Deciphering Parkinson's Disease Progression Using Deep Learning Techniques

Deciphering Parkinson's Disease Progression Using Deep Learning Techniques

연구 분야: Artificial Intelligence

학회: SN Computer Science

Parkinson's disease (PD), a prevalent central nervous system disorder among the elderly, is characterized by the depletion of dopamine-producing brain cells in the substantia nigra pars compacta, leading to motor and non-motor impairments. Traditional diagnostic tools like the Hoehn and Yahr scale and the MDS-UPDRS rely on physical and cognitive assessments for judging the possibility of the disease. These methods are not accurate enough and have not shown promising results over the years. Deep learning solutions, however show promise in early PD detection. Despite significant advances, the following gaps remain: (a) leveraging deep learning for genetic data modelling, (b) utilizing scale-invariant and high-level hand-drawn image features for PD classification. This research addresses these gaps by employing an artificial neural network on the AMP PD dataset, which focuses on protein abundance data, and the New Hand PD Dataset by analyzing hand-drawn images using scale and rotation invariant combined with high-level features. This research demonstrates that deep learning models can outperform traditional benchmark models, offering new avenues for early diagnosis and personalized treatment of PD. By integrating genetic factors and local image features, this work aims to enhance understanding and provide innovative solutions for PD diagnosis at an early stage.