Pneumonia Detection From X-ray Images Using Convolutional Neural Networks

Abstract

In my thesis I examine how the application of artificial intelligence (AI) can be used for automatic classification of pneumonia from normal chest radiographs (X-ray images) and create an analysis on how advantageous certain artificial neural networks for this classification task are based on their performance metrics. Implementing AI for medical image processing applications creates an opportunity to refine and enhance diagnostic accuracy and consistency of radiograph interpretation. It can be beneficial to modern healthcare in improving quality of work medical staff and speeding up treatment and diagnostic procedures. Convolutional neural network architectures used in this thesis include VGGNet, InceptionNet, ResNet and DenseNet. All the networks are pre-trained on ImageNet dataset and are implemented with transfer learning technique.