Righto - in the next release there's a little program that takes in a
design matrix and removes that from the data, that's what you want.
You should be able to put the attached program into $FSLDIR/src/
randomise, edit the Makefile and compile.
changes to Makefile:
XFILES = randomise fdr unconfound
unconfound: unconfound.o
${CXX} ${CXXFLAGS} ${LDFLAGS} -o $@ unconfound.o ${LIBS}
cheers
On 19 May 2007, at 21:05, Rutger Goekoop wrote:
> Hi Steve,
>
> Thanks for your quick reply. I'd like to use MELODIC for further
> studies on
> these data, but since all timeseries contain the same noise
> artefact, I'd
> first like to denoise the data and then use MELODIC on the clean data.
>
> Since MELODIC itself does not seem to be able to find the relevant
> components that completely describe the noise artefact (components
> that I
> could then use to denoise the timeseries using the MELODIC filter
> option), I
> wonder whether there is any other way of filtering out this
> particular noise
> artefact (best modeled by the three regressor below). If a model based
> approach is the only option, I'd like to know how I can use the
> model-based
> output to remove the noise artefact from my timeseries, prior to
> entering
> these into a MELODIC session.
>
> Cheers, Rutger.
>
> -----Original Message-----
> From: FSL - FMRIB's Software Library [mailto:[log in to unmask]]On
> Behalf Of Steve Smith
> Sent: Saturday, May 19, 2007 8:42 AM
> To: [log in to unmask]
> Subject: Re: [FSL] Denoising 0.5 Hz scanner artefact
>
>
> Hi - it's easy to add confound EVs (covariates) in FEAT which would
> be equivalent to what you're doing in SPM - does this not do what you
> want?
> Cheers.
>
>
> On 18 May 2007, at 23:27, Rutger Goekoop wrote:
>
>> Dear all,
>>
>> For some yet unexplained reason, a 0.5 Hz scanner artefact appears
>> in our 800
>> volume PRESTO SENSE timeseries (TR 0.609). This noise sums nearly
>> linearly on
>> top of existing BOLD signal intensity. Analysis shows that it can be
>> eliminated during postprocessing using user-specified regressors in
>> SPM5
>> (with delta functions every 4th scan) as follows:
>>
>> Regressor1:
>> 1 0 0 0 1 0 0 0 1 0 0 0 1 0 0 0 ... etc: complete 800 volumes.
>> Regressor2:
>> 0 1 0 0 0 1 0 0 0 1 0 0 0 1 0 0 ... etc: complete 800 volumes.
>> Regressor3:
>> 0 0 1 0 0 0 1 0 0 0 1 0 0 0 1 0 ... etc: complete 800 volumes.
>>
>> This yields 3 PE images for these 3 regressors with almost 100% noise
>> elimination with good residual SD.
>>
>> Since I want to use MELODIC for further study of these data (and
>> since I'm
>> forever hooked to FSL), I'd like to do something similar in FSL. I
>> tried
>> MELODIC denoising, however MELODIC does not find the relevant
>> components.
>> Instead, nearly all components are contaminated with the 0.5 Hz
>> artefact
>> (possibly because of its regular nature?). Thus, I think I may
>> first need to
>> do a model-based analysis in all individuals (as in SPM5, see
>> above) using
>> custom EVs with delta functions, and then somehow filter out the
>> beta images
>> from the filtered_func_data (using avwmaths?). Or is there perhaps
>> a more
>> sound way of dealing with this problem?
>>
>> Any help would be greatly appreciated,
>>
>> Cheers,
>>
>> Rutger Goekoop.
>
>
> ----------------------------------------------------------------------
> --
> ---
> 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
> ----------------------------------------------------------------------
> --
> ---
>
>
> --
> Mijn Postvak In wordt beschermd door SPAMfighter.
> 132 spam-mails zijn er tot op heden geblokkeerd.
> Download de gratis SPAMfighter via deze link: http://
> www.spamfighter.com/lnl
>
>
> --
> Mijn Postvak In wordt beschermd door SPAMfighter.
> 132 spam-mails zijn er tot op heden geblokkeerd.
> Download de gratis SPAMfighter via deze link: http://
> www.spamfighter.com/lnl
------------------------------------------------------------------------
---
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
------------------------------------------------------------------------
---
|