Dear Experts,
I had 2 questions with regard to vertex wise analysis using qdec.
1) How should one interpret the dependent variables area/pial area and volume when comparing 2 groups with qdec and a vertex wise (as opposed to ROI based) analysis - or are they uninterpretable in this context ? I read briefly about areal as an alternative measure - could someone explain the difference, and how thickness, volume and area related to one another ?
2) When running monte-carlo cluster wise multiple correction in qdec - if one finds a significant cluster is it ok to view them within qdec by highlighting the "mc-z ... " file in the analysis result box, and then changing the minimum threshold to whatever value one selected for the monte carlo simulation ? The reason I ask is that the text ouput given by the monte carlo cluster analysis gives different coorinates in the cluster table to that output given when one clicks on the "find clusters and go to max" button ...
Thanks.
Mahinda
On 04/28/2012 09:45 PM, Mahinda Yogarajah wrote:
Dear Experts,
I had 2 questions with regard to vertex wise analysis using qdec.
- How should one interpret the dependent variables area/pial area and
volume when comparing 2 groups with qdec and a vertex wise (as opposed to ROI based) analysis - or are they uninterpretable in this context ? I read briefly about areal as an alternative measure - could someone explain the difference, and how thickness, volume and area related to one another ?
volume = area * thickness. The "area" of a vertex is the average of the triangles that surround it. For a group analysis, you can think of it as drawing a small circle around a vertex in fsaverage space and asking how big that circle is in each individual subject.
- When running monte-carlo cluster wise multiple correction in qdec -
if one finds a significant cluster is it ok to view them within qdec by highlighting the "mc-z ... " file in the analysis result box, and then changing the minimum threshold to whatever value one selected for the monte carlo simulation ?
No, the mc-z output has values equal to the cluster-wise p-value, so it is not appropriate to apply the vertex-wise threshold.
The reason I ask is that the text ouput given by the monte carlo cluster analysis gives different coorinates in the cluster table to that output given when one clicks on the "find clusters and go to max" button ...
Are you using the "abs" when you do the clustering? Is is possible that it found a negative peak in the table but qdec goes to the positive peak?
Thanks.
Mahinda
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Hi Doug,
Thanks for the swift answer to the earlier questions - can I clairfy some things:
1) Am I safe to use (and attempt to interpret) area/volume in a vertex wise analysis (as opposed to an ROI based analysis) using the basic concept you describe below - I have read on the forums about problems with regard to this - some have mentioned using "areal" instead area and others have mentioned the lack of "jacobian modulation' for volume calculations - what is the current thinking on these issues ...
2) With regard to visualisation of monte carlo simulation output - I made a mistake in my last post - when I visualise in qdec I highlight the mc-z file, and set the threshold to 1.3 which I believe is the default setting for the cluster wise p value - am I right in thinking that the threshold selected in the monte carlo box in qdec is the vertex-wise threshold ? I checked again when I click on find clusters and go to max and the table generated is slightly different to that output from the monte carlo run button as shown below - could you explain what the difference is ...
Thanks.
Mahinda
Output from Monte Carlo run:
ClusterNo Max VtxMax Size(mm^2) TalX TalY TalZ CWP CWPLow CWPHi NVtxs Annot
1 -4.615 105901 851.87 -20.7 -20.3 57.4 0.00550 0.00460 0.00650 2195 precentral
2 -3.482 64220 584.30 -40.4 45.5 -10.6 0.04790 0.04520 0.05060 942 parsorbitalis
Output from Find clusters and go to max
============================================================
Generating cluster stats using min threshold of 1.3...
Found 2 clusters
Contrast: 'lh-Diff-Patient-Control-Cor-thickness-age', 15fwhm, DOF: 34
ClusterNo Max VtxMax Size(mm2) TalX TalY TalZ NVtxs Annotation
--------- --- ------ --------- ---- ---- ---- ----- ----------
1 -2.2596 47 851.87 -22.0 -29.5 52.2 2195 precentral
2 -1.3197 24 584.30 -38.1 50.0 -3.4 942 rostralmiddlefrontal
============================================================
On 04/28/2012 09:45 PM, Mahinda Yogarajah wrote:
Dear Experts,
I had 2 questions with regard to vertex wise analysis using qdec.
- How should one interpret the dependent variables area/pial area and
volume when comparing 2 groups with qdec and a vertex wise (as opposed to ROI based) analysis - or are they uninterpretable in this context ? I read briefly about areal as an alternative measure - could someone explain the difference, and how thickness, volume and area related to one another ?
volume = area * thickness. The "area" of a vertex is the average of the triangles that surround it. For a group analysis, you can think of it as drawing a small circle around a vertex in fsaverage space and asking how big that circle is in each individual subject.
- When running monte-carlo cluster wise multiple correction in qdec -
if one finds a significant cluster is it ok to view them within qdec by highlighting the "mc-z ... " file in the analysis result box, and then changing the minimum threshold to whatever value one selected for the monte carlo simulation ?
No, the mc-z output has values equal to the cluster-wise p-value, so it is not appropriate to apply the vertex-wise threshold.
The reason I ask is that the text ouput given by the monte carlo cluster analysis gives different coorinates in the cluster table to that output given when one clicks on the "find clusters and go to max" button ...
Are you using the "abs" when you do the clustering? Is is possible that it found a negative peak in the table but qdec goes to the positive peak?
Thanks.
Mahinda
Hi Mahinda,
For question 1, if you use the new mris_preproc, as described here: http://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg19821.html then yes, you are safe in computing the areas for each vertex. We have a recent paper that discuss current matters on pointwise area stuff, and that may answer your questions. Please, have a look here: http://www.sciencedirect.com/science/article/pii/S1053811912002996
All the best,
Anderson
On 02/05/12 02:02, Mahinda Yogarajah wrote:
Hi Doug,
Thanks for the swift answer to the earlier questions - can I clairfy some things:
- Am I safe to use (and attempt to interpret) area/volume in a vertex
wise analysis (as opposed to an ROI based analysis) using the basic concept you describe below - I have read on the forums about problems with regard to this - some have mentioned using "areal" instead area and others have mentioned the lack of "jacobian modulation' for volume calculations - what is the current thinking on these issues ...
- With regard to visualisation of monte carlo simulation output - I
made a mistake in my last post - when I visualise in qdec I highlight the mc-z file, and set the threshold to 1.3 which I believe is the default setting for the cluster wise p value - am I right in thinking that the threshold selected in the monte carlo box in qdec is the vertex-wise threshold ? I checked again when I click on find clusters and go to max and the table generated is slightly different to that output from the monte carlo run button as shown below - could you explain what the difference is ...
Thanks.
Mahinda
Output from Monte Carlo run:
ClusterNoMaxVtxMaxSize(mm^2)TalXTalYTalZCWPCWPLowCWPHiNVtxsAnnot
1-4.615105901851.87-20.7-20.357.40.005500.004600.006502195precentral
2-3.48264220584.30-40.445.5-10.60.047900.045200.05060942parsorbitalis
Output from Find clusters and go to max
============================================================
Generating cluster stats using min threshold of 1.3...
Found 2 clusters
Contrast: 'lh-Diff-Patient-Control-Cor-thickness-age', 15fwhm, DOF: 34
ClusterNoMaxVtxMaxSize(mm2)TalXTalYTalZ NVtxs Annotation
1-2.259647851.87-22.0-29.552.2 2195precentral
2-1.319724584.30-38.150.0-3.4942rostralmiddlefrontal
============================================================
On 04/28/2012 09:45 PM, Mahinda Yogarajah wrote:
Dear Experts,
I had 2 questions with regard to vertex wise analysis using qdec.
- How should one interpret the dependent variables area/pial area and
volume when comparing 2 groups with qdec and a vertex wise (as opposed to ROI based) analysis - or are they uninterpretable in this context ? I read briefly about areal as an alternative measure - could someone explain the difference, and how thickness, volume and area related to one another ?
volume = area * thickness. The "area" of a vertex is the average of the triangles that surround it. For a group analysis, you can think of it as drawing a small circle around a vertex in fsaverage space and asking how big that circle is in each individual subject.
- When running monte-carlo cluster wise multiple correction in qdec -
if one finds a significant cluster is it ok to view them within qdec by highlighting the "mc-z ... " file in the analysis result box, and then changing the minimum threshold to whatever value one selected for the monte carlo simulation ?
No, the mc-z output has values equal to the cluster-wise p-value, so it is not appropriate to apply the vertex-wise threshold.
The reason I ask is that the text ouput given by the monte carlo cluster analysis gives different coorinates in the cluster table to that output given when one clicks on the "find clusters and go to max" button ...
Are you using the "abs" when you do the clustering? Is is possible that it found a negative peak in the table but qdec goes to the positive peak?
Thanks.
Mahinda
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.
On 05/01/2012 09:02 PM, Mahinda Yogarajah wrote:
Hi Doug,
Thanks for the swift answer to the earlier questions - can I clairfy some things:
- Am I safe to use (and attempt to interpret) area/volume in a vertex
wise analysis (as opposed to an ROI based analysis) using the basic concept you describe below - I have read on the forums about problems with regard to this - some have mentioned using "areal" instead area and others have mentioned the lack of "jacobian modulation' for volume calculations - what is the current thinking on these issues ...
It is safe if you have the new version of mris_preproc.
- With regard to visualisation of monte carlo simulation output - I
made a mistake in my last post - when I visualise in qdec I highlight the mc-z file, and set the threshold to 1.3 which I believe is the default setting for the cluster wise p value - am I right in thinking that the threshold selected in the monte carlo box in qdec is the vertex-wise threshold ?
Yes
I checked again when I click on find clusters and go to max and the table generated is slightly different to that output from the monte carlo run button as shown below - could you explain what the difference is ...
Are you using "abs"? doug
Thanks.
Mahinda
Output from Monte Carlo run:
ClusterNoMaxVtxMaxSize(mm^2)TalXTalYTalZCWPCWPLowCWPHiNVtxsAnnot
1-4.615105901851.87-20.7-20.357.40.005500.004600.006502195precentral
2-3.48264220584.30-40.445.5-10.60.047900.045200.05060942parsorbitalis
Output from Find clusters and go to max
============================================================
Generating cluster stats using min threshold of 1.3...
Found 2 clusters
Contrast: 'lh-Diff-Patient-Control-Cor-thickness-age', 15fwhm, DOF: 34
ClusterNoMaxVtxMaxSize(mm2)TalXTalYTalZ NVtxs Annotation
1-2.259647851.87-22.0-29.552.2 2195precentral
2-1.319724584.30-38.150.0-3.4942rostralmiddlefrontal
============================================================
On 04/28/2012 09:45 PM, Mahinda Yogarajah wrote:
Dear Experts,
I had 2 questions with regard to vertex wise analysis using qdec.
- How should one interpret the dependent variables area/pial area and
volume when comparing 2 groups with qdec and a vertex wise (as opposed to ROI based) analysis - or are they uninterpretable in this context ? I read briefly about areal as an alternative measure - could someone explain the difference, and how thickness, volume and area related to one another ?
volume = area * thickness. The "area" of a vertex is the average of the triangles that surround it. For a group analysis, you can think of it as drawing a small circle around a vertex in fsaverage space and asking how big that circle is in each individual subject.
- When running monte-carlo cluster wise multiple correction in qdec -
if one finds a significant cluster is it ok to view them within qdec by highlighting the "mc-z ... " file in the analysis result box, and then changing the minimum threshold to whatever value one selected for the monte carlo simulation ?
No, the mc-z output has values equal to the cluster-wise p-value, so it is not appropriate to apply the vertex-wise threshold.
The reason I ask is that the text ouput given by the monte carlo cluster analysis gives different coorinates in the cluster table to that output given when one clicks on the "find clusters and go to max" button ...
Are you using the "abs" when you do the clustering? Is is possible that it found a negative peak in the table but qdec goes to the positive peak?
Thanks.
Mahinda
Hi Doug,
I am using FS build centos 4 64 bit 5.1.1 (downloaded Nov 2011) - presumeably new mris_preproc will not be included in this ? If not is the version Anderson posted link to earlier in this thread (thank you Anderson!) good to use ?
And yes I was using "ABS" - however when I select a unidirectional option (eg neg) the tables generated by run in monte-carlo window, and find clusters and go to max are still slightly different as posted earlier (ie size of clusters and no of vertices in each cluster remains the same, but max, vt max and coordinates differ).
Thanks again.
Mahinda
On Thu, May 3, 2012 at 5:47 PM, Douglas N Greve greve@nmr.mgh.harvard.eduwrote:
On 05/01/2012 09:02 PM, Mahinda Yogarajah wrote:
Hi Doug,
Thanks for the swift answer to the earlier questions - can I clairfy some things:
- Am I safe to use (and attempt to interpret) area/volume in a vertex
wise analysis (as opposed to an ROI based analysis) using the basic concept you describe below - I have read on the forums about problems with regard to this - some have mentioned using "areal" instead area and others have mentioned the lack of "jacobian modulation' for volume calculations - what is the current thinking on these issues ...
It is safe if you have the new version of mris_preproc.
- With regard to visualisation of monte carlo simulation output - I made
a mistake in my last post - when I visualise in qdec I highlight the mc-z file, and set the threshold to 1.3 which I believe is the default setting for the cluster wise p value - am I right in thinking that the threshold selected in the monte carlo box in qdec is the vertex-wise threshold ?
Yes
I checked again when I click on find clusters and go to max and the table
generated is slightly different to that output from the monte carlo run button as shown below - could you explain what the difference is ...
Are you using "abs"?
doug
Thanks.
Mahinda
Output from Monte Carlo run:
ClusterNoMaxVtxMaxSize(mm^2)**TalXTalYTalZCWPCWPLowCWPHiNVtx**sAnnot
1-4.615105901851.87-20.7-20.**357.40.005500.004600.**006502195precentral
2-3.48264220584.30-40.445.5-**10.60.047900.045200.**05060942parsorbitalis
Output from Find clusters and go to max
==============================**==============================
Generating cluster stats using min threshold of 1.3...
Found 2 clusters
Contrast: 'lh-Diff-Patient-Control-Cor-**thickness-age', 15fwhm, DOF: 34
ClusterNoMaxVtxMaxSize(mm2)**TalXTalYTalZ NVtxs Annotation
------------------------------**--------- ----- ----------
1-2.259647851.87-22.0-29.552.2 2195precentral
2-1.319724584.30-38.150.0-3.**4942rostralmiddlefrontal
==============================**==============================
On 04/28/2012 09:45 PM, Mahinda Yogarajah wrote:
Dear Experts,
I had 2 questions with regard to vertex wise analysis using qdec.
- How should one interpret the dependent variables area/pial area and
volume when comparing 2 groups with qdec and a vertex wise (as opposed to ROI based) analysis - or are they uninterpretable in this context ? I read briefly about areal as an alternative measure - could someone explain the difference, and how thickness, volume and area related to one another ?
volume = area * thickness. The "area" of a vertex is the average of the triangles that surround it. For a group analysis, you can think of it as drawing a small circle around a vertex in fsaverage space and asking how big that circle is in each individual subject.
- When running monte-carlo cluster wise multiple correction in qdec -
if one finds a significant cluster is it ok to view them within qdec by highlighting the "mc-z ... " file in the analysis result box, and then changing the minimum threshold to whatever value one selected for the monte carlo simulation ?
No, the mc-z output has values equal to the cluster-wise p-value, so it is not appropriate to apply the vertex-wise threshold.
The reason I ask is that the text ouput given by the monte carlo cluster analysis gives different coorinates in the cluster table to that output given when one clicks on the "find clusters and go to max" button ...
Are you using the "abs" when you do the clustering? Is is possible that it found a negative peak in the table but qdec goes to the positive peak?
Thanks.
Mahinda
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
Bugs: surfer.nmr.mgh.harvard.edu/**fswiki/BugReportinghttp://surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: www.nmr.mgh.harvard.edu/**facility/filedrop/index.htmlhttp://www.nmr.mgh.harvard.edu/facility/filedrop/index.html
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/**compliancelinehttp://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 Winkler -
Douglas N Greve -
Mahinda Yogarajah