On Mon, 3 Apr 2000, Karl Friston wrote:
> Some people think it might be preferable to use non-attenuation
> corrected data with SPECT (to avoid adding extra noise from measured
> attenutation). The only caveat is that the global estimators will
> differ but otherwise it is perfectly OK to proceed without correction.
Is this true? I thought that, unlike PET systems, SPECT systems rarely
have transmission image correction and that correction is usually applied
assuming a uniform attenuation map across the brain (so noise free in that
sense). Also, will it only be a global estimator that changes without
attenuation correction? Surely that assumes the same attenuation gradient
in every subject -- which assumes identical head sizes? [I don't have a
feel for how large such variation may be???]. Once the brain has been
registered without attenuation correction, the attenuation gradient
in the registered image will vary according to the registration
parameters, surely confounding the result in some (small?) way?
Keith
--
Keith Goatman ::: Medical Physics, Southampton General Hospital, UK
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|