A segmentation approach for tissue images using non-dominated sorting GA

Abstract

Medical images are usually used for assisting the doctors to make decisions or diagnoses, the segments are commonly corresponded to different tissue classes, pathologies, and other biologically relevant structures, thus it is very important in the medical diagnose. This paper uses adipose tissue images for example to show the feasibility of the proposed non-dominated sorting Genetic Algorithm (NSGA) model for segmentation. NSGA-based segmentation approach is capable of find the best solution which is close to the Pareto frontier based on the hierarchical structure of population. The experiments show the outperformance of the proposed model over NSGA and sorting Genetic Algorithm (SGA) approaches. The outperformance of the proposed model may attributed to the adaptive determination of the parameters for working out the sharing function which gives the positive impacts on the algorithms.