Parkinson's Disease Diagnosis: A Deep Learning Approach Using CNN AlexNet and MR Scans

Parkinson's disease (PD) is a significant public health challenge, characterized by a neurological disorder resulting from a shortage of dopamine-producing neurons in the brain. The cells responsible for dopamine production play a crucial role in regulating, adapting, and ensuring smooth movements. Achieving precise diagnoses for Parkinson's disease (PD) remains challenging owing to its close association with other neurological disorders, especially in cases where motor and non-motor symptoms are absent. The potential utility of computer-aided diagnosis is evident in detecting the early signs of nervous system issues. In this study, a Convolutional Neural Network (CNN) was introduced, demonstrating its efficacy in distinguishing between individuals with Parkinson's disease (PD) and healthy controls (HC). Additionally, explored the feasibility of the automatic diagnosis of PD using Magnetic Resonance (MR) scans. The Parkinson's Progression Markers Initiative (PPMI) offers openly accessible benchmark T2-weighted MR images for both Parkinson's disease and healthy controls. To identify, train, and test the neural network, the CNN AlexNet with deep learning has been utilized. Subsequently, computed neural network's accuracy, sensitivity, precision, and recall to assess its performance and contrast it with alternative methods.