Performance Analysis of Lung Cancer Classification using Multiple Feature Extraction with SVM and KNN Classifiers

Abstract

All over the world for individual’s death lung malignancy is considered as typical reason. Image processing utilization has increased stage by stage. With tremendous images volume in general radiologist predictions to discover lung malignancy may not be perfect. This paper concentrates at texture feature extraction and classification of lung CT image as normal or affected. Different phases involved are preprocessing, segmentation, feature extraction and classifier. Preprocessing is done using median filter followed by Watershed segmentation. Watershed segmentation is culled for choosing the required region of interest, and then the performance analysis of lung cancer classification is done using multiple features such as GLCM, LBP and HOG and various classifier to choose the best suitable combination of features and classifier for improved classification results. The results and methodology represents further improvements in precision in lung malignancy identification and with enhanced exactness in classification outcomes.