Lung Cancer Detection using Deep Learning
Overview
This project is an AI-based medical imaging system that detects lung cancer from CT scan images using a Convolutional Neural Network (CNN).
The system analyzes lung CT scan images and classifies them into three categories:
- Normal
- Benign (non-cancerous tumor)
- Malignant (cancerous tumor)
A web application built with Streamlit allows users to upload CT scan images and receive predictions from the trained deep learning model.
Technologies Used
- Python
- TensorFlow / Keras
- OpenCV
- NumPy
- Scikit-learn
- Streamlit
Project Structure
lung-cancer-detection
│
├── dataset/
│ ├── normal cases/
│ ├── benign cases/
│ └── malignant cases/
│
├── train_model.py
├── predict.py
├── app.py
├── requirements.txt
└── README.md
Features
- Deep learning model for lung cancer detection
- CT scan image preprocessing using OpenCV
- Classification into Normal, Benign, and Malignant cases
- Streamlit web interface for uploading CT scan images
- Real-time prediction results
Installation
1. Clone the repository
cd lung-cancer-detection-ai
2. Install required libraries
pip install -r requirements.txt
Train the Model
Run the following command to train the CNN model:
python train_model.py
This will generate the trained model file:
lung_cancer_model.h5
Run the Web Application
Start the Streamlit app:
streamlit run app.py
Then open the local link shown in the terminal (usually
http://localhost:8501).
Upload a CT scan image to detect whether it is Normal, Benign, or Malignant.
Dataset
This project uses a lung CT scan dataset containing three classes:
- Normal cases
- Benign cases
- Malignant cases
Example dataset: IQ-OTH/NCCD Lung Cancer Dataset.
How It Works
-
User uploads a lung CT scan image.
-
The image is resized and normalized.
-
The trained CNN model analyzes the image.
-
The model predicts one of the three classes:
Disclaimer
This project is developed for educational and research purposes only and should not be used for real medical diagnosis.
Future Improvements
- Improve model accuracy using Transfer Learning (ResNet50 / EfficientNet)
- Add Grad-CAM visualization to highlight tumor regions
- Deploy the application publicly for online access
- Improve dataset size for better generalization
Author
Manoj Badhan
BTech AI & Robotics Engineering
