Attenuation correction for the NIH ATLAS small animal PET scanner

We evaluated two methods of attenuation correction for the NIH ATLAS small animal PET scanner: 1) a CT-based method that derives 511 keV attenuation coefficients (/spl mu/) by extrapolation from spatially registered CT images; and 2) an analytic method based on the body outline of emission images and an empirical /spl mu/. A specially fabricated attenuation calibration phantom with cylindrical inserts that mimic different body tissues was used to derive the relationship to convert CT values to /spl mu/ for PET. The methods were applied to three test data sets: 1) a uniform cylinder phantom, 2) the attenuation calibration phantom, and 3) a mouse injected with [/sup 18/F] FDG. The CT-based attenuation correction factors were larger in nonuniform regions of the imaging subject, e.g. mouse head, than the analytic method. The two methods had similar correction factors for regions with uniform density and detectable emission source distributions.