Quantitative SPECT Brain Imaging: Effects of Attenuation and Detector Response

Two physical factors that substantially degrade quantitative accuracy in SPECT imaging of the brain are attenuation and detector response. In addition to the physical factors, random noise in the reconstructed image can greatly affect the quantitative measurement. The purpose of this work was to implement two reconstruction methods that compensate for attenuation and detector response, a 3D maximum likelihood-EM method (ML) and a filtered backprojection method (FB) with Metz filter and Chang attenuation compensation, and compare the methods in terms of quantitative accuracy and image noise. The methods were tested on simulated data of the 3D Hoffman brain phantom. The simulation imcorporated attenuation and distance-dependent detector response. Bias and standard deviation of reconstructed voxel intensities were measured in the gray and white matter regions. The results with ML showed that in both the gray and white matter regions as the number of iterations increased, bias decreased and standard deviation increased. Similar results were observed with FB as the Metz filter power increased. In both regions, ML had smaller standard deviation than FB for a given bias. Reconstruction times for the ML method have been greatly reduced through efficient coding, limited source support, and by computing attenuation factors only along rays perpendicular to the detector.