mean thickness covariate, mean area covariate, & mri_anatomical_stats for multiple subjects
Hello freesurfers,
It appears that prior discussions on the listserv have recommended using mean cortical thickness (rather than ICV) as a covariate for thickness analyses (predominantly recommended by Michael Harms). I have three questions about this:
1. Is there a paper that I could cite that recommends using mean cortical thickness rather than ICV? 2. Would the same logic be applied to surface area analyses? i.e. would it make more sense to use mean surface area as a covariate in surface area analyses? If so, which mean surface area calculation should be used? mri_anatomical_stats can produce both pial and white matter mean surface area stats. 3. Is there a way to run mri_anatomical_stats on multiple subjects at once and write to a tablefile (similar to asegstats2table output)?
Thanks!
Laura.
Hi Laura,
- Is there a paper that I could cite that recommends using mean
cortical thickness rather than ICV?
If it helps, we used cortical thickness and area as covariate for the respective analysis of regional thickness and area. Brain volume, which is more closely related to ICV, correlates well with area, but not with thickness. We computed a global thickness average by weighting the thickness of each region by their respective areas. The paper is this: http://surfer.nmr.mgh.harvard.edu/ftp/articles/Winkler2010_Neuroimage.pdf
- Would the same logic be applied to surface area analyses? i.e.
would it make more sense to use mean surface area as a covariate in surface area analyses? If so, which mean surface area calculation should be used? mri_anatomical_stats can produce both pial and white matter mean surface area stats.
Yes, I think so. It seems more logical to have a global measurement of area in the model than a measurement of brain volume. On the other hand, area and thickness are not correlated one to another (as shown in the paper above and also in Panizzon et al, 2009, in Cereb Cortex). I don't think there is a clear answer on which, pial or white, should be used. I'd probably go with the white, as I think it may be more robust to image quality, but I admit this is a rather weak justification and if the images are good, perhaps the pial could be just as good, despite the fact that it somewhat depends on the white for its construction.
- Is there a way to run mri_anatomical_stats on multiple subjects at
once and write to a tablefile (similar to asegstats2table output)?
I think you can use aparcstats2table, then add up all regions in a spreadsheet (or even with awk/gawk). Alternatively, you can use "grep" to pick the WhiteSurfArea for each hemisphere from the ?h.aparc.stats file for each subject.
Hope this helps!
All the best,
Anderson
Hi Anderson,
Thanks, that reference is particularly helpful.
Re: the usage of the white versus pial area question - I believe that the default area calculation in freesurfer is the white surface area, so unless one specifies the pial in calculations, the standard surface area output for surface area by parcellation will be white. This suggests the use of the global measurement of white surface area as a covariate would be an appropriate, whereas if one was specifically using pial surface area in the aparc calculations, it may make more sense to use the global measure of pial surface area as a covariate, correct? As for which one to use in analysis, I'm not sure - conceptually it might be that the pial surface area is more sensitive to atrophy but I don't know if that is born out in the data...
Laura.
On Sat, Mar 9, 2013 at 11:49 AM, Anderson M. Winkler <winkler@fmrib.ox.ac.uk
wrote:
Hi Laura,
- Is there a paper that I could cite that recommends using mean
cortical thickness rather than ICV?
If it helps, we used cortical thickness and area as covariate for the respective analysis of regional thickness and area. Brain volume, which is more closely related to ICV, correlates well with area, but not with thickness. We computed a global thickness average by weighting the thickness of each region by their respective areas. The paper is this: http://surfer.nmr.mgh.harvard.edu/ftp/articles/Winkler2010_Neuroimage.pdf
- Would the same logic be applied to surface area analyses? i.e.
would it make more sense to use mean surface area as a covariate in surface area analyses? If so, which mean surface area calculation should be used? mri_anatomical_stats can produce both pial and white matter mean surface area stats.
Yes, I think so. It seems more logical to have a global measurement of area in the model than a measurement of brain volume. On the other hand, area and thickness are not correlated one to another (as shown in the paper above and also in Panizzon et al, 2009, in Cereb Cortex). I don't think there is a clear answer on which, pial or white, should be used. I'd probably go with the white, as I think it may be more robust to image quality, but I admit this is a rather weak justification and if the images are good, perhaps the pial could be just as good, despite the fact that it somewhat depends on the white for its construction.
- Is there a way to run mri_anatomical_stats on multiple subjects at
once and write to a tablefile (similar to asegstats2table output)?
I think you can use aparcstats2table, then add up all regions in a spreadsheet (or even with awk/gawk). Alternatively, you can use "grep" to pick the WhiteSurfArea for each hemisphere from the ?h.aparc.stats file for each subject.
Hope this helps!
All the best,
Anderson
Hi Laura, I think yes, sounds right. Maybe run both ways and see if the results agree or differ. All the best, Anderson
2013/3/9 Laura M. Tully tully.laura@googlemail.com
Hi Anderson,
Thanks, that reference is particularly helpful.
Re: the usage of the white versus pial area question - I believe that the default area calculation in freesurfer is the white surface area, so unless one specifies the pial in calculations, the standard surface area output for surface area by parcellation will be white. This suggests the use of the global measurement of white surface area as a covariate would be an appropriate, whereas if one was specifically using pial surface area in the aparc calculations, it may make more sense to use the global measure of pial surface area as a covariate, correct? As for which one to use in analysis, I'm not sure - conceptually it might be that the pial surface area is more sensitive to atrophy but I don't know if that is born out in the data...
Laura.
On Sat, Mar 9, 2013 at 11:49 AM, Anderson M. Winkler < winkler@fmrib.ox.ac.uk> wrote:
Hi Laura,
- Is there a paper that I could cite that recommends using mean
cortical thickness rather than ICV?
If it helps, we used cortical thickness and area as covariate for the respective analysis of regional thickness and area. Brain volume, which is more closely related to ICV, correlates well with area, but not with thickness. We computed a global thickness average by weighting the thickness of each region by their respective areas. The paper is this: http://surfer.nmr.mgh.harvard.edu/ftp/articles/Winkler2010_Neuroimage.pdf
- Would the same logic be applied to surface area analyses? i.e.
would it make more sense to use mean surface area as a covariate in surface area analyses? If so, which mean surface area calculation should be used? mri_anatomical_stats can produce both pial and white matter mean surface area stats.
Yes, I think so. It seems more logical to have a global measurement of area in the model than a measurement of brain volume. On the other hand, area and thickness are not correlated one to another (as shown in the paper above and also in Panizzon et al, 2009, in Cereb Cortex). I don't think there is a clear answer on which, pial or white, should be used. I'd probably go with the white, as I think it may be more robust to image quality, but I admit this is a rather weak justification and if the images are good, perhaps the pial could be just as good, despite the fact that it somewhat depends on the white for its construction.
- Is there a way to run mri_anatomical_stats on multiple subjects
at once and write to a tablefile (similar to asegstats2table output)?
I think you can use aparcstats2table, then add up all regions in a spreadsheet (or even with awk/gawk). Alternatively, you can use "grep" to pick the WhiteSurfArea for each hemisphere from the ?h.aparc.stats file for each subject.
Hope this helps!
All the best,
Anderson
--
Laura M. Tully, MA Social Neuroscience & Psychopathology, Harvard University Center for the Assessment and Prevention of Prodromal States, UCLA Semel Institute of Neuroscience ltully@mednet.ucla.edu ltully@fas.harvard.edu 310-267-0170 -- My musings as a young clinical scientist: http://theclinicalbrain.blogspot.com/ Follow me on Twitter: @tully_laura
Hi Laura, If you're looking for another reference that has used this approach, you could see our 2010 paper: http://www.ncbi.nlm.nih.gov/pubmed/20118463
cheers, -MH
Thank you Michael!
LT
On Mon, Mar 11, 2013 at 8:54 AM, Michael Harms mharms@conte.wustl.eduwrote:
Hi Laura, If you're looking for another reference that has used this approach, you could see our 2010 paper: http://www.ncbi.nlm.nih.gov/pubmed/20118463
cheers, -MH
-- Michael Harms, Ph.D.
Conte Center for the Neuroscience of Mental Disorders Washington University School of Medicine Department of Psychiatry, Box 8134 660 South Euclid Ave. Tel: 314-747-6173 St. Louis, MO 63110 Email: mharms@wustl.edu
From: "Laura M. Tully" tully.laura@googlemail.com Date: Saturday, March 9, 2013 5:58 PM To: "Anderson M. Winkler" winkler@fmrib.ox.ac.uk Cc: free freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] mean thickness covariate, mean area covariate, & mri_anatomical_stats for multiple subjects
Hi Anderson,
Thanks, that reference is particularly helpful.
Re: the usage of the white versus pial area question - I believe that the default area calculation in freesurfer is the white surface area, so unless one specifies the pial in calculations, the standard surface area output for surface area by parcellation will be white. This suggests the use of the global measurement of white surface area as a covariate would be an appropriate, whereas if one was specifically using pial surface area in the aparc calculations, it may make more sense to use the global measure of pial surface area as a covariate, correct? As for which one to use in analysis, I'm not sure - conceptually it might be that the pial surface area is more sensitive to atrophy but I don't know if that is born out in the data...
Laura.
On Sat, Mar 9, 2013 at 11:49 AM, Anderson M. Winkler < winkler@fmrib.ox.ac.uk> wrote:
Hi Laura,
- Is there a paper that I could cite that recommends using mean
cortical thickness rather than ICV?
If it helps, we used cortical thickness and area as covariate for the respective analysis of regional thickness and area. Brain volume, which is more closely related to ICV, correlates well with area, but not with thickness. We computed a global thickness average by weighting the thickness of each region by their respective areas. The paper is this: http://surfer.nmr.mgh.harvard.edu/ftp/articles/Winkler2010_Neuroimage.pdf
- Would the same logic be applied to surface area analyses? i.e.
would it make more sense to use mean surface area as a covariate in surface area analyses? If so, which mean surface area calculation should be used? mri_anatomical_stats can produce both pial and white matter mean surface area stats.
Yes, I think so. It seems more logical to have a global measurement of area in the model than a measurement of brain volume. On the other hand, area and thickness are not correlated one to another (as shown in the paper above and also in Panizzon et al, 2009, in Cereb Cortex). I don't think there is a clear answer on which, pial or white, should be used. I'd probably go with the white, as I think it may be more robust to image quality, but I admit this is a rather weak justification and if the images are good, perhaps the pial could be just as good, despite the fact that it somewhat depends on the white for its construction.
- Is there a way to run mri_anatomical_stats on multiple subjects
at once and write to a tablefile (similar to asegstats2table output)?
I think you can use aparcstats2table, then add up all regions in a spreadsheet (or even with awk/gawk). Alternatively, you can use "grep" to pick the WhiteSurfArea for each hemisphere from the ?h.aparc.stats file for each subject.
Hope this helps!
All the best,
Anderson
--
Laura M. Tully, MA Social Neuroscience & Psychopathology, Harvard University Center for the Assessment and Prevention of Prodromal States, UCLA Semel Institute of Neuroscience ltully@mednet.ucla.edu ltully@fas.harvard.edu 310-267-0170 -- My musings as a young clinical scientist: http://theclinicalbrain.blogspot.com/ Follow me on Twitter: @tully_laura _______________________________________________ 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@nmr.mgh.harvard.edu
-
Anderson M. Winkler -
Laura M. Tully -
Michael Harms