Dear Doug,
We are trying to compute partial volume correction in DWI images, using your PEtsurf approach, to correct CSF contamination. Afterwards, we will compute whatever DTI metric from the corrected data and project it to surface.
As you suggested in your previous post (https://www.mail-archive.com/ freesurfer@nmr.mgh.harvard.edu/msg52436.html), we used the psf=0, so the algorithm will just perform TFE.
However, we have some concerns. When we perform this correction in symptomatic subjects (e.g AD patients), we have seen that in certain regions (mainly non atrophied areas), the PVC effect is overestimated. If I am not wrong, the WM coefficient obtained from the SGTM takes into account ALL the superficial WM, and thus, will take intro account both atrophied and non atrophied regions. This is important because, unlike PET, the superficial WM diffusion, resembles the GM diffusion and it would be interesting to treat them as one compartment. Therefore, my question is if there is any option in mri_gtmpvc to use a 3-compartiment model (WM+GM, CSF, SKULL) instead of the default 4-compartment.
Once we have the PVC volume, we project our DTI metric to surface using mri_vol2surf with proj-fract 0.5. However, since we used a GM thr of 30% (as in your NI paper), there are several regions with no signal (i.e 0 in the native surface). How do you solve this issue? Smooth of the data? Exclude those vertices from further analysis?
Thank you in advance for your help!! Regards,
Victor M
sorry, that is not possible with the tools as they are. I had no idea someone would be trying to do something like this when I programmed it! I'll keep it in mind if I make modifications ...
On 7/1/17 6:54 AM, Victor Montal Blancafort wrote:
Dear Doug,
We are trying to compute partial volume correction in DWI images, using your PEtsurf approach, to correct CSF contamination. Afterwards, we will compute whatever DTI metric from the corrected data and project it to surface.
As you suggested in your previous post (https://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg52436.html https://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg52436.html), we used the psf=0, so the algorithm will just perform TFE.
However, we have some concerns. When we perform this correction in symptomatic subjects (e.g AD patients), we have seen that in certain regions (mainly non atrophied areas), the PVC effect is overestimated. If I am not wrong, the WM coefficient obtained from the SGTM takes into account ALL the superficial WM, and thus, will take intro account both atrophied and non atrophied regions. This is important because, unlike PET, the superficial WM diffusion, resembles the GM diffusion and it would be interesting to treat them as one compartment. Therefore, my question is if there is any option in mri_gtmpvc to use a 3-compartiment model (WM+GM, CSF, SKULL) instead of the default 4-compartment.
Once we have the PVC volume, we project our DTI metric to surface using mri_vol2surf with proj-fract 0.5. However, since we used a GM thr of 30% (as in your NI paper), there are several regions with no signal (i.e 0 in the native surface). How do you solve this issue? Smooth of the data? Exclude those vertices from further analysis?
Thank you in advance for your help!! Regards,
Victor M
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Douglas Greve -
Victor Montal Blancafort