Should scatter be corrected in both transmission and emission data for accurate quantitation in cardiac SPET?

Ideally, reliable quantitation in single-photon emission tomography (SPET) requires both emission and transmission data to be scatter free. Although s catter in emission data has been extensively studied, it is not well known how scatter in transmission data affects relative and absolute quantitation in reconstructed images. We studied SPET quantitative accuracy for differe nt amounts of scatter in emission and transmission data using a Utah phanto m and a cardiac Data Spectrum phantom including different attenuating media . Acquisitions over 180 degrees were considered and three projection sets w ere derived: 20% images and Jaszczak and triple-energy-window scatter-corre cted projections. Transmission data were acquired using gadolinium-153 line sources in a 90-110 keV window using a narrow or wide scanning window. The transmission scans were performed either simultaneously with the emission acquisition or 24 h later. Transmission maps were reconstructed using filte red backprojection and mu values were linearly scaled from 100 to 140 keV. Attenuation-corrected images were reconstructed using a conjugate gradient minimal residual algorithm. The mu value underestimation varied between 4% with a narrow transmission window in soft tissue and 22% with a wide window in a material simulating bone. Scatter in the emission and transmission da ta had little effect on the uniformity of activity distribution in the left ventricle wall and in a uniformly hot compartment of the Utah phantom. Cor recting the transmission data for scatter had no impact on contrast between a hot and a cold region or on signal-to-noise ratio (SNR) in regions with uniform activity distribution, while correcting the emission data for scatt er improved contrast and reduced SNR. For absolute quantitation, the most a ccurate results (bias <4% in both phantoms) were obtained when reducing sca tter in both emission and transmission data. In conclusion, trying to obtai n the same amount of scatter in emission and transmission data, in addition to being impractical because of the difficulty in knowing the precise scat ter components, did not yield such accurate absolute activity quantitation as when emission and transmission scatter were reduced.