In the last decades, the classification of images was established as a typical method for diagnosing many abnormalities and diseases. The purpose of an efficient classification method is considered essential in modern diagnostic medicine in order to increase the number of diagnosed patients and decrease the analysis time. The significant storage capabilities of electronic media have enabled research centers to accumulate repositories of classified (labeled) images and mostly of a large number of unclassified (unlabeled) images. Semi-supervised learning algorithms have become a hot topic of research as an alternative to traditional classification methods, seeing as they exploit the explicit classification information of labeled data with the knowledge hidden in the unlabeled data resulting in the creation of powerful and effective classifiers. In this work, we propose a new ensemble self-labeled algorithm, called DTCo, for X-ray classification. The efficacy of the presented algorithm is illustrated by a series of experiments against other state-of-the-art self-labeled methods.
Keywords: Ensemble learning; Lung abnormalities; Self-labeled algorithms; Semi-supervised learning; X-ray classification.