Hi,
On 16 Nov 2007, at 13:57, Agnieszka Burzynska wrote:
> Hi,
> I tried to read all (I hope) posts on this topic, but some questions
> still
> remain:
> 1. My design matrix and contrasts are as below and I run standard
> randomisation for tbss data, does it mean it did 2-tailed test or two
> 1-tailed tests?
Each contrast is a 1-tailed test. Strictly-speaking, if you had no
prior hypothesis about which contrast you were interested in, you
should double the reported p-values for each contrast (to correct for
multiple comparisons across contrasts, i.e., turn it into a two-tailed
test).
> 2. I try to determine appropriate c threshold by looking on tstat
> maps (I
> guess max_tsts and not vox_stats). Do I need my DOF to estimate it
> correct
> for p-value I choose? If so, is my DOF 33? (controls 18, patients 17)?
max_ is the corrected voxelwise p-values (actually 1-p) - they already
take into account the DOF, so you don't need to worry further about
that.
> 3. Function -c is recommended for tbss, but isn`t -C more suitable?
> Is there
> any difference in estimating the c threshold for both -c and -C
> options?
-c (cluster size) is more commonly-used than -C (cluster mass); there
is a little evidence that mass may be more useful. Yes, you'd probably
want to use the same threshold in either case, though it is still an
arbitrary threshold.....
> 4. What are the situations other than analyzing paired data to use -D
> function (demeaning)?
For example, if you haven't put any regressors in your design that
have a non-zero mean.
> 5. I am thinking of the meaning of -c funtion: for very small group,
> the
> cluster forming t-threshold can help to find significant
> differences. Is it
> simply by restricting the area of search? However, high threshold
> like 3 or
> more will again restrict the number of "significant" voxels. Am I
> correct?
Not quite sure what you're asking here.....sadly there is no clear
fixed rule on how to choose this (hence the TFCE option in the latest
version, which you might want to try). See http://www.fmrib.ox.ac.uk/analysis/research/tfce
Cheers.
>
>
> Thank you for all help!
> Aga
>
>
> (design.con)
> /ContrastName1 controls>patients
> /ContrastName2 patients>controls
> /NumWaves 2
> /NumContrasts 2
> /PPheights 1.000000e+00 1.000000e+00
> /RequiredEffect 1.501 1.501
> /Matrix
> 1.000000e+00 -1.000000e+00
> -1.000000e+00 1.000000e+00
>
> design.mat
> /NumWaves 2
> /NumPoints 35
> /PPheights 1.000000e+00 1.000000e+00
>
> /Matrix
> 1.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00
> 0.000000e+00 1.000000e+00
> 0.000000e+00 1.000000e+00
> 0.000000e+00 1.000000e+00
> 0.000000e+00 1.000000e+00
> 0.000000e+00 1.000000e+00
> 0.000000e+00 1.000000e+00
> 0.000000e+00 1.000000e+00
> 0.000000e+00 1.000000e+00
> 0.000000e+00 1.000000e+00
> 0.000000e+00 1.000000e+00
> 0.000000e+00 1.000000e+00
> 0.000000e+00 1.000000e+00
> 0.000000e+00 1.000000e+00
> 0.000000e+00 1.000000e+00
> 0.000000e+00 1.000000e+00
> 0.000000e+00 1.000000e+00
> 0.000000e+00 1.000000e+00
---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director, Oxford University FMRIB Centre
FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------
|