Registration between respiratory-gated PET/CT and high-resolution CT with XCAT simulations: Evaluation and optimization for subsequent PVC

Registration of a high-resolution CT (HRCT) to a low-resolution PET has been shown to enable enhanced PET image quality. The CT can be used as prior information for partial volume correction (PVC) during reconstruction, provided that it is accurately registered to the PET dataset of interest. By means of simulations using an XCAT phantom, this work analyzes the impact of different attenuation correction protocols on HRCT to PET registration. The focus of this work is on cardiac images with only respiratory motion taken into account, with the aim of optimally aligning the left-ventricular region in PET and HRCT. Shallow breathing was simulated and so far only noise-less data were evaluated. Also, a comparison to registration results from a deep breathing phantom has been performed. As expected, results speak in favour of using an ideal, perfectly matching attenuation correction map (ACM). However, when using an average or a single respiratory phase ACM accurate alignment can still be achieved. Registration of non-attenuation-corrected images requires further investigation, as it seems to be strongly dependent on the initial alignment of PET and CT.