Multilevel Modelling of Parkinson’s Disease Symptom Progression in 123I FP-CIT SPECT

Parkinson’s disease (PD) is the second most prevalent neurodegenerative disorder. It is due to the reduction in the neurotransmitter dopamine by the loss of dopamine neurons. Thanks to new high-precision markers, it is possible to study PD before the development of symptoms and thus to model the progression of the disease. In this study, we apply a non-linear decomposition of 123I FP-CIT SPECT imaging in order to unveil the sources of variability, and relate them to symptomatology and disease progression. Isomap was compared to PCA, and multi-level modelling was used to assess the relation of the resulting composite imaging features and different symptomatology biomarkers and neuropsychological scales. The main source of variability was found to be the intensity at the striata, significantly linked to progression scales such as UPDRS or SCOPA. On the other hand, the modelling found additional sources of variance related to rigidity, tremor or other relevant symptoms.