🫁 TB Chest X-Ray Classifier Web Application

Project Overview

This project is a web-based application designed to assist in the classification of Tuberculosis (TB) from chest X-ray images. It leverages deep learning models for both lung segmentation and TB classification, providing visual explanations through Grad-CAM heatmaps. The application features a user-friendly interface for uploading X-ray images and instantly viewing the prediction results along with insightful visualizations.

Features

• Image Upload: Easily upload chest X-ray images via drag-and-drop or file selection.
• TB Classification: Classifies X-ray images as "Normal" or "TB" with a confidence score.
• Lung Segmentation: Visualizes the segmented lung regions from the uploaded X-ray.
• Grad-CAM Heatmaps: Generates and displays Grad-CAM heatmaps to highlight areas of the X-ray that are most influential in the model's classification decision.
• Modern User Interface: A clean, responsive, and intuitive design built with React.js.

Technologies Used

Backend (Python - FastAPI)

• Python 3.x: The core programming language.
• FastAPI: A modern, fast (high-performance) web framework for building APIs with Python 3.7+ based on standard Python type hints.
• TensorFlow/Keras: For loading and running the deep learning models (EfficientNetB0 for classification, U-Net-like for segmentation).
• OpenCV (cv2): Used for image processing tasks such as resizing, color space conversions, and image encoding/decoding.
• NumPy: Fundamental package for numerical computation in Python.
• scikit-image (skimage): For advanced image processing, specifically measure.label and measure.regionprops for mask post-processing.
• SciPy (scipy.ndimage): Used for binary_fill_holes in mask post-processing.
• Uvicorn: An ASGI server for running the FastAPI application.

Frontend (React.js - Vite)

• React.js: A JavaScript library for building user interfaces.
• Vite: A fast frontend build tool that provides a lightning-fast development experience.
• CSS: For styling the application, ensuring a consistent and modern look.

Machine Learning Models

• segmentation_model.h5: A ResUnet++ model for segmenting lung regions from X-ray images.
• best_efficientnetb0_stage2_rgb_fixed.h5: An EfficientNetB0 model trained for the classification of TB from segmented chest X-ray images.

Setup and Installation

Follow these steps to get the project up and running on your local machine.

Prerequisites

• Python 3.8+
• Node.js (LTS version recommended)
• npm or Yarn package manager

1. Clone the Repository

git clone https://github.com/your-username/TB_WEB_APP.git
cd TB_WEB_APP

2. Backend Setup

Navigate to the backend directory, create a virtual environment, install dependencies, and start the FastAPI server.

cd backend
python -m venv venv
.\venv\Scripts\activate # On Windows
# source venv/bin/activate # On macOS/Linux

pip install -r requirements.txt
uvicorn app:app --reload

The backend server will start on http://127.0.0.1:8000.

3. Frontend Setup

Open a new terminal, navigate to the tb-cxr-frontend directory, install dependencies, and start the React development server.

cd tb-cxr-frontend
npm install # or yarn install
npm run dev # or yarn dev

The frontend application will typically open in your browser at http://localhost:5173 (or another port if 5173 is in use).

Usage

1. Upload Image: On the Dashboard page, drag and drop a chest X-ray image into the designated area or click "Browse Files" to select one.
2. Run Prediction: Click the "Run Prediction" button.
3. View Results: The application will display:
   • The classification result (Normal/TB) and confidence score.
   • The original X-ray image.
   • The segmented lung regions.
   • A Grad-CAM heatmap overlaying the segmented lung, indicating areas of importance for the classification.

API Endpoints

POST /api/predict

This endpoint accepts a chest X-ray image and returns the classification, segmented lung image, and Grad-CAM heatmap.

• Method: POST
• Content-Type: multipart/form-data
• Body:
  • image: The chest X-ray image file.
• Response (JSON):

  {
    "classification": {
      "label": "Normal",
      "confidence": 0.9876
    },
    "original_image": "base64_encoded_original_image_png",
    "segmented_lungs": "base64_encoded_segmented_lungs_png",
    "gradcam_on_segmented": "base64_encoded_gradcam_image_png"
  }

Contributing

Contributions are welcome! If you have suggestions for improvements, bug fixes, or new features, please open an issue or submit a pull request.

License

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

Contact

For any questions or inquiries, please reach out to:

• Sheshank Singh
• Email: sheshanksingh2609@gmail.com