Joint reconstruction of activity and attenuation in dynamic PET

Joint reconstruction of attenuation and emission in positron emission tomography (PET) using the maximum likelihood activity and attenuation estimation (MLAA) algorithm was proposed in the past. However, cross-talk between the activity and attenuation estimation limits the usefulness of MLAA for PET data without time-of-flight (TOF) information. This work introduces dynamic MLAA (dMLAA), an extension of the MLAA algorithm for dynamic data, to jointly reconstruct the activity distributions and a single attenuation map. The hypothesis is that using information from multiple dynamic emission frames may improve the estimated attenuation map compared to using static PET data. Preliminary results using dMLAA algorithm showed that use of multiple dynamic emission frames slightly improves the reconstructed attenuation map (especially in bones, cavities and lesion area) compared to using a single emission frame. However, without TOF, the reconstructed map still suffers from ill-posedness of the problem despite the additional dynamic information. The reconstruction may be improved for tracers that present a higher inter- and intra- dynamic frame contrast and edge variability.