I see. Well, you will probably want some combination of cortical thickness maps and subcortical volumes at the least, although of course other information might be important as well (connectivity, myelin content, etc…). These things have different shapes (in the python sense) so you’ll have to sort out a network architecture that lets you combine them

Cheers

Bruce

From: freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mgh.harvard.edu> On Behalf Of REEM ABU BAKR BAHATHIQ
Sent: Thursday, February 24, 2022 11:28 AM
To: Freesurfer support list <freesurfer@nmr.mgh.harvard.edu>
Subject: Re: [Freesurfer] recon-all outpputs for deep learning

        External Email - Use Caution        

Sorry for not making it clear

you are right.

Yes, I want to diagnose autism spectrum disorder from brain images using structural measurements such as the gray matter volumes of regions or at the voxel level.
Then, since deep learning is a black box that simply cannot know what it has learned, there is a so-called "unit of attention" that can point to the most important areas involved in the decision-making of the deep learning model.

في الخميس، 24 فبراير 2022 في 6:19 م تمت كتابة ما يلي بواسطة Fischl, Bruce <BFISCHL@mgh.harvard.edu>:

Sorry, can you explain a bit more about what you want to do. What does “demonstrate morphological variation” mean? Just quantify variability? Are there any variables that you are correlating it with (e.g. disease status, cognitive measure, etc…)? I don’t understand what you are trying to do

Cheers

Bruce

From: freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mgh.harvard.edu> On Behalf Of REEM ABU BAKR BAHATHIQ
Sent: Thursday, February 24, 2022 7:37 AM
To: Freesurfer support list <freesurfer@nmr.mgh.harvard.edu>
Subject: [Freesurfer] recon-all outpputs for deep learning

        External Email - Use Caution        

Hello Freesurfer experts

I want to apply deep learning to structural brain images processed by Recon-all. My goal is to demonstrate the morphological variation(e.g., gray matter volumes or cortical thickness) of brain regions in structural magnetic resonance imaging using convolutional neural networks (CNN). Then interpret those models on the basis of the region of interest.

 But I'm still distracted by the model input . I searched before but still couldn't decide which Recon-all output I should use to perform the task. I think I'll use the folder(mri) but I'm not sure which file.

can any one guide me .

Thanks,

_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
MailScanner has detected a possible fraud attempt from "secure-web.cisco.com" claiming to be https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer