GSA Based Classification of Lung Nodules in CT Images

Abstract

Medical Image processing techniques are extensively used in several medical fields for former detection of diseases. Computer Aided Diagnosis (CAD) systems give way to a great pact of information to analyze and evaluate the accurate structures at short time period. Image quality with detailed edges is the foundation factors of this research, where low pre-processing technique is used for filtering. The adaptive median filter performs spatial processing to preserve detailed edges and smoothen non-impulsive noises. Image segmentation is done using edge detection techniques in lung CT images for Obtaining the Region of Interest (ROI), which is a prior step for feature extraction and selection. Law of universal gravity is used For detecting the differential structures using a new edge detection algorithm, that is Gravitational search algorithm. In this approach the edges are detected by the local variation of intensity values and the movement of agents is computed using Gravitational Search Algorithm (GSA).It gives fast and accurate framework for feature extraction and feature selection. For Feature Extraction Gray Level Co-occurrence Matrix (GLCM) is applicable in extracting statistical texture feature for motion estimation in CT (Computed Tomography) images. In Feature selection Optimal set of Features are selected through classifiers to overcome the computational time and memory exhaustion.

Keywords: Gravitational search algorithm, Gray level Co-occurrence Matrix, Computed Tomography, Computer Aided Diagnosis, Region of Interest