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
|