Dear Ming-Ting
You should create a fixed effect model with three seperate sessions for a single
subject analysis. You enter all the scans in this single subject three session
model. Then you can create a contrast image over the three session for each
subject. Finally, you enter these single contrast images of each subject into
your second level analysis.
Good luck
Frank
"Wu, Ming-Ting" wrote:
> Dear Frank:
> Thank you for your answer.
> To do your suggestion, I need to get a contrast image for each subject
> first. The problem is that each subject only run 3 sessions, it is not
> sufficient to run a one-sample t test. So, how could I get a contrast image
> for a subject run 3 sessions?
>
> Sincerely
>
> Ming-Ting
>
> ================================
> Wu, Ming-Ting, M.D. Depart. of Radiology
> Kaohsiung Veterans General Hospital, 813
> Kaohsiung , Taiwan
> Tel: 886-7-3422121 ext 6205, 6235
> Fax: 886-7-3468301
> Mobil: 0920-040-712
> ----- Original Message -----
> ±H¥óªÌ: Frank Jessen <[log in to unmask]>
> ¦¬¥óªÌ: Wu, Ming-Ting <[log in to unmask]>
> ¶Ç°e¤é´Á: 2000¦~10¤ë25¤é PM 02:01
> ¥D¦®: Re: multi-subject mult-session 2nd level
>
> Dear Ming-Ting,
> I would assume that entering 15 images of 5 subjects with 3 sessions each in
> a
> one-sample t-Test is not correct, because the three session of each subject
> are
> not independent. I think you need to model the three sessions on a single
> subject basis and create one contrast image for each subject, resulting in
> five
> images for the t-test. This however is a very small number for a t-test,
> where
> you should have at least ten images. I think the best way would be to
> include
> at least ten subjects in your study and enter one contrast image per subject
> in
> your statistic.
> Good luck
>
> Frank
>
> "Wu, Ming-Ting" wrote:
>
> > Dear SPM group:
> > I have a basic question which may have been addressed. Please be patient
> to
> > help anyway.
> > we do fMRI study on 5 subjects, each has three sessions, each session
> > generate a contrast image, so there are 15 contrast images.
> > To do one sample t test, how should I input the 15 images? is there a way
> to
> > differentiate my setting from 15 subjects, each have one session?
> >
> > Sincerely
> >
> > Ming-Ting
> >
> > ================================
> > Wu, Ming-Ting, M.D. Depart. of Radiology
> > Kaohsiung Veterans General Hospital, 813
> > Kaohsiung , Taiwan
> > Tel: 886-7-3422121 ext 6205, 6235
> > Fax: 886-7-3468301
> > Mobil: 0920-040-712
>
> --
> Dr. med Frank Jessen
> Klinik für Psychiatrie und Psychotherapie
> Universität Bonn
> Sigmund-Freud-Str. 25
> 53105 Bonn
> Tel: 0228-287-5743
> Fax: 0228-287-6097
--
Dr. med Frank Jessen
Klinik für Psychiatrie und Psychotherapie
Universität Bonn
Sigmund-Freud-Str. 25
53105 Bonn
Tel: 0228-287-5743
Fax: 0228-287-6097
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|