OncoVision AI is a glassmorphic medical web application built with Flask, Python, Scikit-Learn, and Chart.js. It predicts breast cancer malignancy risk in real time from cytological nuclear cell measurements, using interactive sliders, clinical presets, and population-level radar analytics.
🔗 Live Demo: https://breastcancerdetection-ausd.onrender.com/
- 🎨 Medical Glassmorphic UI — dark slate theme with magenta-pink glow accents, blurred glass cards, and subtle animations.
- 🔬 Interactive Diagnostic Panel — 10 adjustable cell nucleus features: Radius, Texture, Perimeter, Area, Smoothness, Compactness, Concavity, Concave Points, Symmetry, and Fractal Dimension.
- ⚡ One-Click Clinical Presets
⚠️ Malignant Case— loads sample malignant measurements✅ Benign Case— loads sample benign measurements🔄 Baseline— resets inputs to dataset mean values
- 📊 Diagnostic Analytics
- Radar chart comparing patient input against benign/malignant population baselines
- Feature importance chart showing classifier weights
- 📦 Zero-Setup Model Backend — automatically trains a baseline classifier if
model.pklisn't found, so the app runs out of the box.
breastCancerDetection/
├── app.py # Flask server & prediction API endpoints
├── requirements.txt # Python dependencies
├── model/ # Trained model artifacts
│ ├── model.pkl # Saved classifier
│ ├── scaler.pkl # Feature scaler
│ └── lr.pkl # Logistic Regression model
├── notebook/ # Model training & experimentation notebooks
├── setup guide/ # Setup instructions / documentation
├── static/
│ ├── css/ # Glassmorphic styling
│ └── js/ # Chart.js logic, API calls, preset handlers
└── templates/
└── index.html # Web app front end
git clone https://github.com/Ayu-this-side/breastCancerDetection.git
cd breastCancerDetectionpython -m venv venv
# Windows
venv\Scripts\activate
# macOS/Linux
source venv/bin/activatepip install -r requirements.txtpython app.pyhttp://127.0.0.1:5000
The prediction engine uses Logistic Regression trained on the Breast Cancer Wisconsin (Diagnostic) dataset, with features scaled before inference. If no pre-trained model is found in model/, the app trains one automatically on startup.
| Layer | Tools |
|---|---|
| Backend | Python, Flask, Scikit-Learn, NumPy, Pandas, Joblib |
| Frontend | HTML5, CSS3 (glassmorphism, custom properties), JavaScript (ES6+) |
| Visualization | Chart.js |
| Fonts | Plus Jakarta Sans, Outfit (Google Fonts) |
OncoVision AI is an educational project demonstrating machine learning integration with web technology. It is not a certified diagnostic tool and should never replace professional medical evaluation.