Dear FreeSurfer experts,
I am trying to use the LME toolbox to analyse the hippocampal change over time in a group and also among 2 groups (2 scans per subject).
When I estimate the parameters with lme_fit_FS I become a warning saying that the matrix is singular and my results are NaN. I guess that my design matrix X is not correct, can you please help me?
I have followed the example from the wiki. My model for the two groups study is the following:
Yij = ß1 + ß2*tij +ß3*Group1 + ß4*Group1 *tij + ß5*Group2 + ß6*Group2 *tij+ ß 7*Genderi + ß8*Agei + ß9*ICVi + b1i + b2i*tij + eij
1) intercept (all ones)
2) time (tij) . Here I have 0s for the first scan of each subject and a number in years for the second scans. Is it correct?
3) one for Group1 and zero otherwise
4) colum 3) .* time
5) one for Group2 and zero otherwise
6) colum 5) .* time
7) Gender
8) Age of the subject at each scan moment.
9) ICV (converted to liters).
Thank you in advance, Yolanda
Hi Yolanda,
since you have the intercept you don't need both group1 and group2 (as one is 1 - the other), so drop beta 5 and 6 terms. If you want to test if the slope is different between the groups differ just look at beta4 which is checking if slope1-slope2 different from zero. If it is negative slope2 is larger, if positive slope1 is larger. Note that if your slopes are negative, e.g. atrophy, then larger means closer to zero = less atrophy.
Best, Martin
On 05/28/2013 11:57 AM, Yolanda Vives wrote:
Dear FreeSurfer experts,
I am trying to use the LME toolbox to analyse the hippocampal change over time in a group and also among 2 groups (2 scans per subject).
When I estimate the parameters with lme_fit_FS I become a warning saying that the matrix is singular and my results are NaN. I guess that my design matrix X is not correct, can you please help me?
I have followed the example from the wiki. My model for the two groups study is the following:
Y_ij = ß_1 + ß_2 *t_ij +ß_3 *Group1 + ß_4 *Group1 *t_ij + ß_5 *Group2
- ß_6 *Group2 *t_ij + ß_7 *Gender_i + ß_8 *Age_i + ß_9 *ICV_i + b_1i +
b_2i *t_ij + e_ij
intercept (all ones)
time (t_ij ) . Here I have 0s for the first scan of each subject
and a number in years for the second scans. Is it correct?
one for Group1 and zero otherwise
colum 3) .* time
one for Group2 and zero otherwise
colum 5) .* time
Gender
Age of the subject at each scan moment.
ICV (converted to liters).
Thank you in advance, Yolanda
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Thank you Martin!
For only one group, would the design matrix be the same but intercept=Group=all ones?
Yij = ß1 + ß2*tij +ß3*Group + ß4*Group *tij + ß5*Genderi + ß6*Agei + ß7*ICVi+ b 1i + b2i*tij + eij
Regards, Yolanda
2013/5/28 Martin Reuter mreuter@nmr.mgh.harvard.edu
Hi Yolanda,
since you have the intercept you don't need both group1 and group2 (as one is 1 - the other), so drop beta 5 and 6 terms. If you want to test if the slope is different between the groups differ just look at beta4 which is checking if slope1-slope2 different from zero. If it is negative slope2 is larger, if positive slope1 is larger. Note that if your slopes are negative, e.g. atrophy, then larger means closer to zero = less atrophy.
Best, Martin
On 05/28/2013 11:57 AM, Yolanda Vives wrote:
Dear FreeSurfer experts,
I am trying to use the LME toolbox to analyse the hippocampal change over time in a group and also among 2 groups (2 scans per subject).
When I estimate the parameters with lme_fit_FS I become a warning saying that the matrix is singular and my results are NaN. I guess that my design matrix X is not correct, can you please help me?
I have followed the example from the wiki. My model for the two groups study is the following:
Yij = ß1 + ß2*tij +ß3*Group1 + ß4*Group1 *tij + ß5*Group2 + ß6*Group2 *tij+ ß 7*Genderi + ß8*Agei + ß9*ICVi + b1i + b2i*tij + eij
intercept (all ones)
time (tij) . Here I have 0s for the first scan of each subject and a
number in years for the second scans. Is it correct?
one for Group1 and zero otherwise
colum 3) .* time
one for Group2 and zero otherwise
colum 5) .* time
Gender
Age of the subject at each scan moment.
ICV (converted to liters).
Thank you in advance, Yolanda
Freesurfer mailing listFreesurfer@nmr.mgh.harvard.eduhttps://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- Martin Reuter, Ph.D. Assistant in Neuroscience - Massachusetts General Hospital Instructor in Neurology - Harvard Medical School MGH / HMS / MIT
A.A.Martinos Center for Biomedical Imaging 149 Thirteenth Street, Suite 2301 Charlestown, MA 02129
Phone: +1-617-724-5652 Email: mreuter@nmr.mgh.harvard.edu reuter@mit.edu Web : http://reuter.mit.edu
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.
For only one group you don't need a group variable. you'd drop beta 3 and 4 terms.
Yolanda Vives yvives@pic.es wrote:
Thank you Martin!
For only one group, would the design matrix be the same but intercept=Group=all ones?
Yij = ß1 + ß2*tij +ß3*Group + ß4*Group *tij + ß5*Genderi + ß6*Agei + ß7*ICVi+ b 1i + b2i*tij + eij
Regards, Yolanda
2013/5/28 Martin Reuter mreuter@nmr.mgh.harvard.edu
Hi Yolanda,
since you have the intercept you don't need both group1 and group2
(as one
is 1 - the other), so drop beta 5 and 6 terms. If you want to test if the slope is different between the groups
differ
just look at beta4 which is checking if slope1-slope2 different from
zero.
If it is negative slope2 is larger, if positive slope1 is larger.
Note that
if your slopes are negative, e.g. atrophy, then larger means closer
to zero
= less atrophy.
Best, Martin
On 05/28/2013 11:57 AM, Yolanda Vives wrote:
Dear FreeSurfer experts,
I am trying to use the LME toolbox to analyse the hippocampal change
over
time in a group and also among 2 groups (2 scans per subject).
When I estimate the parameters with lme_fit_FS I become a warning
saying
that the matrix is singular and my results are NaN. I guess that my
design
matrix X is not correct, can you please help me?
I have followed the example from the wiki. My model for the two
groups
study is the following:
Yij = ß1 + ß2*tij +ß3*Group1 + ß4*Group1 *tij + ß5*Group2 + ß6*Group2
*tij+ ß
7*Genderi + ß8*Agei + ß9*ICVi + b1i + b2i*tij + eij
intercept (all ones)
time (tij) . Here I have 0s for the first scan of each subject and
a
number in years for the second scans. Is it correct?
one for Group1 and zero otherwise
colum 3) .* time
one for Group2 and zero otherwise
colum 5) .* time
Gender
Age of the subject at each scan moment.
ICV (converted to liters).
Thank you in advance, Yolanda
Freesurfer mailing
listFreesurfer@nmr.mgh.harvard.eduhttps://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- Martin Reuter, Ph.D. Assistant in Neuroscience - Massachusetts General Hospital Instructor in Neurology - Harvard Medical School MGH / HMS / MIT
A.A.Martinos Center for Biomedical Imaging 149 Thirteenth Street, Suite 2301 Charlestown, MA 02129
Phone: +1-617-724-5652 Email: mreuter@nmr.mgh.harvard.edu reuter@mit.edu Web : http://reuter.mit.edu
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
-
Martin Reuter -
Yolanda Vives