Description

ABSTRACT The classification of the medical images selection of the optimal features improves the performance of the classification process. The selection of the best features reduces the time complexity and algorithm complexity of the overall process. CT images were used for the identification of the diseases in the lung images. Features were extracted from the images based on B-HOG features, Wavelet features, LBP features and CVH features. The feature selection process were employed based on the Genetic algorithm in which the Fisher criterion is employed for the objective function. The selected features were then classified using five different classifiers to test the efficiency of the selected features. The overall performance of the process were measured based on the performance metrics. The main objective of the process is to select the optimal features from the different type of features extracted using different methods. To compare the performance of different classifiers based on the selected features. To extract different types of features like texture based, intensity based features from the images. To employ best fitness function based on the fisher criterion to select the optimal features. To combine fitness function with Genetic optimization in order to improve the efficiency of Genetic algorithm. The objective functions were initially defined and based on the initial objective function value the genetic optimization steps were employed. Mutation and Crossover operations were the basic steps in the genetic algorithm. In the crossover step the new population of genes were initialized by modifying the previous population of genes. After the crossover step mutation step is employed. In the crossover step child chromosomes were created. After the mutation and the crossover step the stopping condition is verified and the process is repeated till the stopping condition is reached. The selected features were then classified using different classifiers like SVM, Bag Of Features, Naive Bayes, k-NN, Adaboost, inorder to compare the performance of the classifiers. The performance of the process were measured based on the performance metrics like Accuracy, Sensitivity and Specificity.