Hi Jim, and thanks for jumping in Falk. This description is almost all
correct. The one inaccuracy is that the -hires option will downsample
the surface to a number of vertices equal to the number that would have
resulted if recon-all had been run at 1mm. The main difference is that
you will get volume segmentations at the higher resolution with -hires
where as you will get them at 1mm with -conf2hires.
On 9/9/2020 3:13 AM, falk.luesebr...@med.ovgu.de wrote:
External Email - Use Caution
Hi Jim,
The -hires will conform the data to the highest resolution of your
input data, instead of 1mm^3. In case it was an isotropic resolution,
it will stay at native resolution. If it is anisotropic, e.g.
0.3x0.3x0.7, it will be resampled to 0.3mm^3 – which is not really
recommended. The segmentation and surface placement will take place at
that resolution.
The -conf2hires flag will conform your input data to 1mm^3.
Segmentation and initial surface placement will be handled at
conformed (1mm) resolution. Then the initial surface placement is used
at native resolution to refine the surface placement.
With -hires the number of vertices per hemisphere depends on your
input resolution being roughly at 300k at 0.7mm and 500k at 0.5 mm.
With -conf2hires the number of vertices will be around 140k per
hemisphere and, therefore, processing (especially the topologic
correction) will be much faster compared to -hires. The principle
behind -conf2hires was used in the processing pipeline of the HCP
(https://doi.org/10.1016/j.neuroimage.2013.04.127) and has been shown
to improve the segmentation in highly myelinated areas, e.g. around
the visual cortex or central sulcus.
Natalia Zaretskaya and Jon Polimeni
(https://doi.org/10.1016/j.neuroimage.2017.09.060) and I
(https://doi.org/10.1016/j.neuroimage.2012.12.016) could show that the
principles behind the -hires flag yield different results than
downsampling high resolution input data or using 1mm data to being
with. It is hypothesized that using high resolution data with the
-hires flag leads to better surface placement and, therefore, more
accurate cortical thickness measures. However, missing ground truth
always makes it complicated to proof.
To answer your question which is recommended: In the release notes of
FreeSurfer it is written with regards to conf2hires “This was
originally programmed for the HCP. The now-standard hi-res stream
should work just as well, but we kept conf2hires for backwards
compatibility.”
I don’t think someone published a comparison between -hires and
-conf2hires, yet. However, I would use the -hires flag instead of
-conf2hires. I assume the surface placement to be equally good, with
potential benefits towards -hires due to the higher number of
vertices. The drawbacks of -hires (e.g. longer processing time) should
be compensated to some extent by the new denoising feature prior to
the intial surface placement. Personally, I haven’t had much time to
test v7 intensively, though.
Hope this helps.
Best,
Falk
*Von:*freesurfer-boun...@nmr.mgh.harvard.edu
<freesurfer-boun...@nmr.mgh.harvard.edu> *Im Auftrag von *Alexopoulos,
Dimitrios
*Gesendet:* Mittwoch, 9. September 2020 05:27
*An:* freesurfer@nmr.mgh.harvard.edu
*Betreff:* [Freesurfer] -hires vs -conf2hires in 7.1.1.
* External Email - Use Caution *
Does the hires flag still generate segs at the native hi resolution
and place surfaces on the 1mm volumes, whereas conf2hires generates
volumes at 1mm but places surfaces on the hires images?
Which option is recommended and why?
Jim
Get Outlook for Android <https://aka.ms/ghei36>
------------------------------------------------------------------------
The materials in this message are private and may contain Protected
Healthcare Information or other information of a sensitive nature. If
you are not the intended recipient, be advised that any unauthorized
use, disclosure, copying or the taking of any action in reliance on
the contents of this information is strictly prohibited. If you have
received this email in error, please immediately notify the sender via
telephone or return mail.
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
