On Wed, 16 Jan 2008 12:31:57 -0800, Linda Seltzer <[log in to unmask]>
wrote:
>Does anyone written a paper or tutorial applying the well-known concepts
of
>sampling, aliasing, Nyquist theorem, etc. to data analysis in SPM?
>Specifically, TR and sampling rate; spectrum of voxel data vs time, the
>same with noise; how the concept of jitter relates to this; upsampling to
>convolve the delta fns with hemodynamic response (for the regressors) and
>then downsampling the signal back, when each step is done. I would like
to
>see a description from a signal processing engineering viewpoint of how
all
>of the sampling rates and bandwidths are handled. If there is nothing
>formal, but someone has made some notes deriving all of this, I would like
>to read them and learn more. I would like to have an overview of all of
>the sampling rates and spectrums involved.
>Linda Seltzer
>[log in to unmask]
>[log in to unmask]
>=========================================================================
IMHO a useful way to look at this is to not restrict consideration to SPM,
but think about fMRI analysis in general (restricted to the case of
univariate statistics, as opposed to multivariate methods). Then look for
articles in the published literature via Google or PubMed. There's quite
a bit out there.
Just to take an interesting an relevant example, the problems of aliased
noise from cardiac and respiratory cycles is discussed quite frequently.
|