Quantitative SPECT brain imaging: effects of attenuation and detector responseat

Two factors which degrade quantitative accuracy in single photon emission computed tomography (SPECT) imaging of the brain are attenuation and detector response. Two reconstruction methods that compensate for attenuation and detector response (a filtered backprojection method with Metz filter and Chang attenuation compensation and a 3-D maximum-likelihood-EM method) are compared in terms of quantitative accuracy. The methods were tested on simulated data on the 3-D Hoffman brain phantom. The simulation incorporated attenuation and distance-dependent detector response. The root-mean-square (RMS) error was measured in the reconstructed images in the gray matter, white matter, and ventricle regions. The results in the white matter showed that the 3-D maximum-likelihood-expectation-maximization (ML-EM) method over a range of iteration stopping points had a smaller error than the filtered backprojection method over a similar range of power factors. In the gray matter the reverse was found, and this was attributed to an edge artifact in the 3-D ML-EM images. Reconstruction times for the 3-D ML-EM method were greatly reduced through efficient coding, limited source support, and computing attenuation factors only along rays perpendicular to the detector.<
>