Automated segmentation of Giemsa stained microscopic images based on entropy value

This paper proposes a computerized approach to extract and examine the Giemsa stained microscopic medical images. Image examination is an essential procedure in most of the disease identification procedures. In the proposed work, a soft computing approach is implemented to analyze clinical images registered with digital microscope. In this approach, recently developed soft computing procedure known as the Social Group Optimization (SGO) and the entropy function is chosen to mine the Region Of Interest (ROI) from Giemsa stained digital images. In this work, extraction of the blood cell region and the plasmodium species is chosen as the problem. Proposed experimental work is implemented using MATLAB. A relative analysis is presented between the entropy functions, like Kapur and Tsallis. The performance of the Kapur/Tsallis functions is assessed using well known image quality measures and the shape measured based on the GLCM. The experimental outcome confirms that, Kapur's approach is efficient in extracting the ROI from the digital microscopic images with lesser CPU time compared with Tsalli's entropy.