Hello,
The FS release notes for v3.0.5 state: "mris_anatomical_stats produces incorrect results. There is no work- around. A replacement for Linux is here. This fix will appear in the next release."
Is the entire output of 'mris_anatomical_stats' incorrect, or just one part of it? And, does the problem, whatever it is, apply to any previous stable versions (e.g., v3.0.4)?
thanks, Mike Harms
Mike,
I've updated the release notes to be more specific (and less alarming!) on the problem found with mris_anatomical_stats. The new notes state:
"mris_anatomical_stats effectively ignores the <surface name> command line argument when reporting on its curvature measures, and will always report on the pial surface. This does not impact the results output by mris_anatomical_stats when executed by recon-all, as it reports on the pial surface as-is (the <surface name> option of mris_anatomical_stats is not used by recon-all)."
In short, the results produced by mris_anatomical_stats, as called by recon-all, are correct for all distributed versions of Freesurfer.
It is only if the <surface name> argument is used, and a surface other than pial is specified, that the bug exposes itself.
Sorry for the misinformation.
Nick
On Wed, 2007-05-02 at 10:34 -0500, Michael Harms wrote:
Hello,
The FS release notes for v3.0.5 state: "mris_anatomical_stats produces incorrect results. There is no work- around. A replacement for Linux is here. This fix will appear in the next release."
Is the entire output of 'mris_anatomical_stats' incorrect, or just one part of it? And, does the problem, whatever it is, apply to any previous stable versions (e.g., v3.0.4)?
thanks, Mike Harms
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Hi Nick,
Is this "bug" specific to v3.0.5? Because when I specify white vs. pial surfaces as <surface name> options under v3.0.4, I do get different values for the curvature measures, leading me to think the <surface name> option works fine under v3.0.4 ??
Also, could you please clarify whether the default surface for 'mris_anatomical_stats' is the pial or the white surface? Using v3.0.4, when I run 'mri_anatomical_stats' with no <surface name> option, I get the same values for surface area and curvature as when I specify the WHITE surface. But when I specify the pial surface, the values change relative to the default. This leads me to think that the default curvature and surface area values (as executed by recon-all) are actually for the white surface.
Thanks for the additional clarifications, Mike H.
On Wed, 2007-05-02 at 12:47 -0400, Nick Schmansky wrote:
Mike,
I've updated the release notes to be more specific (and less alarming!) on the problem found with mris_anatomical_stats. The new notes state:
"mris_anatomical_stats effectively ignores the <surface name> command line argument when reporting on its curvature measures, and will always report on the pial surface. This does not impact the results output by mris_anatomical_stats when executed by recon-all, as it reports on the pial surface as-is (the <surface name> option of mris_anatomical_stats is not used by recon-all)."
In short, the results produced by mris_anatomical_stats, as called by recon-all, are correct for all distributed versions of Freesurfer.
It is only if the <surface name> argument is used, and a surface other than pial is specified, that the bug exposes itself.
Sorry for the misinformation.
Nick
On Wed, 2007-05-02 at 10:34 -0500, Michael Harms wrote:
Hello,
The FS release notes for v3.0.5 state: "mris_anatomical_stats produces incorrect results. There is no work- around. A replacement for Linux is here. This fix will appear in the next release."
Is the entire output of 'mris_anatomical_stats' incorrect, or just one part of it? And, does the problem, whatever it is, apply to any previous stable versions (e.g., v3.0.4)?
thanks, Mike Harms
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
Mike,
After some investigation, I've updated the ReleaseNotes page again with more accurate details on this bug (which is in v3.0.5 and all prior versions). See the Known Issues section of v3.0.5:
http://surfer.nmr.mgh.harvard.edu/fswiki/ReleaseNotes
If accurate measures of 'average integrated rectified mean curvature', 'average integrated rectified Gaussian curvature', 'folding index', and 'intrinsic curvature index' are important to you, or if you want to use the <surface name> option of mris_anatomical_stats, then you will need to download the fixed mris_anatomical_stats binary as linked in the ReleaseNotes page (or wait till the next release of Freesurfer).
The default surface for mris_anatomical_stats is the white surface.
Nick
On Wed, 2007-05-02 at 12:26 -0500, Michael Harms wrote:
Hi Nick,
Is this "bug" specific to v3.0.5? Because when I specify white vs. pial surfaces as <surface name> options under v3.0.4, I do get different values for the curvature measures, leading me to think the <surface name> option works fine under v3.0.4 ??
Also, could you please clarify whether the default surface for 'mris_anatomical_stats' is the pial or the white surface? Using v3.0.4, when I run 'mri_anatomical_stats' with no <surface name> option, I get the same values for surface area and curvature as when I specify the WHITE surface. But when I specify the pial surface, the values change relative to the default. This leads me to think that the default curvature and surface area values (as executed by recon-all) are actually for the white surface.
Thanks for the additional clarifications, Mike H.
On Wed, 2007-05-02 at 12:47 -0400, Nick Schmansky wrote:
Mike,
I've updated the release notes to be more specific (and less alarming!) on the problem found with mris_anatomical_stats. The new notes state:
"mris_anatomical_stats effectively ignores the <surface name> command line argument when reporting on its curvature measures, and will always report on the pial surface. This does not impact the results output by mris_anatomical_stats when executed by recon-all, as it reports on the pial surface as-is (the <surface name> option of mris_anatomical_stats is not used by recon-all)."
In short, the results produced by mris_anatomical_stats, as called by recon-all, are correct for all distributed versions of Freesurfer.
It is only if the <surface name> argument is used, and a surface other than pial is specified, that the bug exposes itself.
Sorry for the misinformation.
Nick
On Wed, 2007-05-02 at 10:34 -0500, Michael Harms wrote:
Hello,
The FS release notes for v3.0.5 state: "mris_anatomical_stats produces incorrect results. There is no work- around. A replacement for Linux is here. This fix will appear in the next release."
Is the entire output of 'mris_anatomical_stats' incorrect, or just one part of it? And, does the problem, whatever it is, apply to any previous stable versions (e.g., v3.0.4)?
thanks, Mike Harms
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
freesurfer@nmr.mgh.harvard.edu
-
Michael Harms -
Nick Schmansky