Evaluation of Regularized Image Reconstruction for Clinical Positron Emission Tomography

Positron emission tomography (PET) combined with computed tomography (CT) is a widely used noninvasive molecular imaging modality with a broad range of clinical applications in oncology, neurology, and cardiology. Producing imperative image quality and accurate quantification are important driving forces behind the technological advances within PET image reconstruction and system development. To ensure clinical quality and to understand how the modern state-of-the-art PET/CT systems and image reconstruction methods compare with older systems and reconstruction methods they need to be evaluated and assessed in a clinical setting. This thesis summarizes six studies assessing the effect of state-of-the-art image reconstruction methods and the introduction of digital PET on image quality and quantitative outcomes of clinical PET scans in oncology, neurology, and cardiology. The overall aim was to evaluate, optimize, and compare quantitative results of regularized image reconstruction with the current standard reconstruction method used in routine clinical practice, ordered subsets expectation maximization (OSEM). The optimal setting of regularized image reconstruction by block-sequential regularized expectation maximization (BSREM) was found to be tracer dependent, and a potential clinical benefit in terms of image quality measures of BSREM over OSEM was found when applied for whole-body 18 F-FDG, 68 Ga-DOTATOC, 18 F-fluorde, 11 C-acetate, and 68 Ga-PSMA-11 PET imaging. Software-aided assessment of neurodegenerative disease evaluated with 18 F-FDG and 18 F-flutemetamol was affected by image reconstruction methods and should be used with caution when employing other image reconstruction methods than those used for acquisition of the normal database. In contrast, changes in reconstruction settings were shown to not implicate myocardial blood flow (MBF) based on 15 O-water PET analyzed using automated software. This shows that diagnostic MBF cutoff values can be consistently used for 15 O-water. Also, large variations in image noise with three different image reconstruction methods did not impact quantitative cerebral blood flow (CBF) in white and gray matter volumes of interest with 15 O-water brain PET to any large extent. BSREM image reconstruction shows a great potential clinical benefit providing improved image quality measures with a subsequent possibility of shortening image acquisition durations and/or lowering amount of radioactivity needed for each exa