Creating contrast between two groups, controlling for 2 categorical(gender, scanner) and one continous variable(age)
Dear FreeSurfer Experts,
I want to compare structural measures(eg. thickness) between patients and controls controling for the effects of age, gender and scanner (parts of sample were scanned in scanner 1 parts in scanner 2).
I want to answer the question: *do patients and controls differ significantly in lGI when controlling for age (continous variable), gender(male/female), scanner(scanner1/scanner2)?*
*so i have to categorical (gender, scanner) and one continous variable to control for.*
*my FSGD file looks like:*
/GroupDescriptorFile 1 // //Class PatientFemaleScanner1// //Class PatientFemaleScanner2// //Class PatientMaleScanner1// //Class PatientMaleScanner2// //Class ControlFemaleScanner1// //Class ControlFemaleScanner2// //Class ControlMaleScanner1// //Class ControlMaleScanner2// //Variables Age // //Input subject1 ControlMaleScanner1 23.6// //Input subject2 ControlFemaleScanner2 37.7// //Input subject3 PatientFemaleScanner1 32.6/
...
*Total I would have 16 regressors with following contrasts: *
*first conrast:* is there a difference in structural measure(eg. thickness) between *patients and controls* controlling for gender, scanner, age?
0.25 0.25 0.25 0.25 -0.25 -0.25 -0.25 -0.25 0 0 0 0 0 0 0 0
*second contrast:* is there a difference between the two scanners *scanner 1 and 2* controlling for effects of gender, age and group?
0.25 -0.25 0.25 -0.25 0.25 -0.25 0.25 -0.25 0 0 0 0 0 0 0 0
*Questions:*
* Are the Contrasts created ok? * How could I control better for the effect of scanner, as part of the sample was scanned with scanner 1 another part with scanner 2?
Thank you very much for your help!
Geo
The contrasts are correct. I think you have done the right thing (or as good as you can) in controlling for the scanners.
On 8/1/16 10:46 AM, Oana Georgiana Rus wrote:
Dear FreeSurfer Experts,
I want to compare structural measures(eg. thickness) between patients and controls controling for the effects of age, gender and scanner (parts of sample were scanned in scanner 1 parts in scanner 2).
I want to answer the question: *do patients and controls differ significantly in lGI when controlling for age (continous variable), gender(male/female), scanner(scanner1/scanner2)?*
*so i have to categorical (gender, scanner) and one continous variable to control for.*
*my FSGD file looks like:*
/GroupDescriptorFile 1 // //Class PatientFemaleScanner1// //Class PatientFemaleScanner2// //Class PatientMaleScanner1// //Class PatientMaleScanner2// //Class ControlFemaleScanner1// //Class ControlFemaleScanner2// //Class ControlMaleScanner1// //Class ControlMaleScanner2// //Variables Age // //Input subject1 ControlMaleScanner1 23.6// //Input subject2 ControlFemaleScanner2 37.7// //Input subject3 PatientFemaleScanner1 32.6/
...
*Total I would have 16 regressors with following contrasts:
*first conrast:* is there a difference in structural measure(eg. thickness) between *patients and controls* controlling for gender, scanner, age?
0.25 0.25 0.25 0.25 -0.25 -0.25 -0.25 -0.25 0 0 0 0 0 0 0 0
*second contrast:* is there a difference between the two scanners *scanner 1 and 2* controlling for effects of gender, age and group?
0.25 -0.25 0.25 -0.25 0.25 -0.25 0.25 -0.25 0 0 0 0 0 0 0 0
*Questions:*
- Are the Contrasts created ok?
- How could I control better for the effect of scanner, as part of the sample was scanned with scanner 1 another part with scanner 2?
Thank you very much for your help!
Geo
-- Oana Georgiana Rus PhD Student Neuroimaging Center TUM-NIC Klinikum rechts der Isar Technische Universität München Einsteinstr.1 81675 München Raum 5.8
Tel. 089 4140 7971
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Douglas Greve -
Oana Georgiana Rus