Dear Douglas, thank you very much for you kind reply.In the paper I have 3 groups (2 diseases and 1 controls).I have performed cortical thickness analysis among groups, individuating relevant difference in the posterior regions. Next, for one group, I have used qdec to individuate statistically significant regions where NPS test scores and thickness were correlated. 3 nuisance factors was also added to regressor. All results were corrected for multiple comparisons by using a pre-cached cluster-wise Monte-Carlo Simulation.qdec analysis results have individuated a "composite region" (superior parietal cortex and precuneus) where NPS correlated with thinning. Now, if I understand, I should extract the mean thickness from superior parietal cortex and precuneus and I should perform a linear regression (and partial correlation) with an other statistical program (SPSS for example).Is there a better way to perform a regression the most possible similar to qdec computation?
Stefano
----Messaggio originale---- Da: greve@nmr.mgh.harvard.edu Data: 4-dic-2013 17.35 A: stdp82@virgilio.it, freesurfer@nmr.mgh.harvard.edu Ogg: Re: R: Re: [Freesurfer] Linear regression
On 12/02/2013 03:14 PM, stdp82@virgilio.it wrote:
Tank you very much.
A-Do you mean the partial correlation coeff (PCC) at each voxel averaged over region? Finally a linear regression is done between cortical thickness of a composite region and NPS. However the authors discuss every individual region as related to disease. So a table with partials correlations of individuals ROIs with NPS seems useful to support the discussion.
What did you do in the paper? The reviewer mentions a "composite region". Did you average the thickness over that region and then do the regression with NPS?
B- I'd like to obtain the covariance matrix of the regression made between cortical thickness, group and covariates. How can I do it? I'm not sure what you mean. Would this be a different matrix for each vertex? I cite the referee: "The authors must provide the covariance matrix of the regression made between cortical thickness, group and covariates as supplementary information in order to increase the value of the results."
Sorry, I don't understand what the reviewer is asking for. doug
I don't have Matlab for mri_glmfit_pcc matlab.
Stefano
----Messaggio originale---- Da: greve@nmr.mgh.harvard.edu Data: 2-dic-2013 20.47 A: freesurfer@nmr.mgh.harvard.edu Ogg: Re: [Freesurfer] Linear regression
On 12/02/2013 02:22 PM, stdp82@virgilio.it wrote:
Hi list,
I have some questions, please.
I performed linear regression by qdec to find the regions where the cortical thinning correlates with neuropsychological test (NPS), taking in account 3 nuisance factors.
A- I individuated some posterior regions and now I'd like to build a table with partials correlations of individuals ROIs with NPS. How can I obtain the values for each region?
Do you mean the partial correlation coeff (PCC) at each voxel averaged over region? You can get PCC maps using the mri_glmfit_pcc matlab command. You can then do the averaging over region using mri_segstats.
B- I'd like to obtain the covariance matrix of the regression made between cortical thickness,group and covariates. How can I do it?
I'm not sure what you mean. Would this be a different matrix for each vertex?
C- Is there a specific reference for qdec?
No, it is just a GLM so any GLM reference will do.
Thank you very much,
Stefano
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- 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/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/
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.
If you don't have matlab, then I don't know of one. The design matrix is in the glmdir folder (Xg.dat), so if SPSS or some other software can read that in, then the analysis will be the same. You can also run mri_glmfit on the mean thicknesses from your ROI. This will not produce a PCC, but you can at least check the p-values against a program that does produce a PCC doug
On 12/04/2013 01:06 PM, stdp82@virgilio.it wrote:
Dear Douglas, thank you very much for you kind reply. In the paper I have 3 groups (2 diseases and 1 controls). I have performed cortical thickness analysis among groups, individuating relevant difference in the posterior regions. Next, for one group, I have used qdec to individuate statistically significant regions where NPS test scores and thickness were correlated. 3 nuisance factors was also added to regressor. All results were corrected for multiple comparisons by using a pre-cached cluster-wise Monte-Carlo Simulation. qdec analysis results have individuated a "composite region" (superior parietal cortex and precuneus) where NPS correlated with thinning. Now, if I understand, I should extract the mean thickness from superior parietal cortex and precuneus and I should perform a linear regression (and partial correlation) with an other statistical program (SPSS for example). Is there a better way to perform a regression the most possible similar to qdec computation?
Stefano
----Messaggio originale---- Da: greve@nmr.mgh.harvard.edu Data: 4-dic-2013 17.35 A: stdp82@virgilio.it, freesurfer@nmr.mgh.harvard.edu Ogg: Re: R: Re: [Freesurfer] Linear regression
On 12/02/2013 03:14 PM, stdp82@virgilio.it wrote:
Tank you very much.
A-Do you mean the partial correlation coeff (PCC) at each voxel averaged over region? Finally a linear regression is done between cortical thickness of a composite region and NPS. However the authors discuss every individual region as related to disease. So a table with partials correlations of individuals ROIs with NPS seems useful to support the discussion.
What did you do in the paper? The reviewer mentions a "composite region". Did you average the thickness over that region and then do the regression with NPS?
B- I'd like to obtain the covariance matrix of the regression made between cortical thickness, group and covariates. How can I do it? I'm not sure what you mean. Would this be a different matrix for each vertex? I cite the referee: "The authors must provide the covariance matrix of the regression made between cortical thickness, group and covariates as supplementary information in order to increase the value of the results."
Sorry, I don't understand what the reviewer is asking for. doug
I don't have Matlab for mri_glmfit_pcc matlab.
Stefano
----Messaggio originale---- Da: greve@nmr.mgh.harvard.edu Data: 2-dic-2013 20.47 A: freesurfer@nmr.mgh.harvard.edu Ogg: Re: [Freesurfer] Linear regression
On 12/02/2013 02:22 PM, stdp82@virgilio.it wrote:
Hi list,
I have some questions, please.
I performed linear regression by qdec to find the regions where the cortical thinning correlates with neuropsychological test (NPS), taking in account 3 nuisance factors.
A- I individuated some posterior regions and now I'd like to build a table with partials correlations of individuals ROIs with NPS. How can I obtain the values for each region?
Do you mean the partial correlation coeff (PCC) at each voxel averaged over region? You can get PCC maps using the mri_glmfit_pcc matlab command. You can then do the averaging over region using mri_segstats.
B- I'd like to obtain the covariance matrix of the regression made between cortical thickness,group and covariates. How can I do it?
I'm not sure what you mean. Would this be a different matrix for each vertex?
C- Is there a specific reference for qdec?
No, it is just a GLM so any GLM reference will do.
Thank you very much,
Stefano
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- 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/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/
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.
-- 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/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/
If you don't have MATLAB you could try using Octave instead.
On Wed, 4 Dec 2013, Douglas N Greve wrote:
If you don't have matlab, then I don't know of one. The design matrix is in the glmdir folder (Xg.dat), so if SPSS or some other software can read that in, then the analysis will be the same. You can also run mri_glmfit on the mean thicknesses from your ROI. This will not produce a PCC, but you can at least check the p-values against a program that does produce a PCC doug
On 12/04/2013 01:06 PM, stdp82@virgilio.it wrote:
Dear Douglas, thank you very much for you kind reply. In the paper I have 3 groups (2 diseases and 1 controls). I have performed cortical thickness analysis among groups, individuating relevant difference in the posterior regions. Next, for one group, I have used qdec to individuate statistically significant regions where NPS test scores and thickness were correlated. 3 nuisance factors was also added to regressor. All results were corrected for multiple comparisons by using a pre-cached cluster-wise Monte-Carlo Simulation. qdec analysis results have individuated a "composite region" (superior parietal cortex and precuneus) where NPS correlated with thinning. Now, if I understand, I should extract the mean thickness from superior parietal cortex and precuneus and I should perform a linear regression (and partial correlation) with an other statistical program (SPSS for example). Is there a better way to perform a regression the most possible similar to qdec computation?
Stefano
----Messaggio originale---- Da: greve@nmr.mgh.harvard.edu Data: 4-dic-2013 17.35 A: stdp82@virgilio.it, freesurfer@nmr.mgh.harvard.edu Ogg: Re: R: Re: [Freesurfer] Linear regression
On 12/02/2013 03:14 PM, stdp82@virgilio.it wrote:
Tank you very much.
A-Do you mean the partial correlation coeff (PCC) at each voxel averaged over region? Finally a linear regression is done between cortical thickness of a composite region and NPS. However the authors discuss every individual region as related to disease. So a table with partials correlations of individuals ROIs with NPS seems useful to support the discussion.
What did you do in the paper? The reviewer mentions a "composite region". Did you average the thickness over that region and then do the regression with NPS?
B- I'd like to obtain the covariance matrix of the regression made between cortical thickness, group and covariates. How can I do it? I'm not sure what you mean. Would this be a different matrix for each vertex? I cite the referee: "The authors must provide the covariance matrix of the regression made between cortical thickness, group and covariates as supplementary information in order to increase the value of the results."
Sorry, I don't understand what the reviewer is asking for. doug
I don't have Matlab for mri_glmfit_pcc matlab.
Stefano
----Messaggio originale---- Da: greve@nmr.mgh.harvard.edu Data: 2-dic-2013 20.47 A: freesurfer@nmr.mgh.harvard.edu Ogg: Re: [Freesurfer] Linear regression
On 12/02/2013 02:22 PM, stdp82@virgilio.it wrote:
Hi list,
I have some questions, please.
I performed linear regression by qdec to find the regions where the cortical thinning correlates with neuropsychological test (NPS), taking in account 3 nuisance factors.
A- I individuated some posterior regions and now I'd like to build a table with partials correlations of individuals ROIs with NPS. How can I obtain the values for each region?
Do you mean the partial correlation coeff (PCC) at each voxel averaged over region? You can get PCC maps using the mri_glmfit_pcc matlab command. You can then do the averaging over region using mri_segstats.
B- I'd like to obtain the covariance matrix of the regression made between cortical thickness,group and covariates. How can I do it?
I'm not sure what you mean. Would this be a different matrix for each vertex?
C- Is there a specific reference for qdec?
No, it is just a GLM so any GLM reference will do.
Thank you very much,
Stefano
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- 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/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/
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.
-- 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/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/
Dear Stefano, I also want to make a correlation analysis between thickness and behavior data. Would you please send me your precious scripts for group analysis( unpaired t-test and correlation analysis) after recon-all? I want to study it if I get your permission. Thanks. All the best.
2013-12-05
Rujing Zha
发件人:stdp82@virgilio.it 发送时间:2013-12-05 02:06 主题:[Freesurfer] R: Re: R: Re: Linear regression 收件人:"greve"greve@nmr.mgh.harvard.edu,"freesurfer"freesurfer@nmr.mgh.harvard.edu 抄送:
Dear Douglas, thank you very much for you kind reply. In the paper I have 3 groups (2 diseases and 1 controls). I have performed cortical thickness analysis among groups, individuating relevant difference in the posterior regions. Next, for one group, I have used qdec to individuate statistically significant regions where NPS test scores and thickness were correlated. 3 nuisance factors was also added to regressor. All results were corrected for multiple comparisons by using a pre-cached cluster-wise Monte-Carlo Simulation. qdec analysis results have individuated a "composite region" (superior parietal cortex and precuneus) where NPS correlated with thinning. Now, if I understand, I should extract the mean thickness from superior parietal cortex and precuneus and I should perform a linear regression (and partial correlation) with an other statistical program (SPSS for example). Is there a better way to perform a regression the most possible similar to qdec computation?
Stefano
----Messaggio originale---- Da: greve@nmr.mgh.harvard.edu Data: 4-dic-2013 17.35 A: stdp82@virgilio.it, freesurfer@nmr.mgh.harvard.edu Ogg: Re: R: Re: [Freesurfer] Linear regression
On 12/02/2013 03:14 PM, stdp82@virgilio.it wrote:
Tank you very much.
A-Do you mean the partial correlation coeff (PCC) at each voxel averaged over region? Finally a linear regression is done between cortical thickness of a composite region and NPS. However the authors discuss every individual region as related to disease. So a table with partials correlations of individuals ROIs with NPS seems useful to support the discussion.
What did you do in the paper? The reviewer mentions a "composite region". Did you average the thickness over that region and then do the regression with NPS?
B- I'd like to obtain the covariance matrix of the regression made between cortical thickness, group and covariates. How can I do it? I'm not sure what you mean. Would this be a different matrix for each vertex? I cite the referee: "The authors must provide the covariance matrix of the regression made between cortical thickness, group and covariates as supplementary information in order to increase the value of the results."
Sorry, I don't understand what the reviewer is asking for. doug
I don't have Matlab for mri_glmfit_pcc matlab.
Stefano
----Messaggio originale---- Da: greve@nmr.mgh.harvard.edu Data: 2-dic-2013 20.47 A: freesurfer@nmr.mgh.harvard.edu Ogg: Re: [Freesurfer] Linear regression
On 12/02/2013 02:22 PM, stdp82@virgilio.it wrote:
Hi list,
I have some questions, please.
I performed linear regression by qdec to find the regions where the cortical thinning correlates with neuropsychological test (NPS), taking in account 3 nuisance factors.
A- I individuated some posterior regions and now I'd like to build a table with partials correlations of individuals ROIs with NPS. How can I obtain the values for each region?
Do you mean the partial correlation coeff (PCC) at each voxel averaged over region? You can get PCC maps using the mri_glmfit_pcc matlab command. You can then do the averaging over region using mri_segstats.
B- I'd like to obtain the covariance matrix of the regression made between cortical thickness,group and covariates. How can I do it?
I'm not sure what you mean. Would this be a different matrix for each vertex?
C- Is there a specific reference for qdec?
No, it is just a GLM so any GLM reference will do.
Thank you very much,
Stefano
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- 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/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/
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