A Performance Analysis of GA-ELM Classifier in Classification of Abnormality Detection in Electrical Impednce Tomography (EIT) Lung Images

Abstract

This article presents the performance analysis of Genetic Algorithm (GA)-Extreme Learning Machine (ELM) classifier by comparing with Genetic Algorithm (GA)-Support Vector Machine (SVM) classifier for detecting the abnormalities from Electrical impedance tomography images. The machine learning algorithms, Extreme Learning Machine (ELM), and Support Vector Machine (SVM) are used for classification and a Genetic Algorithm (GA) is used as feature selector to reduce the high dimensional features needed for classification. The Gray Level Co-occurrence Matrix (GLCM) and intensity histogram are used for texture feature extraction from the EIT images. The EIT lung images are reconstructed using one step linearized Gauss-Newton (GN) algorithm. Detection of lung injury is one of the critical issue where excessive care has to be taken for better diagnosis and treatment. Any classifier needs to detect the non-ventilated regions with respect to efficiency and performance. The performance analysis of these two classifiers are analyzed based on the benchmark parameters such performance index, sensitivity, specificity, average detection and F-score. From the experimental results it is evident, that the Extreme Learning Machine has performed well compared with the Support Vector Machine and also Extreme Learning Machine classification performance has been increased for genetic algorithm based feature selection.