Hi Jaroslav,
If you use the --Oorig option from the command line, melodic will
also output the non Z-transformed images - if you want to turn these
into %BOLD make sure that you also switch off the variance-
normalisation (use --vn at the command line). The next release
version (this summer) will allow users to request outputs in terms of
%BOLD signal change.
Hope this helps
Christian
On 25 Jun 2007, at 17:19, Jaroslav Hlinka wrote:
> Hi all,
>
>
> Is there any way to extract out something like the raw percentage BOLD
> signal change (say as deviation from average signal) corresponding
> to a
> component?
>
>
> In my understanding, timeseries stored in say t1.txt is normalised to
> variance =1 - so gives no information for such comparison.
>
> MELODIC_IC.img maps are already Z-scores (divided by error estimate
> given
> as the data variance UNEXPLAINED by the extracted components).
> Even if I visually detect say motor area network component in two
> datasets
> and extract the MELODIC z-stat maps, direct comparison is
> ambiguous, since
> the difference in MELODIC z-stats can be attributed to different
> levels of
> fluctuation of motor network as well as different levels of
> unexplained
> noise variance (more unexplained movement, different condition of
> scanner,...), which raw ICAs are divided by.
>
>
> (The only way I have on mind is "filtering out" the component from
> initial
> data, substracting the filtered from initial data, and dividing
> this by
> initial data - but am not sure about how exactly the command line
> filtering option works)
>
> Thanks a lot for any suggestions,
>
> Jaroslav Hlinka
____
Christian F. Beckmann
University Research Lecturer
Oxford University Centre for Functional MRI of the Brain (FMRIB)
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK.
[log in to unmask] http://www.fmrib.ox.ac.uk/~beckmann
tel: +44 1865 222551 fax: +44 1865 222717
|