Friedemann,
> The GUI in FEAT shows "Mixed effects: Siimple OLS". What does it mean?
> Is it the "classical" OLS, which is used by other fMRI-Analyses software
> packages (SPM, LIPSIA ...)? Or is it a simplified OLS for a short
> estimation of fMRI group results and is therefore not comparable with
> the OLS output of other analyses tools? (Whereas for a full "serious"
> higher level analysis it is recommended to use FLAME (stage 2).)
I am not completely clear what you mean by the "simplified OLS for a short
estimation..." approach. What I can say is that the Simple OLS approach in
Feat is the summary statistic approach (each level in the heirarchy of
subjects and sessions is analysed separately and summary statistics are
passed between levels) as outlined by Holmes and Friston:
Holmes, A. and Friston, K. (1998).
Generalisability, random effects & population inference.
In Fourth Int. Conf. on Functional Mapping of the Human Brain, NeuroImage,
volume 7, page S754.
In this simple OLS approach the summary stats passed between levels are
just the parameter estimates or contrasts of parameter estimates. The
inference at the group level is then unbiased as long as balanced designs
are used.
In contrast the full FLAME approach (stage 1 or 2) looks to use the
variance information from lower levels as well. This improves sensitivity
and allows for non-balanced designs:
Woolrich, M.*, Behrens, T.*, Beckmann, C., Jenkinson, M., and Smith, S. (2004).
Multi-Level Linear Modelling for FMRI Group Analysis Using Bayesian
Inference
Neuroimage, 21(4):1732-1747
As far as I know the simple OLS approach is the same as the summary
statistic approaches used in SPM.
--
Cheers, Mark.
Mark Woolrich.
Oxford University Centre for Functional MRI of the Brain (FMRIB),
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK.
Tel: (+44)1865-222782 Homepage: http://www.fmrib.ox.ac.uk/~woolrich
|