Selection of collimator for rCBF studies and evaluation of triple-headed SPET using noise-resolution plots

We investigated the effect of collimator selection on image quality in regional cerebral blood flow (rCBF) studies of the brain performed with 99Tc(m)-HMPAO. A triple-headed SPET system (GE/CGR Neurocam) was used, together with three sets of parallel-hole collimators - general-purpose (GP), high-resolution (HR) and ultra-high-resolution (UHR). Two image quality parameters were used to describe the image quality, namely, noise and resolution. Noise was measured in experimental and Monte-Carlo simulated SPET studies of a cylinder phantom of uniform activity as the pixel root mean square error (RMS) and as the coefficient of variation (CV) of quantitative rCBF values. Resolution was measured as full-width at half-maximum in experimental SPET studies of a line-source. Plots of noise versus resolution for the different collimators were obtained by varying the cut-off frequency of the Hanning filter applied in the reconstruction of transaxial slices. From these noise-resolution plots, we were able to determine which collimator gave the best resolution for a specific noise level. A lowest reasonable noise level may be established by comparison with the inter-observer CV of the quantification method.