Hi,
removing part of the time series is always a bad idea for a variety
of reasons (e.g. if there's a low frequency drift in the data the
cutting will effectively create a step which induces additional error
variance and lowers sensitivity. Also, the estimation of
autocorrelation and the HRF convolution will be affected etc...)
The best way of dealing with this is to use dummy 'nuisance'
regressors in your design, i.e. if you want to cut out 3 volumes from
your series then simply add 3 more EVs to your design, each of which
is 0 almost everywhere and has a single 1 for the volume you want to
remove. You can use the 'Custom (1 entry per volume)' basic shape in
the FEAT GLM setup GUI.
cheers
Christian
On 9 Apr 2007, at 23:18, Anna Engels wrote:
> Hi,
> We have a few subjects with several volumes of artifact during the
> middle
> of the timeseries. Would it be ok to use avwroi and avwmerge (before
> preprocessing) to literally delete these volumes from the
> timeseries? We
> were wondering if this would cause problems because the HDR would
> then not
> be modeled correctly, or for other reasons that we're not aware
> of. Will
> this method of cutting the timeseries into pieces and putting it back
> together again create any problems for statistical analysis?
> Thanks,
> ~Anna
____
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
|