Parkinson's Disease Detection Using Deep Learning Method & Explainable AI

Parkinson's disease (PD) is a neurological disorder characterized by muscle weakness and uncontrollable limb shaking due to the loss of dopamine-producing neurons in the brain. Although there is no cure, early recognition significantly improves patient outcomes. More than 10 million people worldwide are affected by Parkinson's disease, with men being 1.5 times more likely to be affected. This research leverages deep learning methods and explainable AI to enhance diagnostic procedures, facilitating the early identification and treatment of Parkinson's disease. The objective of our project is to develop a web-based system to detect Parkinson's disease using deep learning and explainable AI for highly accurate and interpretable results. We utilized the VGG19 model to train on our hand-drawn spiral dataset. Additionally, we incorporated MRI data, employing a pre-trained VIT model to further improve the diagnostic accuracy. For the implementation of our proposed model, we used Jupyter Notebook and Google Colab, and for our website, we used Streamlit. We achieved an overall accuracy of 97% and a 10% loss for the hand-drawn spiral dataset & 99% accuracy and 1% loss for the MRI dataset on our validation set after model training. To ensure the interpretability of our model's predictions, we integrated explainable AI techniques, specifically using LIME (local interpretable model-agnostic explanations). Users can upload their hand-drawn spiral or MRI images to our website to receive precise diagnostic results. This tool will assist both patients and doctors in diagnosing the disease and guiding treatment plans effectively.