I don't think you can do that. A subject must belong to both a gender and a dx class, so you have to do it as an interaction.
On 07/06/2016 03:38 PM, Timothy Hendrickson wrote:
Hi Doug, I just wanted to make sure that we were on the same page. I may have been a little unclear in
what I would ideally like to do. For #2 what I would like to do is take into account age slopes and total intracranial slopes for
gender, and age slopes and total intracranial slopes for diagnosis. I was not implying that I wanted diagnosis and gender to interact.
Assuming that you understood this previously could you please explain why you would model
the matrix this way? I do not fully understand.
Respectfully,
-Tim
I did not get the same matrix. Here's how I would do it Columns 1-16 model the intercepts for all your classes Columns 17-20 model the ages for your gender-x-diagnosis subclasses Columns 21-24 model the ICV for your gender-x-diagnosis subclasses
On 07/06/2016 02:04 PM, Timothy Hendrickson wrote:
Freesurfer Support,
Thank you for your support with my previous line of questions regarding design matrix creation. I have manually created another design matrix and want to ensure that I designed it correctly.
An example of my FSGD file is as follows below: Class SITE 1-Male-Control Class SITE 1-Male-PATIENT Class SITE 1-Female-Control Class SITE 1-Female-PATIENT Class SITE 2-Male-Control Class SITE 2-Male-PATIENT Class SITE 2-Female-Control Class SITE 2-Female-PATIENT Class SITE 3-Male-Control Class SITE 3-Male-PATIENT Class SITE 3-Female-Control Class SITE 3-Female-PATIENT Class SITE 4-Male-Control Class SITE 4-Male-PATIENT Class SITE 4-Female-Control Class SITE 4-Female-PATIENT Variables age_at_scan total_intracranial_volume study site levels = 1, 2, 3, 4 gender levels = Male, Female diagnosis levels = PATIENT, Control age_at_scan = covariate age total_intracranial_volume = covariate total intracranial volume
What I would ideally like to do is:
- Take into account offset differences amongst diagnosis, gender, and study
site.
- Allowing a difference in age slope, and total intracranial volume slope
amongst the diagnosis and gender levels.
- Modeling the age slope and total intracranial volume slope as the same for
the study site levels. Let's image that the first participant is from site 1, Male, and control, and is 12 and has a TIV of 30,000. The second participant is from site 2, Female, and PATIENT, and is 14, and has a TIV of 25,000. My understanding of the design matrix would be as follows: 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 12 0 0 0 30000 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 14 0 0 0 25000 -Tim
Yes, create your matrix manually.
Those matrix lines are not quite right. The ages are in the correct column, but you need a 1 somewhere in columns 1-8 to indicate the class (ie, site/dx) that the subject is in.
On 05/26/2016 12:57 PM, Timothy Hendrickson wrote:
Hi Doug,
Thank you for such a prompt response. Just to be clear you are recommending that I manually create the matrix file right?
If so I want to ensure that I am understanding how to design the matrix file properly.
Let's imagine that the first participant is a control and is 13 and the second is a patient and is 15. My understanding is that the matrix file would be as follows: 0 0 0 0 0 0 0 0 13 0 0 0 0 0 0 0 0 0 0 15.
-Tim Previous correspondences are below:
You'll need a regressor for each of the 8 classes you describe below. You can use mri_glmfit to generate this (Xg.dat file) You'll need two more regressors for age, one for each diagnosis. If a subject (ie, row) is a control then the two values will be AGE 0. If the subject of the row is a patient, then the two values will be 0 AGE. You can then set up a Controls-Patients age (ie, interaction between dx and age) contrast like [0 0 0 0 0 0 0 0 1 -1] On 05/24/2016 02:30 PM, Timothy Hendrickson wrote:
Freesurfer Support,
I'd like to create a design matrix for a group analysis outside of the DODS and DOSS models. I understand that in order to do this the -X flag must be used. However, I have been unable to find examples of how to do this.
I am hoping to reveal a difference in thickness or gyrification amongst a clinical population. The data set contains two factors: diagnosis, and study site and one covariate: age. Diagnosis has two levels: controls, and patients. Study site has four levels, one level for each location the data has been collected from.
What I would ideally like to do is:
Take into account offset differences amongst diagnosis and study site.
Allowing a difference in age slope amongst the diagnosis levels
Modeling the age slope as the same for the study site levels
My FSGD file is designed as follows
Class SITE 1-Control Class SITE 1-PATIENT Class SITE 2-Control Class SITE 2-PATIENT Class SITE 3-Control Class SITE 3-PATIENT Class SITE 4-Control Class SITE 4-PATIENT Variables age_at_scan
study site levels = 1,2,3 and 4 diagnosis levels = PATIENT and Control age_at_scan = covariate age
Any advice would be greatly appreciated.
Respectfully,
Tim
-- Timothy Hendrickson Department of Psychiatry University of Minnesota Mobile: 507-259-3434 tel:507-259-3434 (texts okay)
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer