortical voxel a voxel ID, and wanted to associate each ID with its corresponding cortical depth. I would normally use the *h.sulc files but I didn't think this would properly translate here since it's a volumetric analysis.

Thank you in advance for your time and help!

Best Panos

-- Panagiotis Fotiadis Ph.D. Candidate | Neuroscience Graduate Group Perelman School of Medicine, University of Pennsylvania