Deep Learning-Based Explainable Artificial Intelligence for Bradykinesia and Tremor Symptoms of Parkinson’s Disease

Parkinson's disease (PD) is the most common neurodegenerative movement disorder attributed to the degeneration of dopaminergic neurons responsible for motor control, leading to symptoms such as tremor, slowness, and rigidity. PD is often confused with essential tremor (ET), another neurological disorder that causes tremors across limbs, due to overlapping symptom characteristics. The dopamine transporter scan effectively distinguishes PD from ET but reports normal baseline imaging findings in some old-onset PD cases. Therefore, this thesis proposes deep learning-based classification models and applies an explainable artificial intelligence framework to identify the unique features of PD bradykinesia, thereby assisting neurologists in differential diagnoses.