Microscopic Image Segmentation Based on Swarm Intelligence Optimization Algorithms

Abstract

The World Health Organization suggests the visual examination of stained sputum smear samples as a preliminary and basic diagnostic technique for diagnosing tuberculosis which is the most common infectious disease in the world. Due to the fact that the visual examination of slide samples performed by expert laboratory technicians requires much time and the process is prone to mistake, an accurate diagnosis of disease is provided with computer aided automatic diagnosis methods. In this study, the usage of swarm intelligence algorithms based on entropy information are proposed for detecting the tuberculosis bacilli as an ovelap approach in segmentation of microscopic images. The microscopic images used in the study are taken from smear samples in which the background concentration is low and bacilli concentration is low and high. An optimum threshold value in gray-level microscopic image is determined using the bi-level entropy based Particle Swarm Optimization, Firefly Algorithm, Cuckoo Search Optimization and Flower Pollination Algorithm. The acquired visual results show that the proposed swarm intelligence algorithms are quite successful in segmentation of microscopic images.