Hello,
thank you for a fast reply. As I am not that familiar with f-test -
do I understand correctly that I should add EV4 1-1-1-1-1-1 and keep
EV4 zero for all contrasts ?
Thank you for fast reply
George
2010/7/20 Stephen Smith <[log in to unmask]>:
> Hi - you should either demean your covariates and then use the -D option, or
> add in an all-1s covariate to model the data mean (and ignore this in the
> contrasts).
> For now it's ok to use uncorrected p-vaules, as with just 6 timepoints from
> 1 subject you will have very low power to find properly significant effects.
> Cheers.
>
>
> On 19 Jul 2010, at 14:19, Jiøí Keller, M.D. wrote:
>
> Dear FSLers,
> I am trying to do a longitudinal TBSS study (so far I have only a
> pilot data of 6 measurements of one subject). I would like to
> correlate the FA changes with three clinical scales. The question is
> "where FA can be explained by one of the scales or by the combination
> of all three". I tried the design as pasted at the end of this mail.
> Is my approach correct ? I ran randomise on that and only received
> some clusters on f1 contrast, but only "uncorrected". I backprojected
> those clusters to the original space after segmening them (taking only
> clusters of three or more voxels) and took only voxels, which were
> correctly back-projected and the results seem reasonable to me. I am
> not sure whether TBSS could be used this way (being MD and not a
> full-grown statistitian), what are the opinions ? Or is TFCE a kind-of
> dogma ?
> Thank you for replies
> George
>
> Design:
> --MAT--
>
> /NumWaves 3
> /NumPoints 6
> /PPheights 2.600000e+01 5.700000e+01 1.800000e+01
>
> /Matrix
> 0.000000e+00 4.000000e+00 0.000000e+00
> 1.000000e+01 2.500000e+01 2.000000e+00
> 2.400000e+01 3.500000e+01 1.000000e+01
> 2.600000e+01 4.500000e+01 1.700000e+01
> 1.600000e+01 5.000000e+01 1.800000e+01
> 2.000000e+01 5.700000e+01 1.700000e+01
> --CON--
> /ContrastName1 RMean
> /ContrastName2 NMean
> /ContrastName3 Vmean
> /NumWaves 3
> /NumContrasts 3
> /PPheights 1.505706e+01 2.234061e+01 8.372717e+00
> /RequiredEffect 199.531 211.371 207.134
>
> /Matrix
> 1.000000e+00 0.000000e+00 0.000000e+00
> 0.000000e+00 1.000000e+00 0.000000e+00
> 0.000000e+00 0.000000e+00 1.000000e+00
> --FTS--
> /NumWaves 3
> /NumContrasts 2
>
> /Matrix
> 1 1 1
> 0 1 1
>
>
>
> ---------------------------------------------------------------------------
> 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
> ---------------------------------------------------------------------------
>
>
>
>
|