 |
 |
Hello! Thanks for the answer. I used the second option and ran 5000 permutations. It takes "its time"... I read that it's normal. Short final question: How many times is it running the permutations? I saw that it finished with the first one, gave the first results and started a second round. Is it normal? So far I haven't found any exact information about it in the previous questions from the mailing list. Thanks a lot again! Yacila On Oct 18, 2014 9:24 PM, "Anderson M. Winkler" <[log in to unmask]> wrote: Hi Yacila, There are two ways to run, and both give the same results: 1) For each subject, subtract timepoint 1 from timepoint 2 (that is, t2-t1), producing 1 volume per subject. Concatenate all in a 4D file, and use a 1-sample t-test in randomise (i.e., the design is a single column with ones, and the option -1 is used). or 2) Don't do the subtractions, making instead a paired t-test. It seems you could use the same design you have, but in randomise, use the option -e to specify a .grp file containing the exchangeability blocks, that here means one block per subject (so, 20 blocks). All the best, Anderson On 17 October 2014 18:16, Deza Araujo, Yacila Isabela <[log in to unmask]<mailto:[log in to unmask]>> wrote: Sorry! 40 rows and 21 columns Yacila Von: FSL - FMRIB's Software Library [mailto:[log in to unmask]<mailto:[log in to unmask]>] Im Auftrag von Deza Araujo, Yacila Isabela Gesendet: Friday, October 17, 2014 7:14 PM An: [log in to unmask]<mailto:[log in to unmask]> Betreff: [FSL] Randomise using GLM Dear FSL experts, I am trying to run dual regression as the last step of my resting state analysis in a sample of 20 subjects with two sessions each one (drug and placebo). I considered a paired T test (same group) and followed the instructions of the GLM for such test. The question is: Should I run the dual regression with my GLM.mat GLM.con or just -1 instead? I tried to use my design so far and I got the error “number of rows in design matrix doesn´t match number of time points”. I have 40 rows (one for each subject condition) and 21 rows (one for modeling the possible group interaction and the rest for every subject). What is the correct design that I should use with my sample then? Many thanks for your help! -- Yacila Deza Araujo, M.Sc. Neuropsych. PhD Student Technische Universität Dresden Faculty of Medicine Carl Gustav Carus Department of Psychiatry & Psychotherapy Section of Systems Neuroscience