Analysis of CT lung images using orthogonal moment features

Abstract

This paper is a study on texture analysis of the CT lung images by the segmentation methods with orthogonal moments features. The CT lung images, normal, cavitary and miliary tuberculosis are taken for study. Owing to the features of orthogonal moments are found suitable for the medical image diagnosis. The segmentation method includes histogram, Watershed and Edge based Segmentation as a base for the evaluation. The orthogonal features used here are Tchebychef, Legendre and Multi resolution enhanced orthogonal polynomial model. In order to get the features, the orthogonal moments are computed from the segmented image. The proposed work is a comparative study on performance in terms of accuracy based on orthogonal moment with segmentation methods. The accuracy is obtained through sensitivity and specificity which are computed from texture features and intensity features. The results reveal that the computation model is an efficacy for medical images.