Cancer Type Prediction using Deep Learning

This repository contains a deep learning-based cancer type prediction system using a trained convolutional neural network (CNN). The model is deployed using Streamlit, allowing users to upload medical images and receive predictions with a probability distribution displayed in a pie chart.

Features

• Upload an image of a tissue sample.
• Get real-time cancer type predictions.
• Visual representation of prediction probabilities using a pie chart.
• Uses a trained CNN model (Model.h5).

Cancer Types Detected

The model predicts five types of tissue conditions:

• Colon Adenocarcinoma
• Colon Benign Tissue
• Lung Adenocarcinoma
• Lung Benign Tissue
• Lung Squamous Cell Carcinoma

Requirements

Ensure you have Python installed along with the following dependencies:

pip install streamlit tensorflow numpy matplotlib

How to Run

1. Clone the repository:

   git clone https://github.com/Uni-Creator/LungCancerClassification.git
   cd LungCancerClassification

2. Place the trained model file (Model.h5) in the project directory.
3. Run the application:

   streamlit run main.py

4. Open the provided local URL in your web browser.

File Descriptions

• main.py: Contains the Streamlit-based web app for cancer type prediction.
• README.md: Documentation for setting up and using the project.
• lung_colon_image_set: Contains about 2% of the original data. You can download the full dataset from: LC25000 Dataset.

Usage

1. Upload an image (JPEG only).
2. The system will process the image and predict the cancer type probabilities.
3. The results will be displayed in a table along with a pie chart visualization.

Example Output

• Prediction Probabilities:
  • Colon Adenocarcinoma: 90.0%
  • Colon Benign Tissue: 10.0%
  • Lung Adenocarcinoma: 0.0%
  • Lung Benign Tissue: 0.0%
  • Lung Squamous Cell Carcinoma: 0.0%

License

This project is open-source. Feel free to modify and improve it!

Acknowledgments

• TensorFlow for deep learning.
• Streamlit for interactive UI development.
• Medical image datasets used for training the model.