Vision Transformer for Pneumonia Classification in X-ray Images

Pneumonia is a common medical condition, usually caused by a lung infection, which causes the tissues in the lungs to become inflamed and affects the functioning of the lungs. Pneumonia ranges from mild pneumonia to life-threatening severity. Identifying the responsible pathogen can be difficult. Diagnosis is often based on symptoms and physical examination, which includes chest X-rays. However, the examination of chest X-rays is a challenging task and is prone to subjective variability. In this study, we focus on the research of a new image classification algorithm for classifying images indicating pneumonia pathology. The proposed method uses the Vision transformer architecture to extract data characteristics and classify the input image as pneumonia or not. Two popular deep learning architectures are compared: Vision transformer and Convolutional Neural Network. In this work, we evaluate Vit-B/16 (for Vision transformer) compared to Convolutional Neural Network algorithms such as MobileNetV2, VGG16, ResNet-50. In this study, the Vision transformer algorithm gives relatively positive classification results with an accuracy of approximately 94%.