Dear Ben,
> Hi, I am a research assistant at Stanford University, Neuroimaging Lab. We are working on quantitizing signal dropout due to susceptibility
> artifact, and I had a short question you might be able to answer
> quickly. When SPM performs single-subject stats it assigns some voxels
> NaN because they lack enough variance to be considered having good
> signal. We have made plots of spmT_???.img that show Tscores falling to
> 0 (or NaN) below a definite intensity value in the smoothed input
> images. Do you know how this cut-point is set? Thanks very much, for
> any information you may have.
SPM99 applies some masking prior to the parameter estimation at each
voxel. The main reason for this is that some voxels (e.g. outside the
brain) might be uninteresting to look at. To save computation time and
hard disk space, SPM99 determines based on some criteria which voxels to
remove from the analysis. Most interestingly for you, there is one
criterion that removes a voxel time series from the analysis, if one
value of this time-series falls below a given threshold. This threshold
is for fMRI data analysis chosen automatically by SPM99.
Have a look at
http://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind0006&L=spm&P=R13693
for a description of the masks applied during the statistical analysis.
This text also contains some advice how to change the intensity
threshold or to switch it off completely.
Stefan
--
Stefan Kiebel
Functional Imaging Laboratory
Wellcome Dept. of Cognitive Neurology
12 Queen Square
WC1N 3BG London, UK
Tel.: +44-(0)20-7833-7478
FAX : -7813-1420
email: [log in to unmask]
|