Detecting Lung Abnormalities From X-rays Using an Improved SSL Algorithm

A significant component in computer-aided medical diagnosis is the automatic detection of lung abnormalities from digital chest X-rays; thus it constitutes a vital first step in radiologic image analysis. During the last decades, the rapid advances of digital technology and chest radiography have ultimately led to the development of large repositories with labeled and unlabeled images. Semi-supervised learning algorithms have become a hot topic of research, exploiting the explicit classification information of labeled images with the knowledge hidden in the unlabeled images. In the present work, we propose a new semi-supervised learning algorithm for the classification of lung abnormalities from X-rays based on an ensemble philosophy. The efficacy of the presented algorithm is demonstrated by numerical experiments, illustrating that reliable prediction models could be developed by incorporating ensemble methodologies in the semi-supervised framework.