I have several sessions from 30 subjects, and I want to test if the HDR changes from session 0 to 1 (S00 vs S01). So far, I’m trying to use the idea from Gang Chen’s 2015 paper on this and am using 3dLME. I have missing data and only one group. When looking at the data, the HDR is decreased at most points along the curve in S01 compared to S00 (20sec total at 0.5sec TR, giving 39 time points). To test for this, I’m using the command:
3dLME -prefix lmeS00vsS01_3 -jobs 8
-corSt ‘Order : AR1’
-gltLabel 1 ‘S00minusS01’ -gltCode 1 ‘Session : 1S00 -1S01’
-dataTable Subj Session Time Order InputFile
P0204 S00 t00 0 iresp_stim01.P0204_S00_masked+tlrc
P0204 S00 t01 1 iresp_stim01.P0204_S00_masked+tlrc
P0204 S00 t02 2 iresp_stim01.P0204_S00_masked+tlrc
Is this the appropriate way to test for this difference?
First, use 3dcalc to obtain the difference between the two sessions at each time point. Then following Example 1 in the 3dLME help.
Thanks so much for the help! I used 3dcalc to get the difference and followed Example 1. The model will run and gives me an Fstat and mean/tstat’s for all of the time points, but the sub-bricks for the GLTs are all zero. Why would this be? I tried several of the 3dLME examples in AFNI_data6/GroupAna_cases to see if there were a problem with my R installation, but they all ran fine and produced outputs.
Here’s the command I’m using. The subbricks for ‘FourTimePoints’ and ‘ExampleTTest’ are all zero.
3dLME -prefix myOutput_S00minusS01 -jobs 8
-corStr ‘Order : AR1’
-glfLabel 1 ‘FourTimePoints’
-glfCode 1 ‘Time : 1Diff03 & 1Diff04 & 1Diff05 & 1Diff06’
-gltLabel 1 ‘ExampleTTest’
-gltCode 1 ‘Time : 1Diff05 +1Diff06’
-dataTable Subj Time Order InputFile
Y4_P0204 Diff00 00 /work/Trey/diffTentHDR/S00minusS01/S00minusS01.Y4_P0204+tlrc.HEAD
Y4_P0204 Diff01 01 /work/Trey/diffTentHDR/S00minusS01/S00minusS01.Y4_P0204+tlrc.HEAD
It does seem that there is some problem with 3dLME. Let me take a close look and get back to you later.