If you make use of the estimated Total Intracranial Volume (eTIV) calculation found in the aseg.stats file of your subject data, then you should be aware that we have recently found a more accurate way to make this calculation. We have found that using the talairach.xfm transform instead of the talairach_with_skull.lta transform, from which the determinant is extracted and multiplied by a scale factor, results in lower variability of the eTIV calc in our test set.
This new way of calculating eTIV will appear in the next release of Freesurfer.
If you want to make use of this new calculation right away, you can download a new version of mri_segstats from here:
ftp://surfer.nmr.mgh.harvard.edu/pub/dist/freesurfer/fixes
and run it using this command:
mri_segstats --subject subjid --etiv-only
and it will output the new eTIV.
Nick
Nick, This is great -- you probably also noticed that the version of the transform used (without or with skull) can make a big difference in the computed scale factor.
Was variability assessed by comparing eTIV values computed from multiple MR sessions from the same subject? Or by comparison to manually traced ICV from a T2w scan? Could you provide some sense of what the actual variability (e.g., coefficient of variation) was for the old vs. new approaches?
Regarding the accuracy of the new eTIV calculation: Since the scale factor is now going to be derived from a transform without the skull, do you have a sense to what degree atrophy and increased sulcal CSF will impact the values? That is, should the new measure be considered more an estimate of "TBV -- total brain volume" (defined as the total volume of everything interior to the outer pial surface, including ventricular CSF), rather than "intracranial volume" per se?
thanks, Mike H.
On Wed, 2009-02-11 at 18:41 -0500, Nick Schmansky wrote:
If you make use of the estimated Total Intracranial Volume (eTIV) calculation found in the aseg.stats file of your subject data, then you should be aware that we have recently found a more accurate way to make this calculation. We have found that using the talairach.xfm transform instead of the talairach_with_skull.lta transform, from which the determinant is extracted and multiplied by a scale factor, results in lower variability of the eTIV calc in our test set.
This new way of calculating eTIV will appear in the next release of Freesurfer.
If you want to make use of this new calculation right away, you can download a new version of mri_segstats from here:
ftp://surfer.nmr.mgh.harvard.edu/pub/dist/freesurfer/fixes
and run it using this command:
mri_segstats --subject subjid --etiv-only
and it will output the new eTIV.
Nick
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Mike,
I've updated our wiki page with most of this info:
http://surfer.nmr.mgh.harvard.edu/fswiki/eTIV
This update, and most of the stats backing it, was done by Martin Reuter here at the NMR Center.
The wiki page shows the difference in variability between the old way of calculating it (talairach_with_skull.lta) and the new (talairach.xfm).
We also look at some longitudinal scans of a control and a patient known to be undergoing atrophy. Results were very good: the eTIV varied the same (less, actually) than the control. Although I think it would take many more example of this to accurately answer your question of whether it is still valid over the _with_skull transform. Maybe Avi knows?
If anything is unclear or you think is missing or lacking from this wiki page, let us know, as we get a lot of questions about eTIV, and we want it to be as good as we think we can get it (at least when using the scheme outlined in the paper).
Nick
On Thu, 2009-02-12 at 12:08 -0600, Michael Harms wrote:
Nick, This is great -- you probably also noticed that the version of the transform used (without or with skull) can make a big difference in the computed scale factor.
Was variability assessed by comparing eTIV values computed from multiple MR sessions from the same subject? Or by comparison to manually traced ICV from a T2w scan? Could you provide some sense of what the actual variability (e.g., coefficient of variation) was for the old vs. new approaches?
Regarding the accuracy of the new eTIV calculation: Since the scale factor is now going to be derived from a transform without the skull, do you have a sense to what degree atrophy and increased sulcal CSF will impact the values? That is, should the new measure be considered more an estimate of "TBV -- total brain volume" (defined as the total volume of everything interior to the outer pial surface, including ventricular CSF), rather than "intracranial volume" per se?
thanks, Mike H.
On Wed, 2009-02-11 at 18:41 -0500, Nick Schmansky wrote:
If you make use of the estimated Total Intracranial Volume (eTIV) calculation found in the aseg.stats file of your subject data, then you should be aware that we have recently found a more accurate way to make this calculation. We have found that using the talairach.xfm transform instead of the talairach_with_skull.lta transform, from which the determinant is extracted and multiplied by a scale factor, results in lower variability of the eTIV calc in our test set.
This new way of calculating eTIV will appear in the next release of Freesurfer.
If you want to make use of this new calculation right away, you can download a new version of mri_segstats from here:
ftp://surfer.nmr.mgh.harvard.edu/pub/dist/freesurfer/fixes
and run it using this command:
mri_segstats --subject subjid --etiv-only
and it will output the new eTIV.
Nick
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Mike,
Another piece of info: the talairach.xfm transform is a registration to an image _with_ skull, to this image:
$FREESURFER_HOME/average/711-2C_as_mni_average_305.4dfp.img
The utility imgreg_4dfp is used to perform this registration. So the eTIV should not be as subject to changes due to atrophy or sulcul depth, as might be the case if our target image did not have skull.
Nick
On Thu, 2009-02-12 at 21:21 -0500, Nick Schmansky wrote:
Mike,
I've updated our wiki page with most of this info:
http://surfer.nmr.mgh.harvard.edu/fswiki/eTIV
This update, and most of the stats backing it, was done by Martin Reuter here at the NMR Center.
The wiki page shows the difference in variability between the old way of calculating it (talairach_with_skull.lta) and the new (talairach.xfm).
We also look at some longitudinal scans of a control and a patient known to be undergoing atrophy. Results were very good: the eTIV varied the same (less, actually) than the control. Although I think it would take many more example of this to accurately answer your question of whether it is still valid over the _with_skull transform. Maybe Avi knows?
If anything is unclear or you think is missing or lacking from this wiki page, let us know, as we get a lot of questions about eTIV, and we want it to be as good as we think we can get it (at least when using the scheme outlined in the paper).
Nick
On Thu, 2009-02-12 at 12:08 -0600, Michael Harms wrote:
Nick, This is great -- you probably also noticed that the version of the transform used (without or with skull) can make a big difference in the computed scale factor.
Was variability assessed by comparing eTIV values computed from multiple MR sessions from the same subject? Or by comparison to manually traced ICV from a T2w scan? Could you provide some sense of what the actual variability (e.g., coefficient of variation) was for the old vs. new approaches?
Regarding the accuracy of the new eTIV calculation: Since the scale factor is now going to be derived from a transform without the skull, do you have a sense to what degree atrophy and increased sulcal CSF will impact the values? That is, should the new measure be considered more an estimate of "TBV -- total brain volume" (defined as the total volume of everything interior to the outer pial surface, including ventricular CSF), rather than "intracranial volume" per se?
thanks, Mike H.
On Wed, 2009-02-11 at 18:41 -0500, Nick Schmansky wrote:
If you make use of the estimated Total Intracranial Volume (eTIV) calculation found in the aseg.stats file of your subject data, then you should be aware that we have recently found a more accurate way to make this calculation. We have found that using the talairach.xfm transform instead of the talairach_with_skull.lta transform, from which the determinant is extracted and multiplied by a scale factor, results in lower variability of the eTIV calc in our test set.
This new way of calculating eTIV will appear in the next release of Freesurfer.
If you want to make use of this new calculation right away, you can download a new version of mri_segstats from here:
ftp://surfer.nmr.mgh.harvard.edu/pub/dist/freesurfer/fixes
and run it using this command:
mri_segstats --subject subjid --etiv-only
and it will output the new eTIV.
Nick
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