Hi
On 27 Nov 2003, at 15:34, Christian Kaufmann wrote:
> Dear all,
>
> from a pica with melodic I got the result files named below. I would
> like to denoise my data as described in the FSL course slides (btw
> thanks!) with melodic. I'm confused about the file
> "filtered_func_data.{img,hdr}" and the fact that melodic denoise gives
> me back 32 bit data although originally they are 16 bit.
> First (to be sure about this), are the "filtered_func_data.{img,hdr}"
> simply the data after prestats?
>
Yes, it's the functional data after motion correction and
spatial/temporal filtering (and any other pre-proc step that you've
specified)
> Second, why 32 bit (or did I miss something in the options ...)?
Melodic internally deals with doubles and writes 32 bit output (since
the melodic maps are Z scores)
> And third, is the denoising option based on the assumption that the
> components derived from pica are additive and are therefore subtracted
> from the original data?
>
Yes, it's a linear additive process and the filter step in melodic is
identical (modulo degree-of freedom issues) to using the melodic time
courses as regressors of no interest in FEAT (when orthogonalising the
rest of the design w.r.t melodic time courses)
hope this helps
christian
> Thank you!
>
> Christian
>
>
>
> ===================
> output from melodic ica:
> eigenvalues_adjusted
> eigenvalues_percent
> example_func.hdr
> example_func.img
> filtered_func_data.hdr
> filtered_func_data.img
> mask.hdr
> mask.img
> melodic_FTmix
> melodic_IC.hdr
> melodic_IC.img
> melodic.log
> melodic_mix
> PPCA
> report
> report.log
> smoothest
> stats
> ==================
>
> Humboldt-Universität zu Berlin
> Department of Psychology
> Clinical Psychology and Behavioural Neuroscience Lab
> Head: Prof. Dr. Kathmann
> Wolfgang Köhler-Haus
> Rudower Chaussee 18
> 12489 Berlin
> Germany
> voice +43 30 2093 4940
> fax +43 30 2093 4859
> e-mail [log in to unmask]
>
--
Christian F. Beckmann
Oxford University Centre for Functional
Magnetic Resonance Imaging of the Brain,
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
Email: [log in to unmask] - http://www.fmrib.ox.ac.uk/~beckmann/
Phone: +44(0)1865 222782 Fax: +44(0)1865 222717
