Hippocampal subfields total volume differences

List overview All Threads
Download

newer

older

mri_ca_label with high resolution...

Running Several Versions of fsfast...

Mojmír Vinkler

13 Jul 2016 13 Jul '16

11:07 a.m.

Hi experts,

I'm using Freesurfer 6.0 dev version (Mac) to perform hippocampal subfields segmentation. I only need to measure total hippocampus volume, but measurements from basic `recon-all` were too volatile and I was hoping that using subfields with T2 modality would reduce variance. However, volume measured by hippocampal subfields is significantly different from volume in aseg.stats.

Here are volumes for right hippocampus from various methods: aseg.stats - 3312.8 rh.hippoSfVolumes-T1.v10.txt (Mode A - only T1 used) - 2883.8 rh.hippoSfVolumes-T2.v10.txt (Mode B - using only additional scan) - 2503.97 rh.hippoSfVolumes-T1-T1T2.v10.txt (Mode B - Multispectral segmentation) - 2596.09

I checked labeling visually and there were no obvious problems. Is such a large difference common? Am I missing something? What number should be most trusted?

Thanks for any hints! Mojmir

Attachments:

attachment.html (text/html — 2.7 KB)

Show replies by date

Iglesias, Eugenio

13 Jul 13 Jul

11:14 a.m.

Hi Mojmir, ASEG typically gives a larger estimate of the volume, though highly correlated with that from the subfield package. We have found the measurements from the subfield package to be a bit more reliable, though. Cheers, Eugenio

Juan Eugenio Iglesias Translational Imaging Group University College London http://www.jeiglesias.com http://cmictig.cs.ucl.ac.uk/

On 13 Jul 2016, at 16:07, Mojmír Vinkler <mojmir.vinkler@gmail.commailto:mojmir.vinkler@gmail.com> wrote:

Hi experts,

I checked labeling visually and there were no obvious problems. Is such a large difference common? Am I missing something? What number should be most trusted?

Thanks for any hints! Mojmir _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edumailto:Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer

Mojmír Vinkler

15 Jul 15 Jul

4 a.m.

Hi Juan,

Thanks! You're right, they really are tightly correlated (although their absolute values differ by 30%). We tried it on couple of subjects, but didn't find subfields to be significantly better than aseg stats though. By the way, is Freesurfer capable of using multiple scans from the same subject (longitudinal study, same modality) to make more precise measurements? For example by using T1 from the same subject / different time instead of T2 in subfields?

Thanks! Mojmir

On Wed, Jul 13, 2016 at 5:15 PM Iglesias, Eugenio e.iglesias@ucl.ac.uk wrote:

...

Hi Mojmir, ASEG typically gives a larger estimate of the volume, though highly correlated with that from the subfield package. We have found the measurements from the subfield package to be a bit more reliable, though. Cheers, Eugenio

Juan Eugenio Iglesias Translational Imaging Group University College London http://www.jeiglesias.com http://cmictig.cs.ucl.ac.uk/

On 13 Jul 2016, at 16:07, Mojmír Vinkler mojmir.vinkler@gmail.com wrote:

Hi experts,

I'm using Freesurfer 6.0 dev version (Mac) to perform hippocampal subfields segmentation. I only need to measure total hippocampus volume, but measurements from basic `recon-all` were too volatile and I was hoping that using subfields with T2 modality would reduce variance. However, volume measured by hippocampal subfields is significantly different from volume in aseg.stats.

Here are volumes for right hippocampus from various methods: aseg.stats - 3312.8 rh.hippoSfVolumes-T1.v10.txt (Mode A - only T1 used) - 2883.8 rh.hippoSfVolumes-T2.v10.txt (Mode B - using only additional scan) - 2503.97 rh.hippoSfVolumes-T1-T1T2.v10.txt (Mode B - Multispectral segmentation)

2596.09

I checked labeling visually and there were no obvious problems. Is such a large difference common? Am I missing something? What number should be most trusted?

Thanks for any hints! Mojmir

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

The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.

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

The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.

Iglesias, Eugenio

6:32 a.m.

Dear Mojmir, using ~400 subjects, we found that the subfield module was a bit better than the ASEG volumes at discriminating Alzheimer’s patients from controls (see the Neuorimage paper). But, in any case, the difference is small. Regarding longitudinal segmentation: we just got a paper about this accepted, see http://www.nmr.mgh.harvard.edu/~iglesias/pdf/Neuroimage_2016_longitudinal.pd... I still need to find time to clean up the code and put it up on FreeSurfer. For the time being, you can run the subfield module on the longitudinally analyzed recons (method “L-INIT” in the paper). Cheers, Eugenio

Juan Eugenio Iglesias Translational Imaging Group University College London http://www.jeiglesias.com http://cmictig.cs.ucl.ac.uk/

On 15 Jul 2016, at 09:00, Mojmír Vinkler <mojmir.vinkler@gmail.commailto:mojmir.vinkler@gmail.com> wrote:

Hi Juan,

Thanks! Mojmir

On Wed, Jul 13, 2016 at 5:15 PM Iglesias, Eugenio <e.iglesias@ucl.ac.ukmailto:e.iglesias@ucl.ac.uk> wrote: Hi Mojmir, ASEG typically gives a larger estimate of the volume, though highly correlated with that from the subfield package. We have found the measurements from the subfield package to be a bit more reliable, though. Cheers, Eugenio

Juan Eugenio Iglesias Translational Imaging Group University College London http://www.jeiglesias.com http://www.jeiglesias.com/ http://cmictig.cs.ucl.ac.uk/

On 13 Jul 2016, at 16:07, Mojmír Vinkler <mojmir.vinkler@gmail.commailto:mojmir.vinkler@gmail.com> wrote:

Hi experts,

I checked labeling visually and there were no obvious problems. Is such a large difference common? Am I missing something? What number should be most trusted?

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

The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail. _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edumailto:Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer

Mojmír Vinkler

26 Jul 26 Jul

2:50 a.m.

Wow, longitudinal segmentation is absolutely awesome! It reduced our measurement error by more than 30%. I can't wait when the hippocampal longitudinal segmentation is ready.

By the way, I want to run (cross-sectional) hippocampal subfields on longitudinal data using additional T2 with lower resolution and I was wondering if I should use multispectral segmentation

recon-all -s <subject_name> -hippocampal-subfields-T1T2 <file name of additional scan> <analysisID>

or just additional scan

recon-all -s <subject_name> -hippocampal-subfields-T2 <file name of additional scan> <analysisID>

Since my T1 and T2 scans are quite well aligned, I wanted to use -hippocampal-subfields-T2, but I saw you recommending T1T2 in mailing list. Thanks!

(I'm also blocked by a bug with .ctf files https://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg48412.html, but I'm fine with waiting for fix)

Thanks! Mojmir

On Fri, Jul 15, 2016 at 12:33 PM Iglesias, Eugenio e.iglesias@ucl.ac.uk wrote:

...

Dear Mojmir, using ~400 subjects, we found that the subfield module was a bit better than the ASEG volumes at discriminating Alzheimer’s patients from controls (see the Neuorimage paper). But, in any case, the difference is small. Regarding longitudinal segmentation: we just got a paper about this accepted, see

http://www.nmr.mgh.harvard.edu/~iglesias/pdf/Neuroimage_2016_longitudinal.pd... I still need to find time to clean up the code and put it up on FreeSurfer. For the time being, you can run the subfield module on the longitudinally analyzed recons (method “L-INIT” in the paper). Cheers, Eugenio

Juan Eugenio Iglesias Translational Imaging Group University College London http://www.jeiglesias.com http://cmictig.cs.ucl.ac.uk/

On 15 Jul 2016, at 09:00, Mojmír Vinkler mojmir.vinkler@gmail.com wrote:

Hi Juan,

Thanks! You're right, they really are tightly correlated (although their absolute values differ by 30%). We tried it on couple of subjects, but didn't find subfields to be significantly better than aseg stats though. By the way, is Freesurfer capable of using multiple scans from the same subject (longitudinal study, same modality) to make more precise measurements? For example by using T1 from the same subject / different time instead of T2 in subfields?

Thanks! Mojmir

On Wed, Jul 13, 2016 at 5:15 PM Iglesias, Eugenio e.iglesias@ucl.ac.uk wrote:

...
Hi Mojmir, ASEG typically gives a larger estimate of the volume, though highly correlated with that from the subfield package. We have found the measurements from the subfield package to be a bit more reliable, though. Cheers, Eugenio

Juan Eugenio Iglesias Translational Imaging Group University College London http://www.jeiglesias.com http://cmictig.cs.ucl.ac.uk/

On 13 Jul 2016, at 16:07, Mojmír Vinkler mojmir.vinkler@gmail.com wrote:

Hi experts,

I'm using Freesurfer 6.0 dev version (Mac) to perform hippocampal subfields segmentation. I only need to measure total hippocampus volume, but measurements from basic `recon-all` were too volatile and I was hoping that using subfields with T2 modality would reduce variance. However, volume measured by hippocampal subfields is significantly different from volume in aseg.stats.

Here are volumes for right hippocampus from various methods: aseg.stats - 3312.8 rh.hippoSfVolumes-T1.v10.txt (Mode A - only T1 used) - 2883.8 rh.hippoSfVolumes-T2.v10.txt (Mode B - using only additional scan) - 2503.97 rh.hippoSfVolumes-T1-T1T2.v10.txt (Mode B - Multispectral segmentation)

2596.09

I checked labeling visually and there were no obvious problems. Is such a large difference common? Am I missing something? What number should be most trusted?

Thanks for any hints! Mojmir

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

The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.

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

The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.

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

The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.

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

The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.

3600

Age (days ago)

3613

Last active (days ago)

freesurfer@nmr.mgh.harvard.edu

4 comments

2 participants

tags (0)

participants (2)

Iglesias, Eugenio
Mojmír Vinkler