LIVE APP: https://cnn-xray.herokuapp.com/
Web Application (flask) for Pneumonia using Convolution Neural Network.
Pneumonia is an infection that inflames the air sacs in one or both lungs. It kills more children younger than 5 years old each year than any other infectious disease, such as HIV infection, malaria, or tuberculosis. Diagnosis is often based on symptoms and physical examination. Chest X-rays may help confirm the diagnosis.
This dataset contains 5,856 validated Chest X-Ray images which is taken from https://data.mendeley.com/datasets/rscbjbr9sj/3. A 13 layered Deep Neural Network was trained on this dataset and the accuracy was around 90%.
Using flask web framework, I created a web application using the model which predicts whether the input x-ray image is infected with pneumonia or not. Thanks to Daniel Kermany, Kang Zhang and Michael Goldbaum for making the dataset publicly and freely available.
Model's accuracy is 89% whereas it's loss is about 0.28.
Here are images showing model's architecture and the screenshots of the web application:
