Dear Expert:
Sorry, I re-organized my question and below are my design matrices for randomise tests for TBSS:
If I have 2 groups of subjects, 4 subjects in the Experimental Group1, and 4 subjects in the Experimental Group2 (so 8 subjects in total), and 2 scan points (Time1 is pre-experiment, and Time2 is post-experiment), and I want to see how the changes between Time1 and Time 2 differ between Group1 and Group2, would the following matrix and contrast work for me?
EB Group1 Group2 S1 S2 S3 S4 S5 S6 S7 S8
1 -1 0 1 0 0 0 0 0 0 0
2 -1 0 0 1 0 0 0 0 0 0
3 -1 0 0 0 1 0 0 0 0 0
4 -1 0 0 0 0 1 0 0 0 0
5 0 -1 0 0 0 0 1 0 0 0
6 0 -1 0 0 0 0 0 1 0 0
7 0 -1 0 0 0 0 0 0 1 0
8 0 -1 0 0 0 0 0 0 0 1
1 1 0 1 0 0 0 0 0 0 0
2 1 0 0 1 0 0 0 0 0 0
3 1 0 0 0 1 0 0 0 0 0
4 1 0 0 0 0 1 0 0 0 0
5 0 1 0 0 0 0 1 0 0 0
6 0 1 0 0 0 0 0 1 0 0
7 0 1 0 0 0 0 0 0 1 0
8 0 1 0 0 0 0 0 0 0 1
Is the EV of Group1 the difference of Time2-Time1 for Group1, and the EV of Group2 the difference of Time2-Time1 for Group2?
And for the contrasts:
Group1 Group2 S1 S2 S3 S4 S5 S6 S7 S8
C1: -1 1 0 0 0 0 0 0 0 0
C2: 1 -1 0 0 0 0 0 0 0 0
C3: 1 0 0 0 0 0 0 0 0 0
C4: 0 1 0 0 0 0 0 0 0 0
C1: Time2 - Time1 increment (or pre-post increment) is higher in Group2 than Group1 ?
C2: Time2 - Time1 increment (or pre-post increment) is higher in Group1 than Group2 ?
C3: Mean Time2 - Time 1 increment (or mean pre-post increment) for Group1 ?
C4: Mean Time2 - Time 1 increment (or mean pre-post increment) for Group2 ?
If the above designs are all correct, then if I want to control for 2 covariates, is below design correct (by adding 2 EVs: EV S9 and EV S10, filled with 0s)?
EB Group1 Group2 S1 S2 S3 S4 S5 S6 S7 S8 S9 S10
1 -1 0 1 0 0 0 0 0 0 0 0 0
2 -1 0 0 1 0 0 0 0 0 0 0 0
3 -1 0 0 0 1 0 0 0 0 0 0 0
4 -1 0 0 0 0 1 0 0 0 0 0 0
5 0 -1 0 0 0 0 1 0 0 0 0 0
6 0 -1 0 0 0 0 0 1 0 0 0 0
7 0 -1 0 0 0 0 0 0 1 0 0 0
8 0 -1 0 0 0 0 0 0 0 1 0 0
1 1 0 1 0 0 0 0 0 0 0 0 0
2 1 0 0 1 0 0 0 0 0 0 0 0
3 1 0 0 0 1 0 0 0 0 0 0 0
4 1 0 0 0 0 1 0 0 0 0 0 0
5 0 1 0 0 0 0 1 0 0 0 0 0
6 0 1 0 0 0 0 0 1 0 0 0 0
7 0 1 0 0 0 0 0 0 1 0 0 0
8 0 1 0 0 0 0 0 0 0 1 0 0
I hope this clarified my previous questions! Looking forward to the confirmation info!
Thank you so much!
Y.H
-----Original Message-----
From: Yuwen Hung
Sent: June 20, 2019 6:55 PM
To: 'FSL - FMRIB's Software Library' <[log in to unmask]>
Subject: RE: Help: longitudinal DTI processing with TBSS
Dear FSL Expert:
Regarding my question below (how to use TBSS for longitudinal dataset of 2 groups and 2 time points), the reason I wanted to look at the interaction effect between Group and Time Point is actually to know whether there is any brain differences between the 2 time point for one group rather than or compared to the other group.
I see in the FSL archive that an alternative way seems to be just look at one group at a time and do a paired t-test in randomise. This way I just need to feed into TBSS all time points of all subjects in that group, and then do the same with another group, separately, right? And any effect found this way can be tested for an interaction effect in SPSS I imagine.
My first question is whether this is a better way to examine my data, compared to finding a way to do 2 x 2 repeated measure mixed design ANOVA. I also don't know how to process TBSS in the latter case.
My second question is, if I go for the first way, looking at paired t-test within-group, is it possible to control for a covariate or two at all in this case?
Your assistance is being greatly appreciated.
Thank you all very much,
Y.H
-----Original Message-----
From: Yuwen Hung
Sent: June 20, 2019 3:35 PM
To: 'FSL - FMRIB's Software Library' <[log in to unmask]>
Subject: Help: longitudinal DTI processing with TBSS
Dear FSL Expert:
I wonder if I could get some help regarding a repeated measure mixed design for DTI data using TBSS.
We have 2 groups of subjects and 2 complete brain scans for each subject (before & after a training course).
The interest is to assess whether there is an interaction between “Group” and “Training (or Time Point)” effect on the diffusion data, which requires a 2 X 2 ANOVA repeated measure mixed design. Although randomise doesn’t do this, I see some alternative solution on the documentation - i.e., testing the group difference (using independent t-test) on the “difference” images of the 2 time points diffusion data (using fslmath/fslmerge functions). I wondered how this can be achieved with TBSS?
Should I run TBSS for individual time point first, creating the mean skeleton 4D data for each time point, and then use fslmath/fslmerge to calculate the difference images from the 2 sets skeleton 4D data, and then run TBSS using randomise independent t-test on the “difference” skeleton data (in 4D)?
Your help would be greatly appreciated.
Thank you very much,
Y.H.
########################################################################
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
|
