Quantitative PET/MRI Evaluation and Application in Dementia

A recently introduced integrated scanner combining simultaneous positron emission tomography (PET) and magnetic resonance imaging (MRI) (PET/MRI) acquisition presents a unique set of opportunities for neuroimaging research and dementia in particular [1]. Among these, the intrinsic co-registration of the images has the potential to reduce errors in multi-modality image registration, which could lead to better quantification of longitudinal volumetric changes in neuroanatomical features that are important for assessing disease progression. PET tracers could provide functional information of brain biochemical processes by radio-labeling specific molecular with high sensitivity and specificity, while MRI could provide superior structural information. Applying MRI anatomical priors to reduce the PET partial volume effect is straightforward, and could further be used to improve spatial resolution of PET images [2]. The hardware challenge in integrating PET and MRI system is the interaction of the PET electronics and MRI system [3]. The special deign of the PET/MRI-PET detector (so called "MR transparent") architecture includes integrated cooling features to assure optimal PET performance, as well as specialized shielding to virtually eliminate magnetic field interference in the PET data processing chain [4]. On the MRI side, larger bore size with cylindrically optimized homogeneity volume has been used for better image quality [4]. Another challenge of the PET/MRI scanner is the derivation of attenuation correction (AC) from MRI for PET images reconstruction [2]. Three commonly-used MR-based AC (MRAC) methods had been proposed: template-based MRAC, Dixon's method [5] for fat/water separation which assigns AC values based on the segmentation of fat, water and relative bone tissues, as well as dual-echo or multi-echo ultra-short TE (UTE) MRI sequence [6] for delineation of bone, air and software tissue.