Parkinson’s Disease Detection Using Spiral Images and Voice Data Set

Abstract

Parkinson’s Disease (PD) is a degenerative neurological disorder that significantly affects motor functions, vocal clarity, and overall coordination. The accurate and early detection of PD is critical for managing symptoms and improving patient outcomes. This base paper presents a dual-modal diagnostic approach that utilizes both voice feature analysis and spiral drawing interpretation to detect Parkinson’s Disease. Voice data is processed to extract features such as jitter, shimmer, and harmonic-to-noise ratio, which are then analyzed using a Random Forest classifier. Parallelly, spiral drawings — commonly used in clinical motor assessments — are classified using a Convolutional Neural Network (CNN) to detect tremor patterns. A web-based application is developed using Streamlit, allowing users to upload data and receive real-time predictions. The integration of two distinct data modalities enhances diagnostic accuracy and provides a non-invasive, cost-effective alternative to traditional methods. Experimental results validate the effectiveness of both models independently, demonstrating that a combined system can provide a more reliable screening tool. This project highlights the potential of AI-powered healthcare tools in supporting early-stage Parkinson’s diagnosis, particularly in remote or resource-limited settings.