Measured attenuation correction using the Maximum Likelihood algorithm

Abstract

Quantitative determination of local radioactivity concentration in positron emission tomography (PET) requires a good attenuation correction procedure to reconstruct the emission image. Using a similar Maximum Likelihood (ML) algorithm as for the reconstruction of the emission image, a method is proposed to reconstruct a transmission image, i.e. a map of absorption coefficients. This reconstructed transmission image is then used to calculate the attenuation correction factors needed for the ML reconstruction of the emission image. This approach takes automatically into account the convolution step in the acquisition process (caused by various smoothing factors, e.g. the detector width). This results in appreciable noise suppression without loss of resolution due to filtering, thus making the reconstructed images easier to interpret. A comparison is made with other estimates for the measured attenuation correction using phantom studies.