Pneumonia Prediction with Chest X-Ray Images using Convolutional Neural Networks
Developed a 19-layer CNN with 5-fold cross-validation, achieving 80% accuracy in classifying chest X-rays
Project information
- University: Rey Juan Carlos University of Madrid
- Course: Medical Image Analysis
- Project date: Sep. 2021 - Dec. 2021
Summary
Designed a convolutional neural network (CNN) using Keras for the classification of chest X-ray images into healthy, viral pneumonia, and bacterial pneumonia categories. The model consists of 19 layers and was trained using 5-fold cross-validation to ensure robustness. Utilized an adaptive optimizer (Adam), with categorical cross-entropy as the loss function, and incorporated an early stopping mechanism to prevent overfitting. TensorFlow and Sklearn were employed to implement and refine the model, achieving a diagnostic accuracy of 80%.