search

gablab / analysis

0 stars 2 forks source link

Is there a bips workflow that does paired group comparison for longitudinal dataset? #7

Open zhenghanQ opened 10 years ago

zhenghanQ commented 10 years ago

I thought workflow 51ef can do something there? I can also learn from an example, if there is one. :)

cdla commented 10 years ago

edit: my bad! thought it was a single group scanned twice. removed as to not have the wrong answer up ;D

satra commented 10 years ago

typically this involves a repeated measures 2x2 anova. that's not easily done. but what you could do is subtract t1 from t2 and do a two sample ttest.

zhenghanQ commented 10 years ago

Actually Carlo's comment helped a lot! I did not express my question clearly in my original post. I did mean to do a paired t-test within a single group. Now the issue is solved with Carlo's help. Yes, workflow 51ef is able to do such analysis. If you are also interested in how to set up the contrast, here is the official explanation: http://fsl.fmrib.ox.ac.uk/fsl/fsl4.0/feat5/detail.html Here is an example of the contrast setting: "reg_contrasts": "def getcontrasts():\n contrasts = [(\"pre-post\",\"T\",[\"prepost\"],[1]),\n (\"post-pre\",\"T\",[\"prepost\"],[-1])]\n return contrasts", You can find my design csv file here: /gablab/p/CASL/Analysis/sent/S1S2_prepost_paired_sent.csv

I understand Satra's comment. If you wanna compare the training effect across two different groups, such analysis will require the access to the training effect at single subject level. In that case, the most direct way is to enter the sessions of each subject separately into the workflow 51ef or f08f and then loop through all the participant. After that step, compare the groups on average of session difference. Is my understanding right?