Parkinson’s Disease (PD) patients show symptoms such as hand tremors, poor hand-eye coordination, and body stiffness. While the exact cause remains unknown, it is believed that genetic factors and exposure to environmental toxins. There is no permanent cure, just treatment to slow down disease progression. Due to a lack of definitive diagnostic tests, clinical assessments are crucial but can lead to misdiagnosis. This paper proposes advanced Convolution Neural Network (CNN)-based models leveraging Magnetic Resonance Imaging (MRI) and 6-[18F]Fluoro-L (Levodopa)-DOPA (dihydroxyphenylalanine) (FDOPA) Positron Emission Tomography (PET) scans. In addition, a novel CNN-based hybrid model that incorporates both MRI and FDOPA PET scans is introduced, which achieved an accuracy of 99%, with precision and recall of 0.99 and a minimal false-negative rate, which is vital for effective PD detection. The dataset used for the models comprises T2-weighted MRI scans and FDOPA PET scans, extracted from the Parkinson’s Progression Markers Initiative (PPMI) database.