Pneumonia Detection and Analysis System

Overview

This project implements an AI-powered Pneumonia Detection and Analysis System using a Convolutional Neural Network (CNN). The system is designed to detect pneumonia from chest X-ray images, providing predictions and detailed analysis through an interactive user interface built with Streamlit.

Features

• Pneumonia Detection: Utilizes a custom-trained CNN model to detect pneumonia in chest X-ray images with high accuracy.
• Interactive User Interface: Built using Streamlit for a seamless and user-friendly experience.
• Security: Implements AES-256 encryption for image data to comply with HIPAA regulations.

Project Structure

pneumonia-detection/
├── chest_xray/                   # Directory for dataset and image data          
│   ├── train/
│   ├── val/                   
│   └── test/
├── Images                        # Directory for plotting images                   
├── models/                       # Directory for trained models
│   ├── pneumonia_cnn.pth         # Custom-trained CNN model
│   └── pretrained_model.pth      # Pre-trained model (e.g., ResNet)
├── pneumoniaCNN.ipynb
├── transfer_learning.ipynb       # Transfer learning       
├── data_loader.py            
├── CNN.py                  
├── train.py                                        
├── UI.py                         # Interactive UI using Streamlit
├── pipeline.py                   # Pipeline for training and inference                  
├── Dockerfile                    # Docker configuration for containerized app
├── requirements.txt              # Python dependencies
└── README.md                     # Project documentation

Installation

Prerequisites

• Python 3.8+
• PyTorch
• Streamlit

Setup Instructions

1. Clone the repository:

   git clone https://github.com/codewithdark-git/Pneumonia-Detection-from-Chest-X-Rays.git
   cd Pneumonia-Detection-from-Chest-X-Rays

2. Install the required dependencies:

   pip install -r requirements.txt

Running the Application

• To run the application locally with the Streamlit UI:

  streamlit run UI.py

• To run the pipeline for training the model:

  python pipeline.py

Model Details

The system employs the PneumoniaCNN architecture, a custom-built CNN with three convolutional blocks and a fully connected layer. The model is trained using chest X-ray images, and both custom and pretrained models (such as ResNet) are available for comparison.

• Custom Model Accuracy:

  • Training Accuracy: 85%
  • Testing Accuracy: 89%

• Pretrained Model Accuracy:

  • Testing Accuracy: 90%

Security and Privacy

• Data Encryption: All uploaded X-ray images are encrypted using AES-256 encryption.
• HIPAA Compliance: The system adheres to HIPAA regulations for secure handling of medical data.

Future Enhancements

• Mobile App: Develop a mobile version of the app for easier access in healthcare environments.
• Batch Processing: Implement a system for batch processing of multiple X-ray images.
• Model Optimization: Experiment with more advanced CNN architectures to improve accuracy and inference speed.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• PyTorch: For building the deep learning models.
• Streamlit: For creating the interactive UI.

---