PneumCNN

A Convolutional Neural Network approach for diagnosis of Pneumonia from chest X-ray reports.

Getting Started

1. Clone this repository
2. To set the virtual environment and install dependencies run

    virtualenv -p python3 venv  # to be created only once
    source venv/bin/activate
    pip install -r requirements.txt
   

3. To run the streamlit app, run

   streamlit run streamlit.py
   

Project Layout

cnn.ipynb: is the main notebook that has the network model, The model is saved after training into a JSON format, and the weights are saved in an h5 file.

load_model.py: has a function test which loads the model and the weights. Then it predicts whether the input image (in the form of an numpy.ndarray) is pneumonic or not

streamlit.py: is a minimal GUI made using Streamlit. Users can upload an image in the app. The app invokes the test function from the load_model.py file and passes the image in the form of a numpy.ndarray