Chest X-Ray Disease Prediction Web App

This is a Flask web application that uses a trained PyTorch model to predict diseases from chest X-ray images.

Features

• Upload X-ray images via drag-and-drop or file selector
• Preview uploaded images before prediction
• Disease prediction with probability percentages
• Visual results with plots showing the original image and probability distribution

Setup Instructions

1. Ensure you have Python installed (version 3.7 or higher recommended)

2. Activate the virtual environment:

   .\venv\Scripts\activate
   

3. Install the required packages (if not already done):

   pip install flask torch torchvision pillow numpy matplotlib
   

4. Place your trained model in the model directory:

   model/best_model.pth
   

   Note: You need to have a trained model. The model should be a PyTorch checkpoint file.

5. Run the Flask application:

   python app.py
   

6. Open your web browser and navigate to:

   http://127.0.0.1:5000/
   

Using the Web App

1. Drag and drop a chest X-ray image onto the upload area or click the "Select File" button to choose an image file.
2. After uploading, a preview of the image will be displayed.
3. Click the "Predict Diseases" button to analyze the image.
4. Results will show on the right side, including:
   • A visual plot of the X-ray and disease probabilities
   • A table listing all diseases and their predicted probabilities

Model Information

The model uses a DenseNet121 architecture pre-trained on ImageNet and fine-tuned for multi-label classification of 14 different lung diseases commonly found in chest X-rays:

• Atelectasis
• Cardiomegaly
• Effusion
• Infiltration
• Mass
• Nodule
• Pneumonia
• Pneumothorax
• Consolidation
• Edema
• Emphysema
• Fibrosis
• Pleural Thickening
• Hernia