Detection of Parkinson's Disease Using Spiral Models

Parkinson’s Disease (PD) is a steadily worsening neurological condition that affects millions of individuals around the world. Early diagnosis remains a major challenge due to subtle initial symptoms and limited access to trained neurologists. This paper proposes an automated and explainable deep-learning approach designed to identify PD at an early stage using spiral drawing tests. Using transfer learning with VGG19 and explainable AI methods such as LIME, the proposed model identifies tremor- induced distortions in drawings and highlights regions contributing to classification. The model demonstrates high accuracy and interpretability, positioning it as a reliable tool for assisting clinical diagnosis.