To Smooth or Not to Smooth? Bias and Efficiency in fMRI Time-Series
Analysis, K.J. Friston, O. Josephs, E. Zarahn, A.P. Holmes, S. Rouquette,
J.B. Poline. Neuroimage 12: 196-208 (2000). First sentence in abstract:
"This paper concerns temporal filtering in fMRI time series analysis." The
article talks about filtering the design matrix in a way that doesn't cause
statistical bias. Re spatial filtering, the matlab image processing
toolbox should have such routines.
Linda
> I wonder if this might be one of the papers with the same title about
> spatial smoothing?
> Max
>
> Wang, J., Wang, Z., Aguirre, G. K., & Detre, J. A. (2005). To smooth or
> not
> to smooth? ROC analysis of perfusion fMRI data. Magnetic Resonance
> Imaging,
> 23(1), 75-81.
>
> On Nov 6, 2007 2:49 PM, Laura Mancini < [log in to unmask]>
> wrote:
>
> > With "to smooth or not to smooth" you mean the paper from Friston?
> > Because I think that one is on temporal smoothing, not spatial
> > smoothing.
> >
> > Best,
> > Laura
> >
> >
> >
> > -----Original Message-----
> > From: SPM (Statistical Parametric Mapping) [mailto: [log in to unmask]]
> > On Behalf Of cyril pernet
> > Sent: 06 November 2007 14:11
> > To: [log in to unmask]
> > Subject: Re: [SPM] Is it appropriate to smooth like this?
> >
> >
> > Hi Shaquanda
> > > Hi Cyril,
> > >
> > > Thanks for the wisdom. That is what I thought, but it is nice to know
> > > that is the case. If I may, I have one more questions: Do you know of
> > any references that address the issue of spacial smoothing for
> irregular
> > volumes (not necessarily an "imaging" reference). I have tried to find
> > this, but have had difficulty.
> > >
> > the most popular paper is 'to smooth or not to smooth?' neuroimage
> 2002;
> >
> > from there in the ref or using the link in pubmed you'll get other
> > references
> >
> > plus the good chapter on
> >
> http://documents.wolfram.com/applications/eda/SmoothingDataFillingMissin
> > gDataAndNonparametricFitting.html
> >
> > best
> > cyril
> >
> >
> > > Thank you once again very much for your time and thoughts. Regards,
> > > Shaqunada
> > >
> > > ----- Original Message ----
> > > From: cyril pernet <[log in to unmask]>
> > > To: Shaquanda Jones < [log in to unmask]>; SPM
> > > <[log in to unmask]>
> > > Sent: Monday, November 5, 2007 6:20:58 AM
> > > Subject: Re: [SPM] Is it appropriate to smooth like this?
> > >
> > >
> > > Hi Shaquanda
> > >
> > >> I have heard that when you have a highly non-isotropic scan volume
> > >>
> > > that it is appropriate to use a non-isotropic smoothing kernel. Is
> > > this true?
> > >
> > >>
> > >>
> > > well I don't know if one can say it is true or not, but spatial
> > > smoothing will render your data data more normally distributed which
> > is
> > >
> > > always good when you want to do some stats .. anyway I have no
> > > mathematical/statistical explanations here but the usual thing to do
> > is
> > >
> > > to smooth 1 to 1.5 times the voxel size for individual data or about
> 3
> > > times for group analyses, i.e. if you normalize at 2x2x2 for a group
> > > analysis, then a 6x6x6 FWHM (or higher) will do - now say you want to
> > > do
> > > some individual analyses then for e.g. a non-isotropic 3x3x5 voxel
> > > size,
> > > a 4.5x4.5x7.5 FWHM should be ok ..
> > >
> > > hope this helps
> > > cyril
> > >
> > >
> > >
> > >
> > >
> > > __________________________________________________
> > > Do You Yahoo!?
> > > Tired of spam? Yahoo! Mail has the best spam protection around
> > > http://mail.yahoo.com
> > >
> > >
> > >
> >
> >
> > --
> > Dr Cyril Pernet,
> > fMRI Lead Researcher SINAPSE
> > SFC Brain Imaging Research Center
> > Department of Clinical Neurosciences
> > University of Edinburgh
> > Western General Hospital
> > Crewe Road
> > Edinburgh
> > EH4 2XU
> > Scotland, UK
> >
> > [log in to unmask]
> > tel: +44(0)1315373661
> > http://www.dcn.ed.ac.uk/bic/
> > http://www.dcn.ed.ac.uk/bic/sinapse/sinapse.asp
> >
>
|