Dear Freesurfer experts,
I have three groups g0, g1, and g2.
I am trying to use LME model with random effects y-int and time from the base line.
Based on the LME tutorial and the questions and answers from Freesurfer support list:
Y=B1+B2*t+B3*g1+B4*g1*t+B5*g2+B6*g2*t+ ...
The slope of g0 : 0 1 0 0 0 0 The slope of g1: 0 1 0 1 0 0 The slope of g2: 0 1 0 0 0 1
The slope difference between g0 and g1: 0 0 0 1 0 0 The slope difference between g0 and g2: 0 0 0 0 0 1 The slope difference between g1 and g2: 0 0 0 -1 0 1
1) Is this contrast matrix correct to observe pairwise slope difference?
In the tutorial the contrast matrix C = [zeros(3,3) [1 0 0 0 0; -1 0 1 0 0; 0 0 -1 0 1] zeros(3,6)]
2) Is it necessary to combine contrast matrix of the pair difference the same as the tutorial?
If so, my contrast matrix can be C=[0 0 0 1 0 0; 0 0 0 -1 0 1]
3) How can we interpret the result for this contrast matrix?
4) How can I obtain B1, B2, ... ?
[sz1,sz2] = lme_plannedSampleSize(Zi,ZiCol,Dhat,phisqhat,effsz,... dr,pw,alpha,gr_pr)
5) For sample size calculation how can I get Dhat ( Estimated random effects covariance matrix) and phisqhat (Estimated intra-subject variability)?
With regards
Getaneh
Hi Getaneh,
please find my responses below.
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: freesurfer@nmr.mgh.harvard.edu freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] LME Date: Mon, 8 Jan 2018 22:54:49 +0100
Dear Freesurfer experts,
I have three groups g0, g1, and g2.
I am trying to use LME model with random effects y-int and time from the base line.
Based on the LME tutorial and the questions and answers from Freesurfer support list:
Y=B1+B2*t+B3*g1+B4*g1*t+B5*g2+B6*g2*t+ …
The slope of g0 : 0 1 0 0 0 0 The slope of g1: 0 1 0 1 0 0 The slope of g2: 0 1 0 0 0 1
The slope difference between g0 and g1: 0 0 0 1 0 0 The slope difference between g0 and g2: 0 0 0 0 0 1 The slope difference between g1 and g2: 0 0 0 -1 0 1
1) Is this contrast matrix correct to observe pairwise slope difference?
In the tutorial the contrast matrix C = [zeros(3,3) [1 0 0 0 0; -1 0 1 0 0; 0 0 -1 0 1] zeros(3,6)]
This is a correct matrix, but only for the particular design given in the tutorial, i.e. 4 groups and several covariates. Also, it is not strictly speaking a pairwise test, but will indicate if there is *any* difference between the slopes of the groups (i.e. it will give only one F- and p-value). If you specifically want to test differences between particular pairs, use contrast matrices with just a single row.
2) Is it necessary to combine contrast matrix of the pair difference the same as the tutorial?
If so, my contrast matrix can be C=[0 0 0 1 0 0; 0 0 0 -1 0 1]
It makes sense to look at the overall effect prior to looking at pairwise differences. If you don't have any additional covariates, that matrix is OK for looking the overall effect. Otherwise, add as many columns of zeros as you have covariates, similar to the tutorial.
3) How can we interpret the result for this contrast matrix?
This is an F-test to detect if there is any difference between the slopes of the groups.
4) How can I obtain B1, B2, … ?
Consider the following command from the tutorial:
total_hipp_vol_stats = lme_fit_FS(X,[1 2],Y(:,1)+Y(:,2),ni);
Then B1 and B2 will be the first two entries of total_hipp_vol_stats.Bhat
[sz1,sz2] = lme_plannedSampleSize(Zi,ZiCol,Dhat,phisqhat,effsz,... dr,pw,alpha,gr _pr)
5) For sample size calculation how can I get Dhat ( Estimated random effects covariance matrix) and phisqhat (Estimated intra-subject variability)?
If you are planning a new design, these estimates should be based on prior knowledge and can be derived e.g. from a pilot study or from the literature.
Best regards,
Kersten
With regards
Getaneh
Hi Kersten.
Thank you very much .
I am not planning a new design for sample size calculation.
Can you please help me how to find those values?
Thank you
Getaneh
________________________________________ From: freesurfer-bounces@nmr.mgh.harvard.edu freesurfer-bounces@nmr.mgh.harvard.edu on behalf of Diers, Kersten /DZNE Kersten.Diers@dzne.de Sent: Tuesday, January 9, 2018 5:56 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME
Hi Getaneh,
please find my responses below.
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: freesurfer@nmr.mgh.harvard.edu freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] LME Date: Mon, 8 Jan 2018 22:54:49 +0100
Dear Freesurfer experts,
I have three groups g0, g1, and g2.
I am trying to use LME model with random effects y-int and time from the base line.
Based on the LME tutorial and the questions and answers from Freesurfer support list:
Y=B1+B2*t+B3*g1+B4*g1*t+B5*g2+B6*g2*t+ …
The slope of g0 : 0 1 0 0 0 0 The slope of g1: 0 1 0 1 0 0 The slope of g2: 0 1 0 0 0 1
The slope difference between g0 and g1: 0 0 0 1 0 0 The slope difference between g0 and g2: 0 0 0 0 0 1 The slope difference between g1 and g2: 0 0 0 -1 0 1
1) Is this contrast matrix correct to observe pairwise slope difference?
In the tutorial the contrast matrix C = [zeros(3,3) [1 0 0 0 0; -1 0 1 0 0; 0 0 -1 0 1] zeros(3,6)]
This is a correct matrix, but only for the particular design given in the tutorial, i.e. 4 groups and several covariates. Also, it is not strictly speaking a pairwise test, but will indicate if there is *any* difference between the slopes of the groups (i.e. it will give only one F- and p-value). If you specifically want to test differences between particular pairs, use contrast matrices with just a single row.
2) Is it necessary to combine contrast matrix of the pair difference the same as the tutorial?
If so, my contrast matrix can be C=[0 0 0 1 0 0; 0 0 0 -1 0 1]
It makes sense to look at the overall effect prior to looking at pairwise differences. If you don't have any additional covariates, that matrix is OK for looking the overall effect. Otherwise, add as many columns of zeros as you have covariates, similar to the tutorial.
3) How can we interpret the result for this contrast matrix?
This is an F-test to detect if there is any difference between the slopes of the groups.
4) How can I obtain B1, B2, … ?
Consider the following command from the tutorial:
total_hipp_vol_stats = lme_fit_FS(X,[1 2],Y(:,1)+Y(:,2),ni);
Then B1 and B2 will be the first two entries of total_hipp_vol_stats.Bhat
[sz1,sz2] = lme_plannedSampleSize(Zi,ZiCol,Dhat,phisqhat,effsz,... dr,pw,alpha,gr _pr)
5) For sample size calculation how can I get Dhat ( Estimated random effects covariance matrix) and phisqhat (Estimated intra-subject variability)?
If you are planning a new design, these estimates should be based on prior knowledge and can be derived e.g. from a pilot study or from the literature.
Best regards,
Kersten
With regards
Getaneh
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harvard.ed...
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 https://urldefense.proofpoint.com/v2/url?u=http-3A__www.partners.org_complia... . 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.
Hi Getaneh,
the estimates for D and phisq for a given analysis are contained within the data structure that is returned by the lme_fit_FS procedure.
I.e. for the tutorial data, the command was:
total_hipp_vol_stats = lme_fit_FS(X,[1 2],Y(:,1)+Y(:,2),ni);
Dhat and phisqhat are fields of the 'total_hipp_vol_stats' structure, and can be accessed by typing 'total_hipp_vol_stats.Dhat' and 'total_hipp_vol_stats.phisqhat', respectively.
Best regards,
Kersten
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: Freesurfer support list freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME Date: Tue, 9 Jan 2018 23:02:08 +0100
Hi Kersten.
Thank you very much .
I am not planning a new design for sample size calculation.
Can you please help me how to find those values?
Thank you
Getaneh
________________________________________ From: freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mg h.harvard.edu> on behalf of Diers, Kersten /DZNE <Kersten.Diers@dzne.de
Sent: Tuesday, January 9, 2018 5:56 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME
Hi Getaneh,
please find my responses below.
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: freesurfer@nmr.mgh.harvard.edu freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] LME Date: Mon, 8 Jan 2018 22:54:49 +0100
Dear Freesurfer experts,
I have three groups g0, g1, and g2.
I am trying to use LME model with random effects y-int and time from the base line.
Based on the LME tutorial and the questions and answers from Freesurfer support list:
Y=B1+B2*t+B3*g1+B4*g1*t+B5*g2+B6*g2*t+ …
The slope of g0 : 0 1 0 0 0 0 The slope of g1: 0 1 0 1 0 0 The slope of g2: 0 1 0 0 0 1
The slope difference between g0 and g1: 0 0 0 1 0 0 The slope difference between g0 and g2: 0 0 0 0 0 1 The slope difference between g1 and g2: 0 0 0 -1 0 1
1) Is this contrast matrix correct to observe pairwise slope difference?
In the tutorial the contrast matrix C = [zeros(3,3) [1 0 0 0 0; -1 0 1 0 0; 0 0 -1 0 1] zeros(3,6)]
This is a correct matrix, but only for the particular design given in the tutorial, i.e. 4 groups and several covariates. Also, it is not strictly speaking a pairwise test, but will indicate if there is *any* difference between the slopes of the groups (i.e. it will give only one F- and p-value). If you specifically want to test differences between particular pairs, use contrast matrices with just a single row.
2) Is it necessary to combine contrast matrix of the pair difference the same as the tutorial?
If so, my contrast matrix can be C=[0 0 0 1 0 0; 0 0 0 -1 0 1]
It makes sense to look at the overall effect prior to looking at pairwise differences. If you don't have any additional covariates, that matrix is OK for looking the overall effect. Otherwise, add as many columns of zeros as you have covariates, similar to the tutorial.
3) How can we interpret the result for this contrast matrix?
This is an F-test to detect if there is any difference between the slopes of the groups.
4) How can I obtain B1, B2, … ?
Consider the following command from the tutorial:
total_hipp_vol_stats = lme_fit_FS(X,[1 2],Y(:,1)+Y(:,2),ni);
Then B1 and B2 will be the first two entries of total_hipp_vol_stats.Bhat
[sz1,sz2] = lme_plannedSampleSize(Zi,ZiCol,Dhat,phisqhat,effsz,... dr,pw,alpha,gr _pr)
5) For sample size calculation how can I get Dhat ( Estimated random effects covariance matrix) and phisqhat (Estimated intra-subject variability)?
If you are planning a new design, these estimates should be based on prior knowledge and can be derived e.g. from a pilot study or from the literature.
Best regards,
Kersten
With regards
Getaneh
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=w3Go ND77ORaREflQR6B1sxt8pIPGPzd8Oh_hzfJ_rhk&s=7GFfEuDrrPYWE- NE_mVYzsQUEvfZizwe853OSP2n_Zo&e=
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 https://urldefense.proofpoint.com/v2/url?u=http-3A__www.partners.org_co mplianceline&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1VapJQ8JVoF8oWH1AgfQ&r=f 2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=w3GoND77ORaREflQR6B1sxt8pI PGPzd8Oh_hzfJ_rhk&s=_JumR_-uS8QiP3xZfqqQ73pib0Pi-_MI1tUhsm2z5bA&e= . 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 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.
Dear Kersten,
Thank you so much.
When I run the procedure with left and right cortical thickness data I get different phisqhat values Phisqhat:0.2530 for the left and Phisqhat:0.0903 for the right
The Dhat matrix is also different.
Is that Ok?
If so , shall I consider two sample sizes for the left and right?
With regards
Getaneh
-----Original Message----- From: freesurfer-bounces@nmr.mgh.harvard.edu [mailto:freesurfer-bounces@nmr.mgh.harvard.edu] On Behalf Of Diers, Kersten /DZNE Sent: Wednesday, January 10, 2018 7:10 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME
Hi Getaneh,
the estimates for D and phisq for a given analysis are contained within the data structure that is returned by the lme_fit_FS procedure.
I.e. for the tutorial data, the command was:
total_hipp_vol_stats = lme_fit_FS(X,[1 2],Y(:,1)+Y(:,2),ni);
Dhat and phisqhat are fields of the 'total_hipp_vol_stats' structure, and can be accessed by typing 'total_hipp_vol_stats.Dhat' and 'total_hipp_vol_stats.phisqhat', respectively.
Best regards,
Kersten
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: Freesurfer support list freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME Date: Tue, 9 Jan 2018 23:02:08 +0100
Hi Kersten.
Thank you very much .
I am not planning a new design for sample size calculation.
Can you please help me how to find those values?
Thank you
Getaneh
________________________________________ From: freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mg h.harvard.edu> on behalf of Diers, Kersten /DZNE <Kersten.Diers@dzne.de
Sent: Tuesday, January 9, 2018 5:56 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME
Hi Getaneh,
please find my responses below.
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: freesurfer@nmr.mgh.harvard.edu freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] LME Date: Mon, 8 Jan 2018 22:54:49 +0100
Dear Freesurfer experts,
I have three groups g0, g1, and g2.
I am trying to use LME model with random effects y-int and time from the base line.
Based on the LME tutorial and the questions and answers from Freesurfer support list:
Y=B1+B2*t+B3*g1+B4*g1*t+B5*g2+B6*g2*t+ …
The slope of g0 : 0 1 0 0 0 0 The slope of g1: 0 1 0 1 0 0 The slope of g2: 0 1 0 0 0 1
The slope difference between g0 and g1: 0 0 0 1 0 0 The slope difference between g0 and g2: 0 0 0 0 0 1 The slope difference between g1 and g2: 0 0 0 -1 0 1
1) Is this contrast matrix correct to observe pairwise slope difference?
In the tutorial the contrast matrix C = [zeros(3,3) [1 0 0 0 0; -1 0 1 0 0; 0 0 -1 0 1] zeros(3,6)]
This is a correct matrix, but only for the particular design given in the tutorial, i.e. 4 groups and several covariates. Also, it is not strictly speaking a pairwise test, but will indicate if there is *any* difference between the slopes of the groups (i.e. it will give only one F- and p-value). If you specifically want to test differences between particular pairs, use contrast matrices with just a single row.
2) Is it necessary to combine contrast matrix of the pair difference the same as the tutorial?
If so, my contrast matrix can be C=[0 0 0 1 0 0; 0 0 0 -1 0 1]
It makes sense to look at the overall effect prior to looking at pairwise differences. If you don't have any additional covariates, that matrix is OK for looking the overall effect. Otherwise, add as many columns of zeros as you have covariates, similar to the tutorial.
3) How can we interpret the result for this contrast matrix?
This is an F-test to detect if there is any difference between the slopes of the groups.
4) How can I obtain B1, B2, … ?
Consider the following command from the tutorial:
total_hipp_vol_stats = lme_fit_FS(X,[1 2],Y(:,1)+Y(:,2),ni);
Then B1 and B2 will be the first two entries of total_hipp_vol_stats.Bhat
[sz1,sz2] = lme_plannedSampleSize(Zi,ZiCol,Dhat,phisqhat,effsz,... dr,pw,alpha,gr _pr)
5) For sample size calculation how can I get Dhat ( Estimated random effects covariance matrix) and phisqhat (Estimated intra-subject variability)?
If you are planning a new design, these estimates should be based on prior knowledge and can be derived e.g. from a pilot study or from the literature.
Best regards,
Kersten
With regards
Getaneh
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=w3Go ND77ORaREflQR6B1sxt8pIPGPzd8Oh_hzfJ_rhk&s=7GFfEuDrrPYWE- NE_mVYzsQUEvfZizwe853OSP2n_Zo&e=
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 https://urldefense.proofpoint.com/v2/url?u=http-3A__www.partners.org_co mplianceline&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1VapJQ8JVoF8oWH1AgfQ&r=f 2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=w3GoND77ORaREflQR6B1sxt8pI PGPzd8Oh_hzfJ_rhk&s=_JumR_-uS8QiP3xZfqqQ73pib0Pi-_MI1tUhsm2z5bA&e= . 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 mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harvard.ed...
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 https://urldefense.proofpoint.com/v2/url?u=http-3A__www.partners.org_complia... . 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 mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harvard.ed...
Hi,
I am not entirely sure that I understand the goal of this particular analysis, so I don't have any suggestions at the moment.
Maybe you could give some more detail, if this is still an issue.
Best regards,
Kersten
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: Freesurfer support list freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME Date: Wed, 10 Jan 2018 17:58:58 +0100
Dear Kersten,
Thank you so much.
When I run the procedure with left and right cortical thickness data I get different phisqhat values Phisqhat:0.2530 for the left and Phisqhat:0.0903 for the right
The Dhat matrix is also different.
Is that Ok?
If so , shall I consider two sample sizes for the left and right?
With regards
Getaneh
-----Original Message----- From: freesurfer-bounces@nmr.mgh.harvard.edu [mailto:freesurfer-bounces @nmr.mgh.harvard.edu] On Behalf Of Diers, Kersten /DZNE Sent: Wednesday, January 10, 2018 7:10 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME
Hi Getaneh,
the estimates for D and phisq for a given analysis are contained within the data structure that is returned by the lme_fit_FS procedure.
I.e. for the tutorial data, the command was:
total_hipp_vol_stats = lme_fit_FS(X,[1 2],Y(:,1)+Y(:,2),ni);
Dhat and phisqhat are fields of the 'total_hipp_vol_stats' structure, and can be accessed by typing 'total_hipp_vol_stats.Dhat' and 'total_hipp_vol_stats.phisqhat', respectively.
Best regards,
Kersten
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: Freesurfer support list freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME Date: Tue, 9 Jan 2018 23:02:08 +0100
Hi Kersten.
Thank you very much .
I am not planning a new design for sample size calculation.
Can you please help me how to find those values?
Thank you
Getaneh
________________________________________ From: freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mg h.harvard.edu> on behalf of Diers, Kersten /DZNE <Kersten.Diers@dzne.d e
Sent: Tuesday, January 9, 2018 5:56 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME
Hi Getaneh,
please find my responses below.
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: freesurfer@nmr.mgh.harvard.edu freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] LME Date: Mon, 8 Jan 2018 22:54:49 +0100
Dear Freesurfer experts,
I have three groups g0, g1, and g2.
I am trying to use LME model with random effects y-int and time from the base line.
Based on the LME tutorial and the questions and answers from Freesurfer support list:
Y=B1+B2*t+B3*g1+B4*g1*t+B5*g2+B6*g2*t+ …
The slope of g0 : 0 1 0 0 0 0 The slope of g1: 0 1 0 1 0 0 The slope of g2: 0 1 0 0 0 1
The slope difference between g0 and g1: 0 0 0 1 0 0 The slope difference between g0 and g2: 0 0 0 0 0 1 The slope difference between g1 and g2: 0 0 0 -1 0 1
1) Is this contrast matrix correct to observe pairwise slope difference?
In the tutorial the contrast matrix C = [zeros(3,3) [1 0 0 0 0; -1 0 1 0 0; 0 0 -1 0 1] zeros(3,6)]
This is a correct matrix, but only for the particular design given in the tutorial, i.e. 4 groups and several covariates. Also, it is not strictly speaking a pairwise test, but will indicate if there is *any* difference between the slopes of the groups (i.e. it will give only one F- and p-value). If you specifically want to test differences between particular pairs, use contrast matrices with just a single row.
2) Is it necessary to combine contrast matrix of the pair difference the same as the tutorial?
If so, my contrast matrix can be C=[0 0 0 1 0 0; 0 0 0 -1 0 1]
It makes sense to look at the overall effect prior to looking at pairwise differences. If you don't have any additional covariates, that matrix is OK for looking the overall effect. Otherwise, add as many columns of zeros as you have covariates, similar to the tutorial.
3) How can we interpret the result for this contrast matrix?
This is an F-test to detect if there is any difference between the slopes of the groups.
4) How can I obtain B1, B2, … ?
Consider the following command from the tutorial:
total_hipp_vol_stats = lme_fit_FS(X,[1 2],Y(:,1)+Y(:,2),ni);
Then B1 and B2 will be the first two entries of total_hipp_vol_stats.Bhat
[sz1,sz2] = lme_plannedSampleSize(Zi,ZiCol,Dhat,phisqhat,effsz,... dr,pw,alpha,gr _pr)
5) For sample size calculation how can I get Dhat ( Estimated random effects covariance matrix) and phisqhat (Estimated intra-subject variability)?
If you are planning a new design, these estimates should be based on prior knowledge and can be derived e.g. from a pilot study or from the literature.
Best regards,
Kersten
With regards
Getaneh
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=w3Go ND77ORaREflQR6B1sxt8pIPGPzd8Oh_hzfJ_rhk&s=7GFfEuDrrPYWE- NE_mVYzsQUEvfZizwe853OSP2n_Zo&e=
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 https://urldefense.proofpoint.com/v2/ur l?u=http-3A__www.partners.org_co mplianceline&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1VapJQ8JVoF8oWH1AgfQ&r=f 2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=w3GoND77ORaREflQR6B1sxt8pI PGPzd8Oh_hzfJ_rhk&s=_JumR_-uS8QiP3xZfqqQ73pib0Pi-_MI1tUhsm2z5bA&e= . 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 mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=hFe7 jo1OaOiJPYJN1sciBovML-jRiS7s_7h32CzMnCY&s=1X6nJq9b-3c- 2R4AgFUDMpC2l7itlL1ORo6Ipf9ljLs&e=
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 https://urldefense.proofpoint.com/v2/ur l?u=http- 3A__www.partners.org_complianceline&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1 VapJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=hFe 7jo1OaOiJPYJN1sciBovML- jRiS7s_7h32CzMnCY&s=lcIqRosuNio5NrM5LeJnk6gIGAk_7570yXenCjzgEdk&e= . 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 mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=hFe7 jo1OaOiJPYJN1sciBovML-jRiS7s_7h32CzMnCY&s=1X6nJq9b-3c- 2R4AgFUDMpC2l7itlL1ORo6Ipf9ljLs&e=
_______________________________________________ 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.
Hi Kersten,
I am sorry I did not make the question more specific.
I have three groups. I want see the power analysis by checking the sample size calculation for the cortical thickness analysis .
[sz1,sz2] = lme_plannedSampleSize(Zi,ZiCol,Dhat,phisqhat,effsz,... dr,pw,alpha,gr_pr)
my question is how can I get Dhat ( Estimated random effects covariance matrix) and phisqhat (Estimated intra-subject variability) to calculate the sample size of the pairs .
Based on your suggestion to get Dhat and phisqhat I have to run " lme_fit_FS(X,[1,2],y ,ni)"
When I run
lme_fit_FS(X,[1,2],y ,ni), where Y is obtained from the following
[Y,mri] = fs_read_Y('lh.thickness_sm10.mgh'), I get error.
Which function with what argument is good to use to find Dhat and phisqhat?
Are these values the same for left and right cortical thickness data?
I am sorry for the inconvenience .
with regards
Getaneh
________________________________________ From: freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mgh.ha "rvard.edu> on behalf of Diers, Kersten /DZNE Kersten.Diers@dzne.de Sent: Thursday, January 11, 2018 3:50 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME
Hi,
I am not entirely sure that I understand the goal of this particular analysis, so I don't have any suggestions at the moment.
Maybe you could give some more detail, if this is still an issue.
Best regards,
Kersten
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: Freesurfer support list freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME Date: Wed, 10 Jan 2018 17:58:58 +0100
Dear Kersten,
Thank you so much.
When I run the procedure with left and right cortical thickness data I get different phisqhat values Phisqhat:0.2530 for the left and Phisqhat:0.0903 for the right
The Dhat matrix is also different.
Is that Ok?
If so , shall I consider two sample sizes for the left and right?
With regards
Getaneh
-----Original Message----- From: freesurfer-bounces@nmr.mgh.harvard.edu [mailto:freesurfer-bounces @nmr.mgh.harvard.edu] On Behalf Of Diers, Kersten /DZNE Sent: Wednesday, January 10, 2018 7:10 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME
Hi Getaneh,
the estimates for D and phisq for a given analysis are contained within the data structure that is returned by the lme_fit_FS procedure.
I.e. for the tutorial data, the command was:
total_hipp_vol_stats = lme_fit_FS(X,[1 2],Y(:,1)+Y(:,2),ni);
Dhat and phisqhat are fields of the 'total_hipp_vol_stats' structure, and can be accessed by typing 'total_hipp_vol_stats.Dhat' and 'total_hipp_vol_stats.phisqhat', respectively.
Best regards,
Kersten
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: Freesurfer support list freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME Date: Tue, 9 Jan 2018 23:02:08 +0100
Hi Kersten.
Thank you very much .
I am not planning a new design for sample size calculation.
Can you please help me how to find those values?
Thank you
Getaneh
________________________________________ From: freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mg h.harvard.edu> on behalf of Diers, Kersten /DZNE <Kersten.Diers@dzne.d e
Sent: Tuesday, January 9, 2018 5:56 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME
Hi Getaneh,
please find my responses below.
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: freesurfer@nmr.mgh.harvard.edu freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] LME Date: Mon, 8 Jan 2018 22:54:49 +0100
Dear Freesurfer experts,
I have three groups g0, g1, and g2.
I am trying to use LME model with random effects y-int and time from the base line.
Based on the LME tutorial and the questions and answers from Freesurfer support list:
Y=B1+B2*t+B3*g1+B4*g1*t+B5*g2+B6*g2*t+ …
The slope of g0 : 0 1 0 0 0 0 The slope of g1: 0 1 0 1 0 0 The slope of g2: 0 1 0 0 0 1
The slope difference between g0 and g1: 0 0 0 1 0 0 The slope difference between g0 and g2: 0 0 0 0 0 1 The slope difference between g1 and g2: 0 0 0 -1 0 1
1) Is this contrast matrix correct to observe pairwise slope difference?
In the tutorial the contrast matrix C = [zeros(3,3) [1 0 0 0 0; -1 0 1 0 0; 0 0 -1 0 1] zeros(3,6)]
This is a correct matrix, but only for the particular design given in the tutorial, i.e. 4 groups and several covariates. Also, it is not strictly speaking a pairwise test, but will indicate if there is *any* difference between the slopes of the groups (i.e. it will give only one F- and p-value). If you specifically want to test differences between particular pairs, use contrast matrices with just a single row.
2) Is it necessary to combine contrast matrix of the pair difference the same as the tutorial?
If so, my contrast matrix can be C=[0 0 0 1 0 0; 0 0 0 -1 0 1]
It makes sense to look at the overall effect prior to looking at pairwise differences. If you don't have any additional covariates, that matrix is OK for looking the overall effect. Otherwise, add as many columns of zeros as you have covariates, similar to the tutorial.
3) How can we interpret the result for this contrast matrix?
This is an F-test to detect if there is any difference between the slopes of the groups.
4) How can I obtain B1, B2, … ?
Consider the following command from the tutorial:
total_hipp_vol_stats = lme_fit_FS(X,[1 2],Y(:,1)+Y(:,2),ni);
Then B1 and B2 will be the first two entries of total_hipp_vol_stats.Bhat
[sz1,sz2] = lme_plannedSampleSize(Zi,ZiCol,Dhat,phisqhat,effsz,... dr,pw,alpha,gr _pr)
5) For sample size calculation how can I get Dhat ( Estimated random effects covariance matrix) and phisqhat (Estimated intra-subject variability)?
If you are planning a new design, these estimates should be based on prior knowledge and can be derived e.g. from a pilot study or from the literature.
Best regards,
Kersten
With regards
Getaneh
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=w3Go ND77ORaREflQR6B1sxt8pIPGPzd8Oh_hzfJ_rhk&s=7GFfEuDrrPYWE- NE_mVYzsQUEvfZizwe853OSP2n_Zo&e=
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 https://urldefense.proofpoint.com/v2/ur l?u=http-3A__www.partners.org_co mplianceline&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1VapJQ8JVoF8oWH1AgfQ&r=f 2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=w3GoND77ORaREflQR6B1sxt8pI PGPzd8Oh_hzfJ_rhk&s=_JumR_-uS8QiP3xZfqqQ73pib0Pi-_MI1tUhsm2z5bA&e= . 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 mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=hFe7 jo1OaOiJPYJN1sciBovML-jRiS7s_7h32CzMnCY&s=1X6nJq9b-3c- 2R4AgFUDMpC2l7itlL1ORo6Ipf9ljLs&e=
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 https://urldefense.proofpoint.com/v2/ur l?u=http- 3A__www.partners.org_complianceline&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1 VapJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=hFe 7jo1OaOiJPYJN1sciBovML- jRiS7s_7h32CzMnCY&s=lcIqRosuNio5NrM5LeJnk6gIGAk_7570yXenCjzgEdk&e= . 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 mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=hFe7 jo1OaOiJPYJN1sciBovML-jRiS7s_7h32CzMnCY&s=1X6nJq9b-3c- 2R4AgFUDMpC2l7itlL1ORo6Ipf9ljLs&e=
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harvard.ed...
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 https://urldefense.proofpoint.com/v2/url?u=http-3A__www.partners.org_complia... . 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 mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harvard.ed...
Hi Getaneh,
thanks for the additional information.
I suspect that one source of confusion was that we have to distinguish the univariate and the mass-univariate processing stream in the LME toolbox.
The univariate stream is about a single measure like left or right or bilateral hippocampal volume, whereas the mass-univariate stream does statistical modeling and testing for all vertices in a hemisphere, for example to evaluate cortical thickness.
For a large part of my previous replies I was assuming that we are talking about the univariate stream, where we need to use the 'lme_fit_FS' script. In contrast, the 'lme_mass_fit_Rgw' script, among others, is used for estimation in the mass-univariate stream.
In that sense, it is not surprising to get error messages when using the 'lme_fit_FS' script with mass-univariate vertex data as in your example.
The bad news is that power analysis is only implemented for the univariate stream, but not the mass-univariate stream. This is because mass-univariate processing returns thousands of different estimates for phisq and Dhat, among others, one for each vertex in a hemisphere. We can get them from the data structure returned by 'lme_mass_fit_Rgw', but they are simply too many to submit to the power analysis.
In that sense, I currently see no viable way for power analysis for cortical thickness data within the LME toolbox unless we can reduce it to using estimates of a single (as opposed to multi-vertex) effect and its variabilty.
Likewise, it is not surprising that estimates for Dhat and phisqhat are different for e.g. the left and right hippocampus. One might speculate, though, that they should be somewhat similar, and I would be somewhat suprised if sample size calculations based on them result in very different sample size estimates.
I have one more remark about the lme_plannedSampleSize script: I would only recommend using it prospectively. As far as I can see it is not intended nor can it be meaningfully used retrospectively, i.e. for assessing the properties of a study that has already been conducted.
Best regards,
Kersten
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: freesurfer@nmr.mgh.harvard.edu freesurfer@nmr.mgh.harvard.edu Cc: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Subject: Re: [Freesurfer] LME Date: Thu, 11 Jan 2018 21:09:21 +0100
Hi Kersten,
I am sorry I did not make the question more specific.
I have three groups. I want see the power analysis by checking the sample size calculation for the cortical thickness analysis .
[sz1,sz2] = lme_plannedSampleSize(Zi,ZiCol,Dhat,phisqhat,effsz,... dr,pw,alpha,gr _pr)
my question is how can I get Dhat ( Estimated random effects covariance matrix) and phisqhat (Estimated intra-subject variability) to calculate the sample size of the pairs .
Based on your suggestion to get Dhat and phisqhat I have to run " lme_fit_FS(X,[1,2],y ,ni)"
When I run
lme_fit_FS(X,[1,2],y ,ni), where Y is obtained from the following
[Y,mri] = fs_read_Y('lh.thickness_sm10.mgh'), I get error.
Which function with what argument is good to use to find Dhat and phisqhat?
Are these values the same for left and right cortical thickness data?
I am sorry for the inconvenience .
with regards
Getaneh
________________________________________ From: freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mg h.ha "rvard.edu> on behalf of Diers, Kersten /DZNE <Kersten.Diers@dzne. de> Sent: Thursday, January 11, 2018 3:50 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME
Hi,
I am not entirely sure that I understand the goal of this particular analysis, so I don't have any suggestions at the moment.
Maybe you could give some more detail, if this is still an issue.
Best regards,
Kersten
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: Freesurfer support list freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME Date: Wed, 10 Jan 2018 17:58:58 +0100
Dear Kersten,
Thank you so much.
When I run the procedure with left and right cortical thickness data I get different phisqhat values Phisqhat:0.2530 for the left and Phisqhat:0.0903 for the right
The Dhat matrix is also different.
Is that Ok?
If so , shall I consider two sample sizes for the left and right?
With regards
Getaneh
-----Original Message----- From: freesurfer-bounces@nmr.mgh.harvard.edu [mailto:freesurfer-bounces @nmr.mgh.harvard.edu] On Behalf Of Diers, Kersten /DZNE Sent: Wednesday, January 10, 2018 7:10 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME
Hi Getaneh,
the estimates for D and phisq for a given analysis are contained within the data structure that is returned by the lme_fit_FS procedure.
I.e. for the tutorial data, the command was:
total_hipp_vol_stats = lme_fit_FS(X,[1 2],Y(:,1)+Y(:,2),ni);
Dhat and phisqhat are fields of the 'total_hipp_vol_stats' structure, and can be accessed by typing 'total_hipp_vol_stats.Dhat' and 'total_hipp_vol_stats.phisqhat', respectively.
Best regards,
Kersten
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: Freesurfer support list freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME Date: Tue, 9 Jan 2018 23:02:08 +0100
Hi Kersten.
Thank you very much .
I am not planning a new design for sample size calculation.
Can you please help me how to find those values?
Thank you
Getaneh
________________________________________ From: freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mg h.harvard.edu> on behalf of Diers, Kersten /DZNE <Kersten.Diers@dzne.d e
Sent: Tuesday, January 9, 2018 5:56 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME
Hi Getaneh,
please find my responses below.
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: freesurfer@nmr.mgh.harvard.edu freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] LME Date: Mon, 8 Jan 2018 22:54:49 +0100
Dear Freesurfer experts,
I have three groups g0, g1, and g2.
I am trying to use LME model with random effects y-int and time from the base line.
Based on the LME tutorial and the questions and answers from Freesurfer support list:
Y=B1+B2*t+B3*g1+B4*g1*t+B5*g2+B6*g2*t+ …
The slope of g0 : 0 1 0 0 0 0 The slope of g1: 0 1 0 1 0 0 The slope of g2: 0 1 0 0 0 1
The slope difference between g0 and g1: 0 0 0 1 0 0 The slope difference between g0 and g2: 0 0 0 0 0 1 The slope difference between g1 and g2: 0 0 0 -1 0 1
1) Is this contrast matrix correct to observe pairwise slope difference?
In the tutorial the contrast matrix C = [zeros(3,3) [1 0 0 0 0; -1 0 1 0 0; 0 0 -1 0 1] zeros(3,6)]
This is a correct matrix, but only for the particular design given in the tutorial, i.e. 4 groups and several covariates. Also, it is not strictly speaking a pairwise test, but will indicate if there is *any* difference between the slopes of the groups (i.e. it will give only one F- and p-value). If you specifically want to test differences between particular pairs, use contrast matrices with just a single row.
2) Is it necessary to combine contrast matrix of the pair difference the same as the tutorial?
If so, my contrast matrix can be C=[0 0 0 1 0 0; 0 0 0 -1 0 1]
It makes sense to look at the overall effect prior to looking at pairwise differences. If you don't have any additional covariates, that matrix is OK for looking the overall effect. Otherwise, add as many columns of zeros as you have covariates, similar to the tutorial.
3) How can we interpret the result for this contrast matrix?
This is an F-test to detect if there is any difference between the slopes of the groups.
4) How can I obtain B1, B2, … ?
Consider the following command from the tutorial:
total_hipp_vol_stats = lme_fit_FS(X,[1 2],Y(:,1)+Y(:,2),ni);
Then B1 and B2 will be the first two entries of total_hipp_vol_stats.Bhat
[sz1,sz2] = lme_plannedSampleSize(Zi,ZiCol,Dhat,phisqhat,effsz,... dr,pw,alpha,gr _pr)
5) For sample size calculation how can I get Dhat ( Estimated random effects covariance matrix) and phisqhat (Estimated intra-subject variability)?
If you are planning a new design, these estimates should be based on prior knowledge and can be derived e.g. from a pilot study or from the literature.
Best regards,
Kersten
With regards
Getaneh
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=w3Go ND77ORaREflQR6B1sxt8pIPGPzd8Oh_hzfJ_rhk&s=7GFfEuDrrPYWE- NE_mVYzsQUEvfZizwe853OSP2n_Zo&e=
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 https://urldefense.proofpoint.com/v2/ur l?u=http-3A__www.partners.org_co mplianceline&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1VapJQ8JVoF8oWH1AgfQ&r=f 2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=w3GoND77ORaREflQR6B1sxt8pI PGPzd8Oh_hzfJ_rhk&s=_JumR_-uS8QiP3xZfqqQ73pib0Pi-_MI1tUhsm2z5bA&e= . 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 mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=hFe7 jo1OaOiJPYJN1sciBovML-jRiS7s_7h32CzMnCY&s=1X6nJq9b-3c- 2R4AgFUDMpC2l7itlL1ORo6Ipf9ljLs&e=
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 https://urldefense.proofpoint.com/v2/ur l?u=http- 3A__www.partners.org_complianceline&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1 VapJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=hFe 7jo1OaOiJPYJN1sciBovML- jRiS7s_7h32CzMnCY&s=lcIqRosuNio5NrM5LeJnk6gIGAk_7570yXenCjzgEdk&e= . 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 mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=hFe7 jo1OaOiJPYJN1sciBovML-jRiS7s_7h32CzMnCY&s=1X6nJq9b-3c- 2R4AgFUDMpC2l7itlL1ORo6Ipf9ljLs&e=
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=5QnF HjjF94uyRV4EjwoCOEmFnHWg1Fidi6TqnPB6SEI&s=7acSXCq_P7kH1f_GHHvsOPRBaYwz7 FHXlD9BopUWVWg&e=
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 https://urldefense.proofpoint.com/v2/url?u=http-3A__www.partners.org_co mplianceline&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1VapJQ8JVoF8oWH1AgfQ&r=f 2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=5QnFHjjF94uyRV4EjwoCOEmFnH Wg1Fidi6TqnPB6SEI&s=opPTPIhD9uIAxefj3KJreZs1N4UXeDIcD1TUJA2CBic&e= . 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 mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=5QnF HjjF94uyRV4EjwoCOEmFnHWg1Fidi6TqnPB6SEI&s=7acSXCq_P7kH1f_GHHvsOPRBaYwz7 FHXlD9BopUWVWg&e=
_______________________________________________ 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.
Kersten
Thank you so much.
With regards
Getaneh
-----Original Message----- From: freesurfer-bounces@nmr.mgh.harvard.edu [mailto:freesurfer-bounces@nmr.mgh.harvard.edu] On Behalf Of Diers, Kersten /DZNE Sent: Friday, January 12, 2018 9:50 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME
Hi Getaneh,
thanks for the additional information.
I suspect that one source of confusion was that we have to distinguish the univariate and the mass-univariate processing stream in the LME toolbox.
The univariate stream is about a single measure like left or right or bilateral hippocampal volume, whereas the mass-univariate stream does statistical modeling and testing for all vertices in a hemisphere, for example to evaluate cortical thickness.
For a large part of my previous replies I was assuming that we are talking about the univariate stream, where we need to use the 'lme_fit_FS' script. In contrast, the 'lme_mass_fit_Rgw' script, among others, is used for estimation in the mass-univariate stream.
In that sense, it is not surprising to get error messages when using the 'lme_fit_FS' script with mass-univariate vertex data as in your example.
The bad news is that power analysis is only implemented for the univariate stream, but not the mass-univariate stream. This is because mass-univariate processing returns thousands of different estimates for phisq and Dhat, among others, one for each vertex in a hemisphere. We can get them from the data structure returned by 'lme_mass_fit_Rgw', but they are simply too many to submit to the power analysis.
In that sense, I currently see no viable way for power analysis for cortical thickness data within the LME toolbox unless we can reduce it to using estimates of a single (as opposed to multi-vertex) effect and its variabilty.
Likewise, it is not surprising that estimates for Dhat and phisqhat are different for e.g. the left and right hippocampus. One might speculate, though, that they should be somewhat similar, and I would be somewhat suprised if sample size calculations based on them result in very different sample size estimates.
I have one more remark about the lme_plannedSampleSize script: I would only recommend using it prospectively. As far as I can see it is not intended nor can it be meaningfully used retrospectively, i.e. for assessing the properties of a study that has already been conducted.
Best regards,
Kersten
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: freesurfer@nmr.mgh.harvard.edu freesurfer@nmr.mgh.harvard.edu Cc: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Subject: Re: [Freesurfer] LME Date: Thu, 11 Jan 2018 21:09:21 +0100
Hi Kersten,
I am sorry I did not make the question more specific.
I have three groups. I want see the power analysis by checking the sample size calculation for the cortical thickness analysis .
[sz1,sz2] = lme_plannedSampleSize(Zi,ZiCol,Dhat,phisqhat,effsz,... dr,pw,alpha,gr _pr)
my question is how can I get Dhat ( Estimated random effects covariance matrix) and phisqhat (Estimated intra-subject variability) to calculate the sample size of the pairs .
Based on your suggestion to get Dhat and phisqhat I have to run " lme_fit_FS(X,[1,2],y ,ni)"
When I run
lme_fit_FS(X,[1,2],y ,ni), where Y is obtained from the following
[Y,mri] = fs_read_Y('lh.thickness_sm10.mgh'), I get error.
Which function with what argument is good to use to find Dhat and phisqhat?
Are these values the same for left and right cortical thickness data?
I am sorry for the inconvenience .
with regards
Getaneh
________________________________________ From: freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mg h.ha "rvard.edu> on behalf of Diers, Kersten /DZNE <Kersten.Diers@dzne. de> Sent: Thursday, January 11, 2018 3:50 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME
Hi,
I am not entirely sure that I understand the goal of this particular analysis, so I don't have any suggestions at the moment.
Maybe you could give some more detail, if this is still an issue.
Best regards,
Kersten
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: Freesurfer support list freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME Date: Wed, 10 Jan 2018 17:58:58 +0100
Dear Kersten,
Thank you so much.
When I run the procedure with left and right cortical thickness data I get different phisqhat values Phisqhat:0.2530 for the left and Phisqhat:0.0903 for the right
The Dhat matrix is also different.
Is that Ok?
If so , shall I consider two sample sizes for the left and right?
With regards
Getaneh
-----Original Message----- From: freesurfer-bounces@nmr.mgh.harvard.edu [mailto:freesurfer-bounces @nmr.mgh.harvard.edu] On Behalf Of Diers, Kersten /DZNE Sent: Wednesday, January 10, 2018 7:10 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME
Hi Getaneh,
the estimates for D and phisq for a given analysis are contained within the data structure that is returned by the lme_fit_FS procedure.
I.e. for the tutorial data, the command was:
total_hipp_vol_stats = lme_fit_FS(X,[1 2],Y(:,1)+Y(:,2),ni);
Dhat and phisqhat are fields of the 'total_hipp_vol_stats' structure, and can be accessed by typing 'total_hipp_vol_stats.Dhat' and 'total_hipp_vol_stats.phisqhat', respectively.
Best regards,
Kersten
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: Freesurfer support list freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME Date: Tue, 9 Jan 2018 23:02:08 +0100
Hi Kersten.
Thank you very much .
I am not planning a new design for sample size calculation.
Can you please help me how to find those values?
Thank you
Getaneh
________________________________________ From: freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mg h.harvard.edu> on behalf of Diers, Kersten /DZNE <Kersten.Diers@dzne.d e
Sent: Tuesday, January 9, 2018 5:56 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME
Hi Getaneh,
please find my responses below.
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: freesurfer@nmr.mgh.harvard.edu freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] LME Date: Mon, 8 Jan 2018 22:54:49 +0100
Dear Freesurfer experts,
I have three groups g0, g1, and g2.
I am trying to use LME model with random effects y-int and time from the base line.
Based on the LME tutorial and the questions and answers from Freesurfer support list:
Y=B1+B2*t+B3*g1+B4*g1*t+B5*g2+B6*g2*t+ …
The slope of g0 : 0 1 0 0 0 0 The slope of g1: 0 1 0 1 0 0 The slope of g2: 0 1 0 0 0 1
The slope difference between g0 and g1: 0 0 0 1 0 0 The slope difference between g0 and g2: 0 0 0 0 0 1 The slope difference between g1 and g2: 0 0 0 -1 0 1
1) Is this contrast matrix correct to observe pairwise slope difference?
In the tutorial the contrast matrix C = [zeros(3,3) [1 0 0 0 0; -1 0 1 0 0; 0 0 -1 0 1] zeros(3,6)]
This is a correct matrix, but only for the particular design given in the tutorial, i.e. 4 groups and several covariates. Also, it is not strictly speaking a pairwise test, but will indicate if there is *any* difference between the slopes of the groups (i.e. it will give only one F- and p-value). If you specifically want to test differences between particular pairs, use contrast matrices with just a single row.
2) Is it necessary to combine contrast matrix of the pair difference the same as the tutorial?
If so, my contrast matrix can be C=[0 0 0 1 0 0; 0 0 0 -1 0 1]
It makes sense to look at the overall effect prior to looking at pairwise differences. If you don't have any additional covariates, that matrix is OK for looking the overall effect. Otherwise, add as many columns of zeros as you have covariates, similar to the tutorial.
3) How can we interpret the result for this contrast matrix?
This is an F-test to detect if there is any difference between the slopes of the groups.
4) How can I obtain B1, B2, … ?
Consider the following command from the tutorial:
total_hipp_vol_stats = lme_fit_FS(X,[1 2],Y(:,1)+Y(:,2),ni);
Then B1 and B2 will be the first two entries of total_hipp_vol_stats.Bhat
[sz1,sz2] = lme_plannedSampleSize(Zi,ZiCol,Dhat,phisqhat,effsz,... dr,pw,alpha,gr _pr)
5) For sample size calculation how can I get Dhat ( Estimated random effects covariance matrix) and phisqhat (Estimated intra-subject variability)?
If you are planning a new design, these estimates should be based on prior knowledge and can be derived e.g. from a pilot study or from the literature.
Best regards,
Kersten
With regards
Getaneh
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=w3Go ND77ORaREflQR6B1sxt8pIPGPzd8Oh_hzfJ_rhk&s=7GFfEuDrrPYWE- NE_mVYzsQUEvfZizwe853OSP2n_Zo&e=
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 https://urldefense.proofpoint.com/v2/ur l?u=http-3A__www.partners.org_co mplianceline&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1VapJQ8JVoF8oWH1AgfQ&r=f 2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=w3GoND77ORaREflQR6B1sxt8pI PGPzd8Oh_hzfJ_rhk&s=_JumR_-uS8QiP3xZfqqQ73pib0Pi-_MI1tUhsm2z5bA&e= . 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 mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=hFe7 jo1OaOiJPYJN1sciBovML-jRiS7s_7h32CzMnCY&s=1X6nJq9b-3c- 2R4AgFUDMpC2l7itlL1ORo6Ipf9ljLs&e=
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 https://urldefense.proofpoint.com/v2/ur l?u=http- 3A__www.partners.org_complianceline&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1 VapJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=hFe 7jo1OaOiJPYJN1sciBovML- jRiS7s_7h32CzMnCY&s=lcIqRosuNio5NrM5LeJnk6gIGAk_7570yXenCjzgEdk&e= . 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 mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=hFe7 jo1OaOiJPYJN1sciBovML-jRiS7s_7h32CzMnCY&s=1X6nJq9b-3c- 2R4AgFUDMpC2l7itlL1ORo6Ipf9ljLs&e=
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=5QnF HjjF94uyRV4EjwoCOEmFnHWg1Fidi6TqnPB6SEI&s=7acSXCq_P7kH1f_GHHvsOPRBaYwz7 FHXlD9BopUWVWg&e=
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 https://urldefense.proofpoint.com/v2/url?u=http-3A__www.partners.org_co mplianceline&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1VapJQ8JVoF8oWH1AgfQ&r=f 2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=5QnFHjjF94uyRV4EjwoCOEmFnH Wg1Fidi6TqnPB6SEI&s=opPTPIhD9uIAxefj3KJreZs1N4UXeDIcD1TUJA2CBic&e= . 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 mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=5QnF HjjF94uyRV4EjwoCOEmFnHWg1Fidi6TqnPB6SEI&s=7acSXCq_P7kH1f_GHHvsOPRBaYwz7 FHXlD9BopUWVWg&e=
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harvard.ed...
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 https://urldefense.proofpoint.com/v2/url?u=http-3A__www.partners.org_complia... . 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 mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harvard.ed...
External Email - Use Caution
Dear Kersten,
I have a question about how to interpret cortical thickness difference between groups.
I have three groups(A, B, and C) and five time points(t=0,0.5, 1,2,3).
To see the difference between group A and group B, I constructed the following contrast matrix using the tutorial provided by Freesurfer
C = [zeros(1,3) [1 0 0] zeros(1,4)]; CM.C=[zeros(1,3) [1 0 0] zeros(1,4)];
I run the following functions
rhstats = lme_mass_fit_Rgw(X,[1 2],rhY,ni,rhTh0,rhRgs,rhsphere); F_rhstats = lme_mass_F(rhstats,CM); fs_write_fstats(F_rhstats,rhmri,'/backup/cross/rh_AB_cross_sig.mgh','sig');
nv=length(rhstats); Beta2 = zeros(1,nv); for i=1:nv if ~isempty(rhstats(i).Bhat) Beta2(i) = rhstats(i).Bhat(2); end; end;
rhmri1 = rhmri; rhmri1.volsz(4) = 1; fs_write_Y(Beta2,rhmri1,'/backup/cross/rh_AB_cross_rhBeta2.mgh');
[detvtx,sided_pval,pth] = lme_mass_FDR2(F_rhstats.pval,F_rhstats.sgn,rhcortex,0.05,0);
fs_write_Y(sided_pval,rhmri1,'/backup/cross/rh_AB_cross_spval.mgh');
When I view rh_AB_cross_sig.mgh using tksurfer I see regions with red and blue color .
1) How do I know the difference between the two groups for 5 time points? 2) Does red means thickness of group A is larger than thickness of group B?
With regards
Getaneh
-----Original Message----- From: freesurfer-bounces@nmr.mgh.harvard.edu [mailto:freesurfer-bounces@nmr.mgh.harvard.edu] On Behalf Of Diers, Kersten /DZNE Sent: Friday, January 12, 2018 9:50 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME
Hi Getaneh,
thanks for the additional information.
I suspect that one source of confusion was that we have to distinguish the univariate and the mass-univariate processing stream in the LME toolbox.
The univariate stream is about a single measure like left or right or bilateral hippocampal volume, whereas the mass-univariate stream does statistical modeling and testing for all vertices in a hemisphere, for example to evaluate cortical thickness.
For a large part of my previous replies I was assuming that we are talking about the univariate stream, where we need to use the 'lme_fit_FS' script. In contrast, the 'lme_mass_fit_Rgw' script, among others, is used for estimation in the mass-univariate stream.
In that sense, it is not surprising to get error messages when using the 'lme_fit_FS' script with mass-univariate vertex data as in your example.
The bad news is that power analysis is only implemented for the univariate stream, but not the mass-univariate stream. This is because mass-univariate processing returns thousands of different estimates for phisq and Dhat, among others, one for each vertex in a hemisphere. We can get them from the data structure returned by 'lme_mass_fit_Rgw', but they are simply too many to submit to the power analysis.
In that sense, I currently see no viable way for power analysis for cortical thickness data within the LME toolbox unless we can reduce it to using estimates of a single (as opposed to multi-vertex) effect and its variabilty.
Likewise, it is not surprising that estimates for Dhat and phisqhat are different for e.g. the left and right hippocampus. One might speculate, though, that they should be somewhat similar, and I would be somewhat suprised if sample size calculations based on them result in very different sample size estimates.
I have one more remark about the lme_plannedSampleSize script: I would only recommend using it prospectively. As far as I can see it is not intended nor can it be meaningfully used retrospectively, i.e. for assessing the properties of a study that has already been conducted.
Best regards,
Kersten
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: freesurfer@nmr.mgh.harvard.edu freesurfer@nmr.mgh.harvard.edu Cc: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Subject: Re: [Freesurfer] LME Date: Thu, 11 Jan 2018 21:09:21 +0100
Hi Kersten,
I am sorry I did not make the question more specific.
I have three groups. I want see the power analysis by checking the sample size calculation for the cortical thickness analysis .
[sz1,sz2] = lme_plannedSampleSize(Zi,ZiCol,Dhat,phisqhat,effsz,... dr,pw,alpha,gr _pr)
my question is how can I get Dhat ( Estimated random effects covariance matrix) and phisqhat (Estimated intra-subject variability) to calculate the sample size of the pairs .
Based on your suggestion to get Dhat and phisqhat I have to run " lme_fit_FS(X,[1,2],y ,ni)"
When I run
lme_fit_FS(X,[1,2],y ,ni), where Y is obtained from the following
[Y,mri] = fs_read_Y('lh.thickness_sm10.mgh'), I get error.
Which function with what argument is good to use to find Dhat and phisqhat?
Are these values the same for left and right cortical thickness data?
I am sorry for the inconvenience .
with regards
Getaneh
________________________________________ From: freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mg h.ha "rvard.edu> on behalf of Diers, Kersten /DZNE <Kersten.Diers@dzne. de> Sent: Thursday, January 11, 2018 3:50 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME
Hi,
I am not entirely sure that I understand the goal of this particular analysis, so I don't have any suggestions at the moment.
Maybe you could give some more detail, if this is still an issue.
Best regards,
Kersten
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: Freesurfer support list freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME Date: Wed, 10 Jan 2018 17:58:58 +0100
Dear Kersten,
Thank you so much.
When I run the procedure with left and right cortical thickness data I get different phisqhat values Phisqhat:0.2530 for the left and Phisqhat:0.0903 for the right
The Dhat matrix is also different.
Is that Ok?
If so , shall I consider two sample sizes for the left and right?
With regards
Getaneh
-----Original Message----- From: freesurfer-bounces@nmr.mgh.harvard.edu [mailto:freesurfer-bounces @nmr.mgh.harvard.edu] On Behalf Of Diers, Kersten /DZNE Sent: Wednesday, January 10, 2018 7:10 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME
Hi Getaneh,
the estimates for D and phisq for a given analysis are contained within the data structure that is returned by the lme_fit_FS procedure.
I.e. for the tutorial data, the command was:
total_hipp_vol_stats = lme_fit_FS(X,[1 2],Y(:,1)+Y(:,2),ni);
Dhat and phisqhat are fields of the 'total_hipp_vol_stats' structure, and can be accessed by typing 'total_hipp_vol_stats.Dhat' and 'total_hipp_vol_stats.phisqhat', respectively.
Best regards,
Kersten
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: Freesurfer support list freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME Date: Tue, 9 Jan 2018 23:02:08 +0100
Hi Kersten.
Thank you very much .
I am not planning a new design for sample size calculation.
Can you please help me how to find those values?
Thank you
Getaneh
________________________________________ From: freesurfer-bounces@nmr.mgh.harvard.edu <freesurfer-bounces@nmr.mg h.harvard.edu> on behalf of Diers, Kersten /DZNE <Kersten.Diers@dzne.d e
Sent: Tuesday, January 9, 2018 5:56 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] LME
Hi Getaneh,
please find my responses below.
-----Original Message----- From: "Tefera, Getaneh B" Getaneh.B.Tefera@uth.tmc.edu Reply-to: Freesurfer support list freesurfer@nmr.mgh.harvard.edu To: freesurfer@nmr.mgh.harvard.edu freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] LME Date: Mon, 8 Jan 2018 22:54:49 +0100
Dear Freesurfer experts,
I have three groups g0, g1, and g2.
I am trying to use LME model with random effects y-int and time from the base line.
Based on the LME tutorial and the questions and answers from Freesurfer support list:
Y=B1+B2*t+B3*g1+B4*g1*t+B5*g2+B6*g2*t+ …
The slope of g0 : 0 1 0 0 0 0 The slope of g1: 0 1 0 1 0 0 The slope of g2: 0 1 0 0 0 1
The slope difference between g0 and g1: 0 0 0 1 0 0 The slope difference between g0 and g2: 0 0 0 0 0 1 The slope difference between g1 and g2: 0 0 0 -1 0 1
1) Is this contrast matrix correct to observe pairwise slope difference?
In the tutorial the contrast matrix C = [zeros(3,3) [1 0 0 0 0; -1 0 1 0 0; 0 0 -1 0 1] zeros(3,6)]
This is a correct matrix, but only for the particular design given in the tutorial, i.e. 4 groups and several covariates. Also, it is not strictly speaking a pairwise test, but will indicate if there is *any* difference between the slopes of the groups (i.e. it will give only one F- and p-value). If you specifically want to test differences between particular pairs, use contrast matrices with just a single row.
2) Is it necessary to combine contrast matrix of the pair difference the same as the tutorial?
If so, my contrast matrix can be C=[0 0 0 1 0 0; 0 0 0 -1 0 1]
It makes sense to look at the overall effect prior to looking at pairwise differences. If you don't have any additional covariates, that matrix is OK for looking the overall effect. Otherwise, add as many columns of zeros as you have covariates, similar to the tutorial.
3) How can we interpret the result for this contrast matrix?
This is an F-test to detect if there is any difference between the slopes of the groups.
4) How can I obtain B1, B2, … ?
Consider the following command from the tutorial:
total_hipp_vol_stats = lme_fit_FS(X,[1 2],Y(:,1)+Y(:,2),ni);
Then B1 and B2 will be the first two entries of total_hipp_vol_stats.Bhat
[sz1,sz2] = lme_plannedSampleSize(Zi,ZiCol,Dhat,phisqhat,effsz,... dr,pw,alpha,gr _pr)
5) For sample size calculation how can I get Dhat ( Estimated random effects covariance matrix) and phisqhat (Estimated intra-subject variability)?
If you are planning a new design, these estimates should be based on prior knowledge and can be derived e.g. from a pilot study or from the literature.
Best regards,
Kersten
With regards
Getaneh
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=w3Go ND77ORaREflQR6B1sxt8pIPGPzd8Oh_hzfJ_rhk&s=7GFfEuDrrPYWE- NE_mVYzsQUEvfZizwe853OSP2n_Zo&e=
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 https://urldefense.proofpoint.com/v2/ur l?u=http-3A__www.partners.org_co mplianceline&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1VapJQ8JVoF8oWH1AgfQ&r=f 2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=w3GoND77ORaREflQR6B1sxt8pI PGPzd8Oh_hzfJ_rhk&s=_JumR_-uS8QiP3xZfqqQ73pib0Pi-_MI1tUhsm2z5bA&e= . 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 mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=hFe7 jo1OaOiJPYJN1sciBovML-jRiS7s_7h32CzMnCY&s=1X6nJq9b-3c- 2R4AgFUDMpC2l7itlL1ORo6Ipf9ljLs&e=
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 https://urldefense.proofpoint.com/v2/ur l?u=http- 3A__www.partners.org_complianceline&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1 VapJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=hFe 7jo1OaOiJPYJN1sciBovML- jRiS7s_7h32CzMnCY&s=lcIqRosuNio5NrM5LeJnk6gIGAk_7570yXenCjzgEdk&e= . 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 mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=hFe7 jo1OaOiJPYJN1sciBovML-jRiS7s_7h32CzMnCY&s=1X6nJq9b-3c- 2R4AgFUDMpC2l7itlL1ORo6Ipf9ljLs&e=
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=5QnF HjjF94uyRV4EjwoCOEmFnHWg1Fidi6TqnPB6SEI&s=7acSXCq_P7kH1f_GHHvsOPRBaYwz7 FHXlD9BopUWVWg&e=
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 https://urldefense.proofpoint.com/v2/url?u=http-3A__www.partners.org_co mplianceline&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1VapJQ8JVoF8oWH1AgfQ&r=f 2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=5QnFHjjF94uyRV4EjwoCOEmFnH Wg1Fidi6TqnPB6SEI&s=opPTPIhD9uIAxefj3KJreZs1N4UXeDIcD1TUJA2CBic&e= . 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 mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harva rd.edu_mailman_listinfo_freesurfer&d=DwIGaQ&c=6vgNTiRn9_pqCD9hKx9JgXN1V apJQ8JVoF8oWH1AgfQ&r=f2w_aQllkjZisjakf5SS_8h7m7gQVFtFYcFC7K0Nk8k&m=5QnF HjjF94uyRV4EjwoCOEmFnHWg1Fidi6TqnPB6SEI&s=7acSXCq_P7kH1f_GHHvsOPRBaYwz7 FHXlD9BopUWVWg&e=
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harvard.ed...
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 https://urldefense.proofpoint.com/v2/url?u=http-3A__www.partners.org_complia... . 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 mailing list Freesurfer@nmr.mgh.harvard.edu https://urldefense.proofpoint.com/v2/url?u=https-3A__mail.nmr.mgh.harvard.ed...
freesurfer@nmr.mgh.harvard.edu
-
Diers, Kersten /DZNE -
Tefera, Getaneh B