ANALYSIS OF LUNG CT IMAGES BASED ON FISHER CRITERION AND GENETIC OPTIMIZATION

Abstract: In this paper a simple technique for the identification of lung diseases from CT (Computed Tomography) image is proposed. The overall performance of the classification process is improved by the proper classification and selection of optimal features from the Common Computed Tomography (CT) Imaging Signs of Lung diseases (CISLs). Here the feature selection process is performed based on the Genetic algorithm in which the fisher criterion is used for the objective function and used to employ the best fitness function. Now the selected features are classified using the different classifiers such as Support Vector Machine (SVM), Bag of Features, Bayesian, k-Nearest Neighbor (k-NN) and Ada Boost (Ada) classifiers. Eventually the comparison among the classifiers is done based on performance.

Keywords: Computed Tomography (CT), Common CT Imaging Signs of Lung diseases (CISLs), Genetic Optimization, Fisher Criterion