Kohkichi Hosoda wrote:
> Hi,
>
> I am a newbie to SPM2, so please forgive me for asking primitive
> questions.
>
> Now I am trying to compare some values in a single patient with a
> group of 27 normal controls on MRI. How can SPM2 best be used to
> compare a single patient with a group of controls in order to
> establish the pattern of regional reduction in the patient?
>
> I normalized each image and then tried to use "Two sample t-test" in
> "Basic models". For "Source images" I selected 27 control images and
> then the patient image. For "set up analysis" I typed "c c c... 27th c
> p." I input all "no" to "grand mean scaling", "Threshold masking",
> "threshold", "Implicit mask", "explicitly mask", "Global calculation",
> "non-sphericity correction". Then I clicked "estimate". After
> calculation I clicked "Results". In SPM contrast manager, I selected
> "Define new contrast". I entered "control>patient" for name,
> "t-contrast for type" and "1 -1" for contrast. Or should I enter "-1 1"?
>
This is correct.
As the control images were entered first then, for control>patient,
1 -1 is the appropriate contrast.
(Check: your design matrix should look like
X=[ones(26,1),zeros(26,1); 0 1];
imagesc(X))
You may also be interested in the F-contrast [1 -1] which looks for
significant positive *or* negative changes.
Best,
Will.
> I would appreciate it very much if somebody could teach me whether the
> procedures described above are correct or wrong.
>
> Sincerely,
>
> --
> Kohkichi Hosoda M.D.
>
> Department of Neurosurgery,
> Hyogo Emergency Medical Center/Kobe Red Cross Hospital,
> 1-3-1 Wakinohama-Kaigan-Dori, Chuo-ku, Kobe, 651-0073, JAPAN
> Telephone 078-241-3131
> Fax 078-241-2772
> e-mail address : [log in to unmask]
>
>
--
William D. Penny
Wellcome Department of Imaging Neuroscience
University College London
12 Queen Square
London WC1N 3BG
Tel: 020 7833 7475
FAX: 020 7813 1420
Email: [log in to unmask]
URL: http://www.fil.ion.ucl.ac.uk/~wpenny/
|