Abstract: In The Parkinson's disease (PD) is a progressive neurodegenerative disorder that affects movement, muscle control, and balance. Early diagnosis of PD is essential for timely intervention and management of the disease. This project proposes a novel PD detection system that leverages spiral drawings and convolutional neural networks (CNNs) for accurate and efficient diagnosis. The proposed system consists of two main components: data collection and analysis. Patients are instructed to draw spirals using a digital pen or touchscreen device, capturing the subtle motor impairments characteristic of PD. These drawings are then pre processed and fed into a CNN model for feature extraction and classification. The CNN model is trained on a dataset of spiral drawings from both PD patients and healthy individuals. Transfer learning techniques are employed to fine-tune a pre-trained CNN architecture, enhancing the model's ability to detect subtle patterns indicative of PD. The developed system is implemented using Python and Flask, providing a user-friendly web interface for data collection and analysis. The system aims to improve the accuracy and efficiency of PD detection compared to existing methods, offering a non-invasive and cost effective solution for early diagnosis and monitoring of PD patients.
Keywords: Parkinson, CNN, Deep learning, Spiral Drawings.
|
DOI:
10.17148/IJARCCE.2024.13807
