Project overview
Developed a convolutional neural network to classify chest X-ray images into healthy, viral pneumonia and bacterial pneumonia categories. The model used a 19-layer architecture and achieved 80% diagnostic accuracy.
Approach
Implemented the model in Keras using TensorFlow and Scikit-learn. Applied 5-fold cross-validation, the Adam optimiser and categorical cross-entropy loss, with early stopping to improve robustness and reduce overfitting.
Outcome
The model achieved 80% accuracy in distinguishing between healthy cases, viral pneumonia and bacterial pneumonia in chest X-ray images. The results demonstrate the potential of deep learning for automated medical-image classification.