A Deep Learning Web Application for Parkinson’s Detection via Eye Movement Analysis
NeuroVision is a Flask-based web application that uses a deep learning model to detect Parkinson’s disease from eye movement images. Built for neurologists and medical researchers, it provides:
⚡ Real-time AI predictions: Instantly classify images as Parkinson’s Detected or No Parkinson’s
🔐 User authentication: Secure login/signup using a SQLite database
🧭 Neurologist locator: Search and locate specialists via an interactive map
💻 Modern UI: Responsive single-page interface with a sleek glassmorphism design
🔬 Developed as part of a research project to bridge AI and clinical diagnosis.
| Component | Technologies Used |
|---|---|
| Frontend | HTML5, CSS3 (Flexbox/Grid), JavaScript (Vanilla), Leaflet.js (Map) |
| Backend | Python, Flask (REST API), SQLAlchemy (ORM) |
| AI Model | TensorFlow (Convolutional Neural Network for classification) |
| Database | SQLite (Lightweight, serverless) |
| Dev Tools | VS Code, Git |
Single-page layout for seamless experience
Drag-and-drop image uploads with preview
Animated modals for login and signup
Upload eye movement images for instant predictions
CNN model trained on preprocessed data (resizing, normalization)
Displays prediction confidence and visual feedback
Interactive, searchable table of neurologists
Embedded Leaflet.js map with clickable location markers
Passwords hashed with Bcrypt
Session handling using localStorage
Modular Flask backend for future scalability