AbstractWe investigated the glucose metabolism in an adeno-associated viral vector based alpha-synuclein rat model for Parkinson’s disease (PD) using longitudinal ¹⁸F-FDG PET imaging, which resulted in an improved characterization of this animal model. We generated a PD specific pattern (PDSP) based on a multivariate classification approach to differentiate between a PD and control group at a late disease stage, where the neurodegeneration is considered nearly complete. In particular, we applied a principal component analysis prior to classification by a support vector machine (SVM). Moreover, by using a SVM for regression to predict corresponding motor scores, a PD motor pattern (PDMP) was derived as well. The PDSP mainly corresponds to the PDMP and overlaps to a large extent with the human pattern. We were able to quantify disease expression at previous time points by projecting onto the PDSP and PDMP. While a univariate analysis indicated metabolic changes which did not persist through time, both PDSP and PDMP were able to differentiate significantly (p-value < 0.05) between the PD and control group at week 4, 6 and 9 post injection, while no significant differences were obtained at baseline and at week 3, which is in accordance with the animal model.