All great questions, Liwen!

It’s the same data, but with the subfields partitioned into body/head when appropriate, and with the nuclei of the amygdala. And there were also some changes to the optimization algorithm of the atlas building method. So, when you build the atlas, it looks a tiny bit different (see attached captures of the 6.0 and dev atlases).
I wouldn’t say the fit of the atlas is more accurate, but having the subdivision into head/boday/tail may indeed help understand distinct subfield functions along the anterior-posterior axis.
Interesting finding! The fit of the two atlases to the image is almost identical, since the head/body divisions are not considered in the process. I would lean towards using dev and then merging the volumes of the anterior and posterior sections of each subregion, if you think the parcellation is too fine.
I wish I knew the answer! There’s a bunch of things we want to change (adding regions of entorhinal cortex to the atlas, porting the code to Python), but I don’t know whether this will happen before or after the release of the next stable FS version (which, again, nobody knows when will be). I would say it’s perfectly fine to run subjects with recon-all and then running the dev hippo/amygdala module, as long as you report it in the paper. As of today, the atlas in dev is the only one that has the nuclei of the amygdala…

I hope this helps,

Juan Eugenio Iglesias

Senior research fellow

CMIC (UCL), MGH (HMS) and CSAIL (MIT)

http://www.jeiglesias.com

From: <freesurfer-bounces@nmr.mgh.harvard.edu> on behalf of Liwen Zhang <lw.zhang@yahoo.com>
Reply-To: Freesurfer support list <freesurfer@nmr.mgh.harvard.edu>
Date: Tuesday, 24 September 2019 at 06:18
To: Freesurfer support list <freesurfer@nmr.mgh.harvard.edu>
Cc: Helen Juan Zhou <helen.zhou@duke-nus.edu.sg>
Subject: Re: [Freesurfer] help! Hippocampal subfield segmentation with FS development version

External Email - Use Caution

Dear Eugenio,

Many thanks for your prompt reply and explanation. May I ask few follow-up questions, please?

1) You mentioned "we rebuilt the atlas ... change the code...", would you please explain a little bit more about what you mean with "rebuilt"? In the FS dev, does the "rebuilt" mean that you use different segmentation algorithm on the same FS v.6.0 ex-vivo atlas template, or even the ex-vivo atlas template per se is already different in FS dev from that in FS v.6.0?

2) Related to 1), given the volume discrepancy between the two FS versions, from your perspective, would you say the current hippocampal subfield segmentation in FS dev is more accurate compared with FS v.6.0 algorithm? Does this further HBT segmentation within subfields help us understand distinct subfield functions along the anterior-posterior axis?

3) In our analyses in patients with atrophy, we noticed that the results between FS dev. and FS v.6.0 were relatively more consistent for patients with less severe atrophy, but inconsistent for patients with more severe atrophy. In your opinion, would you suggest that we use the FS dev to segment the hippocampus in patients with atrophy, or stay with the stable FS v.6.0 version at this moment, since those subfields are already small?

4) For the latest FS dev version, may I know whether you have any plans to further consolidate/optimize the algorithm, or it can be considered as the stable version?

Apologize for so many questions, but I am really interested in this method and find it very helpful.Greatly appreciate your help and reply in advance!

Best,

Liwen

On Monday, 23 September 2019, 8:04:13 pm GMT+8, Iglesias Gonzalez, Juan E. <JIGLESIASGONZALEZ@mgh.harvard.edu> wrote:

Dear Liwen,

We rebuilt the atlas and there were also minor changes in the code, so the volume won’t be exactly the same.

Having said that, if you plot 6 vs dev volumes, they should be highly correlated.

Cheers,

Juan Eugenio Iglesias

Senior research fellow

CMIC (UCL), MGH (HMS) and CSAIL (MIT)

http://www.jeiglesias.com

From: <freesurfer-bounces@nmr.mgh.harvard.edu> on behalf of Liwen Zhang <lw.zhang@yahoo.com>
Reply-To: Freesurfer support list <freesurfer@nmr.mgh.harvard.edu>
Date: Sunday, 22 September 2019 at 10:40
To: "freesurfer@nmr.mgh.harvard.edu" <freesurfer@nmr.mgh.harvard.edu>
Subject: [Freesurfer] help! Hippocampal subfield segmentation with FS development version

External Email - Use Caution

Hello freesurfer experts,

I previously applied hippocampal subfield segmentation using FS v.6.0. I am now trying to devide subfields into head/body where applicable using FS development version (latest version). However, I encountered the following problems/quesetions:

1) X represents certain subfield (volume)

X_head + X_body (from FS dev) ~= X (from FS v.6.0)

What possibly causes this volume discrepancy between FS dev and FS v.6.0? Is the head/body segmentation template in FS dev based on the same ex-vivo template as used in FS v.6.0?

2) Related to 1), subfield volume as obtained from FS v.6.0 generally larger than that from FS dev (not for all subfields though)

I processed T1 images with FS v.6.0 and then applied hippocampal segmentation with both FS v.6.0 and FS dev.

Any suggestions/comments would be greatly appreicated!

Thank you for your time and help in advance!

Best,

Liwen

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