In order to assess the relative effect of scatter in comparison to att
enuation and collimator blur, we Imaged concentric cylinder, cold-lesi
on, and thorax-like phantoms with the emission-transmission computed t
omography (ETCT) system, a novel device capable of both x-ray CT and S
PECT. For these experiments, the ETCT system was outfitted with a sing
le channel of low-noise electronics which permitted acquisition of SPE
CT data with 3 keV FWHM energy resolution for good scatter rejection.
Reference images were reconstructed with object-specific attenuation c
orrection and accurate collimator response compensation. These images
were then compared to Images which were reconstructed without attenuat
ion correction, without collimator response, or with data that had bee
n blurred to simulate 13 keV FWHM energy resolution. The results show
that although scatter represents a measurable error, its effect is sma
ll compared to the effect of attenuation and collimator blur. As part
of this study, we have also compared scatter rejection to the well-kno
wn dual energy window scatter correction method and to a technique whi
ch incorporates an estimate of the scatter distribution directly in th
e back-projector of the ML-EM algorithm. It is shown that scatter corr
ection may be preferable to scatter rejection due to noise considerati
ons.