Dear Experts,
I have some related questions:
1) Am I right in thinking that the default settings for recon-all produce aparc files where the areas are based on white matter and not pia ? 2) Therefore am I right in assuming that when I use aparcstats2table across my subjects using "--meas area", the area is also going to be based on white matter unless I redo the processing using mris_anatomical with a "pial" flag 3) If the above is correct, why is the default to measure areas based on white matter - intuitively this does not make sense to me or seem as useful as gray matter ROI areas (particularly as from my understanding there are further assumptions made in the algorithsm that does the labelling of underlying white matter from the cortex) - are their papers where white matter areas are being reportted ? 4) In a vertex wise analysis of area (using newer version of mris_preproc) what nuisance factors are people using - does it make sense to use total brain surface area (rather than say ICV as one might use in thickness) and in which case does it make a difference to use total pial or white matter area ? 5) Finally what do the dependent measures - intensity_deep/intensity_superficial, intensity_deep.mgz/intensity_superficial.mgz and white K/H measure in qdec ...
Thanks.
Mahinda
Hi,
Sorry to re-post but I wanted to amend my questions on the basis of further reading I have done in the forum. Ignore questions 1 to 3.
Instead ....
4) In a vertex wise analysis of area (using newer version of mris_preproc) what nuisance factors are people using - does it make sense to use total brain surface area (rather than say ICV as one might use in thickness) and in which case does it make a difference to use total pial or white matter area ?
I have read some of the discussions on the forum about this, and some of the references posted. I wonder whether someone might tell me whether they agree with my thinking. I have 2 groups (controls and a patient gp) - I am interested in the global as well as local effects of group on my dependent measures (thickness, area, volume) and hence have only included age and ICV as nuisance factors in my vertex wise analysis - is this reasonble ? - if I was more interested in local/regional effects would it be then reasonable to include mean hemispheric thickness, mean GM volume, or mean GM/WM area as nuisance factors in a vertex wise analysis ... Furthermore, am I right in thinking if I was carrying out an ROI based analysis of any of these dependent measures then by definition I should include the appropriate correction factor to ensure that I am not just seeing a global effect ?
5) What do the dependent measures - intensity_deep/intensity_superficial, intensity_deep.mgz/intensity_superficial.mgz and white K/H measure in qdec ...
6) An additional question - I note in the recon-all.log file that mris_anatomical_stats is run with the flags "-mgz -cortex ... - what are these flags for ?
Thanks.
Mahinda
On Mon, May 21, 2012 at 8:03 PM, Mahinda Yogarajah y.mahinda@gmail.comwrote:
Dear Experts,
I have some related questions:
- Am I right in thinking that the default settings for recon-all produce
aparc files where the areas are based on white matter and not pia ? 2) Therefore am I right in assuming that when I use aparcstats2table across my subjects using "--meas area", the area is also going to be based on white matter unless I redo the processing using mris_anatomical with a "pial" flag 3) If the above is correct, why is the default to measure areas based on white matter - intuitively this does not make sense to me or seem as useful as gray matter ROI areas (particularly as from my understanding there are further assumptions made in the algorithsm that does the labelling of underlying white matter from the cortex) - are their papers where white matter areas are being reportted ? 4) In a vertex wise analysis of area (using newer version of mris_preproc) what nuisance factors are people using - does it make sense to use total brain surface area (rather than say ICV as one might use in thickness) and in which case does it make a difference to use total pial or white matter area ? 5) Finally what do the dependent measures - intensity_deep/intensity_superficial, intensity_deep.mgz/intensity_superficial.mgz and white K/H measure in qdec ...
Thanks.
Mahinda
On 05/22/2012 04:18 PM, Mahinda Yogarajah wrote:
Hi,
Sorry to re-post but I wanted to amend my questions on the basis of further reading I have done in the forum. Ignore questions 1 to 3.
Instead ....
- In a vertex wise analysis of area (using newer version of
mris_preproc) what nuisance factors are people using - does it make sense to use total brain surface area (rather than say ICV as one might use in thickness) and in which case does it make a difference to use total pial or white matter area ?
I have read some of the discussions on the forum about this, and some of the references posted. I wonder whether someone might tell me whether they agree with my thinking. I have 2 groups (controls and a patient gp) - I am interested in the global as well as local effects of group on my dependent measures (thickness, area, volume) and hence have only included age and ICV as nuisance factors in my vertex wise analysis - is this reasonble ? - if I was more interested in local/regional effects would it be then reasonable to include mean hemispheric thickness, mean GM volume, or mean GM/WM area as nuisance factors in a vertex wise analysis ... Furthermore, am I right in thinking if I was carrying out an ROI based analysis of any of these dependent measures then by definition I should include the appropriate correction factor to ensure that I am not just seeing a global effect ?
It probably does not make sense to use ICV for surface area. People sometimes use total surface area, but I think it is an open question (as is the exact method for correction: nuisance regressor or scaling). But accounting for the global effect should reveal local effects by definition. And, yes, you would want to do the same thing for your ROI analysis.
- What do the dependent measures -
intensity_deep/intensity_superficial, intensity_deep.mgz/intensity_superficial.mgz and white K/H measure in qdec ...
These are curvature measures. I can't remember what the K and H are. Bruce?
- An additional question - I note in the recon-all.log file that
mris_anatomical_stats is run with the flags "-mgz -cortex ... - what are these flags for ?
-mgz just says to look for an mgz file. At one point, we had another format, but that was years ago; but the -mgz remains. -cortex passes the cortical label for computing statistics across all of cortex.
doug
Thanks.
Mahinda
On Mon, May 21, 2012 at 8:03 PM, Mahinda Yogarajah <y.mahinda@gmail.com mailto:y.mahinda@gmail.com> wrote:
Dear Experts, I have some related questions: 1) Am I right in thinking that the default settings for recon-all produce aparc files where the areas are based on white matter and not pia ? 2) Therefore am I right in assuming that when I use aparcstats2table across my subjects using "--meas area", the area is also going to be based on white matter unless I redo the processing using mris_anatomical with a "pial" flag 3) If the above is correct, why is the default to measure areas based on white matter - intuitively this does not make sense to me or seem as useful as gray matter ROI areas (particularly as from my understanding there are further assumptions made in the algorithsm that does the labelling of underlying white matter from the cortex) - are their papers where white matter areas are being reportted ? 4) In a vertex wise analysis of area (using newer version of mris_preproc) what nuisance factors are people using - does it make sense to use total brain surface area (rather than say ICV as one might use in thickness) and in which case does it make a difference to use total pial or white matter area ? 5) Finally what do the dependent measures - intensity_deep/intensity_superficial, intensity_deep.mgz/intensity_superficial.mgz and white K/H measure in qdec ... Thanks. Mahinda
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
intensity.deep and superficial are estimates of the intensity of the bottom of the cortex (deep) and the top of the cortex (superficial). These aren't generated by default, although we do have some tools for doing so.
cheers Bruce On Tue, 22 May 2012, Douglas N Greve wrote:
On 05/22/2012 04:18 PM, Mahinda Yogarajah wrote:
Hi,
Sorry to re-post but I wanted to amend my questions on the basis of further reading I have done in the forum. Ignore questions 1 to 3.
Instead ....
- In a vertex wise analysis of area (using newer version of
mris_preproc) what nuisance factors are people using - does it make sense to use total brain surface area (rather than say ICV as one might use in thickness) and in which case does it make a difference to use total pial or white matter area ?
I have read some of the discussions on the forum about this, and some of the references posted. I wonder whether someone might tell me whether they agree with my thinking. I have 2 groups (controls and a patient gp) - I am interested in the global as well as local effects of group on my dependent measures (thickness, area, volume) and hence have only included age and ICV as nuisance factors in my vertex wise analysis - is this reasonble ? - if I was more interested in local/regional effects would it be then reasonable to include mean hemispheric thickness, mean GM volume, or mean GM/WM area as nuisance factors in a vertex wise analysis ... Furthermore, am I right in thinking if I was carrying out an ROI based analysis of any of these dependent measures then by definition I should include the appropriate correction factor to ensure that I am not just seeing a global effect ?
It probably does not make sense to use ICV for surface area. People sometimes use total surface area, but I think it is an open question (as is the exact method for correction: nuisance regressor or scaling). But accounting for the global effect should reveal local effects by definition. And, yes, you would want to do the same thing for your ROI analysis.
- What do the dependent measures -
intensity_deep/intensity_superficial, intensity_deep.mgz/intensity_superficial.mgz and white K/H measure in qdec ...
These are curvature measures. I can't remember what the K and H are. Bruce?
- An additional question - I note in the recon-all.log file that
mris_anatomical_stats is run with the flags "-mgz -cortex ... - what are these flags for ?
-mgz just says to look for an mgz file. At one point, we had another format, but that was years ago; but the -mgz remains. -cortex passes the cortical label for computing statistics across all of cortex.
doug
Thanks.
Mahinda
On Mon, May 21, 2012 at 8:03 PM, Mahinda Yogarajah <y.mahinda@gmail.com mailto:y.mahinda@gmail.com> wrote:
Dear Experts, I have some related questions: 1) Am I right in thinking that the default settings for recon-all produce aparc files where the areas are based on white matter and not pia ? 2) Therefore am I right in assuming that when I use aparcstats2table across my subjects using "--meas area", the area is also going to be based on white matter unless I redo the processing using mris_anatomical with a "pial" flag 3) If the above is correct, why is the default to measure areas based on white matter - intuitively this does not make sense to me or seem as useful as gray matter ROI areas (particularly as from my understanding there are further assumptions made in the algorithsm that does the labelling of underlying white matter from the cortex) - are their papers where white matter areas are being reportted ? 4) In a vertex wise analysis of area (using newer version of mris_preproc) what nuisance factors are people using - does it make sense to use total brain surface area (rather than say ICV as one might use in thickness) and in which case does it make a difference to use total pial or white matter area ? 5) Finally what do the dependent measures - intensity_deep/intensity_superficial, intensity_deep.mgz/intensity_superficial.mgz and white K/H measure in qdec ... Thanks. Mahinda
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Bruce Fischl -
Douglas N Greve -
Mahinda Yogarajah