Comparison of back-propagation ANN and Naϊve-Bayes method in identifying lung X-ray images

Abstract

This paper presents the results of a comparison of two methods for the identification of X-ray images of the lungs, namely the ANN back-propagation method and the Nai"ve-Bayes method. The ANN back-propagation method works based on the final weighting of each network as a result of the identification training process. The Nai"ve-Bayes method works based on the probabilities of each characteristic of the identification training process. The process of identifying X-ray images of the lungs is carried out based on the extraction of histogram features on X-ray images of the lungs to be identified into three groups, namely X-ray images of normal lungs, X-ray images of lung cancer, and an X- ray image of the lung affected by effusion. This study aims to compare the strengths or weaknesses between the weight- based ANN back-propagation method and the probability-based Nai"ve-Bayes method in the ability to identify lung X-ray images. The results of this study indicate that the Nai"ve-Bayes method is superior to ANN back-propagation in identifying X-ray images of lungs.