Design Matrix Creation - Freesurfer

6 Jul 2016

      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:
1) Take into account offset differences amongst diagnosis, gender, and
study site.
2) Allowing a difference in age slope, and total intracranial volume
slope amongst the diagnosis and gender levels.
3) 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)