Parkinson’s Disease Detection via MRI:A Machine Learning Approach with XAI Integration

Parkinson's disease (PD) is a progressive neurological ailment that affects millions of individuals globally. Improving patient outcomes requires early identification and treatment.Despite advances in Machine Learning and Deep Learning forParkinson's disease detection, the explainability of these models remains a significant challenge. The development of Deep Learning and Machine learning approaches has greatly aided PD identification. Still, these models' explainability—which frequently functions as "black boxes," making them challenging to interpret in clinical settings—remains a significant obstacle. This work focuses on using MRI data to identify Parkinson's disease using three Machine Learning algorithms: Random Forest, K-Nearest Neighbours, and Naive Bayes. XAI methodologies, namely LIME (Local to overcome the drawbacks of model interpretability.LIME improves the transparency and trustworthiness of these algorithms by offering insights into the primary variables influencing the model's predictions. By combining XAI with ML models, this approach ensures high-performance detection and improves physicians' understanding of the decision-making process. The findings show that reliable ML techniques combined with explainable frameworks like as LIME can enhance the accuracy and efficacy of Parkinson's disease detection models in medical applications. By bridging the gap between cutting-edge technology and medical practice, this method makes it simpler for medical personnel to comprehend and have trust in ML-based tools for early PD diagnosis.