raghu venkatram wrote:
> Hi Will,
>
> Thank you for the much needed/valueable information.
>
> On the same lines and setup, interpreting the t maps and the respective
> contrast images,
>
> * the t maps give me an idea of how well the 2 conditions were modelled
>
The t-statistic resulting from say the 1 -1 contrast
is equal to (effect_size)/(effect_variability) where the
effect is the difference in response between the
two conditions.
> * also the tmaps are not thresholded to a given p value
>
They are not thresholded per se. But when you
display them in SPM you only really want to display
thresholded maps where the threshold
is chosen st. the chance of a false positive anywhere
in the image is less than 0.05 (this is the FWE=0.05 option).
So the threshold is useful for focussing on either
strong effects or weak but consistent effects.
> * can i interpret the contrast images(intensity values) to be a good
> indicator of the increase or decrease in metabolism for areas where the
> correspoding t values are higher than a threshold?
>
>
yes. the contrast images correspond to the (effect_size).
best wishes,
Will.
>
> Thank you in advance
>
> Raghu
>
> >From: Will Penny
>
> >To: raghu venkatra
>
> >CC: [log in to unmask]
>
> >Subject: Re: paired t test setup/ second level analysis
>
> >Date: Tue, 29 Jul 2003 11:33:00 +0100
>
> >
>
> >
>
> >
>
> >raghu venkatra wrote:
>
> >
>
> >>Hello SPMers,
>
> >>
>
> >>I ran a paired t test on a 4-subject study. The data I have
>
> >>consists of two
>
> >>scans per subject (drug/vehicle). I chose the PET population main
>
> >>effect
>
> >>(paired t test). The null hypothesis I believe I am testing is that
>
> >>the
>
> >>drug has no effect. I used a [1,-1] contrast for increase in
>
> >>metabolism and
>
> >>[-1,1] for decrease in metabolism.
>
> >>I am confused as to if I am doing a model analysis here, and will
>
> >>have to
>
> >>use the two contrast images for a second level analysis using the
>
> >>basic
>
> >>model two-sample t test.
>
> >>
>
> >
>
> >
>
> >I would recommend that you don't do a 2nd-level analysis (ie.
>
> >a random effect analysis) with this data because you
>
> >don't have enough subjects.
>
> >
>
> >
>
> >>Some questions:
>
> >>*Do the SPM_T*.img maps indicate increase/decrease in metabolism
>
> >>for the
>
> >>contrats?
>
> >
>
> >
>
> >Yes. [1 -1] would be relative increase in metabolism due to drug.
>
> >Usually you'd do a [1 -1] contrast for each subject at the first
>
> >level
>
> >and then do a one sample t-test at the 2nd. But with 4 subjects
>
> >this gives you only 3 degrees of freedom which is'nt enough.
>
> >
>
> >
>
> >>*If I do perform a second level analysis, two sample t test on the
>
> >>con*.img
>
> >>images, for the above two contrasts, what is the null hypothesis?
>
> >
>
> >
>
> >The null hypothesis would be that, in the population at large, there
>
> >is no
>
> >difference in activation between the two conditions.
>
> >
>
> >
>
> >>*in general does first level analysis, maps basically an analysis
>
> >>of the
>
> >>estimated model?
>
> >
>
> >
>
> >I would recommend just analysing each subject separately and
>
> >reporting the
>
> >results as a series of case studies (unless you can get data on
>
> >another 8 subjects).
>
> >
>
> >So 'at the first level' you plot maps of regions where
>
> >you can reject the null hypothesis: In subject X at location y there
>
> >is
>
> >no difference in activation between the two conditions.
>
> >
>
> >Hope this helps, Will.
>
> >
>
> >
>
> >>
>
> >>
>
> >>Thank you in advance.
>
> >>Raghu Venkatram
>
> >>
>
> >>
>
> >>
>
> >
>
> >
>
> >--
>
> >William D. Penny
>
> >Wellcome Department of Imaging Neuroscience
>
> >University College London
>
> >12 Queen Square
>
> >London WC1N 3BG
>
> >
>
> >Tel: 020 7833 7478
>
> >FAX: 020 7813 1420
>
> >Email: [log in to unmask]
>
> >URL: http://www.fil.ion.ucl.ac.uk/~wpenny/
>
> >
>
> >
>
>
> ------------------------------------------------------------------------
> Add photos to your messages with MSN 8.
> <http://g.msn.com/8HMHENUS/2749??PS=> Get 2 months FREE*.
--
William D. Penny
Wellcome Department of Imaging Neuroscience
University College London
12 Queen Square
London WC1N 3BG
Tel: 020 7833 7478
FAX: 020 7813 1420
Email: [log in to unmask]
URL: http://www.fil.ion.ucl.ac.uk/~wpenny/
|