Abstract

Abstract: Lungdiseases are very serious health problems in the life of people. These diseasesinclude chronic obstructive pulmonary disease, pneumonia, asthma, tuberculosis,and lung diseases. The timely diagnosis of lung diseases is very important.Many methods have been developed for this purpose. In this paper, wedemonstrate the feasibility of classifying the lung pathologies in lung X-raysusing conventional and deep learning approaches. In the paper, convolutionalneural networks (CNNs) are presented for the diagnosis of lung diseases. Thearchitecture of CNN and its design principle are presented. For comparativepurpose, backpropagation neural networks (BPNNs) with supervised learning,competitive neural networks (CpNNs) with unsupervised learning are alsoconstructed for diagnosis lung diseases. All the considered networks CNN, BPNN,and CpNN are trained and tested on the same lung X-ray database, and theperformance of each network is discussed.