Comparison of Segmentation Algorithms for Leukemia Classification

Abstract

Leukemia is a deadly cancer that results from the proliferation of non-differentiated white blood cells in blood as compared to the other two types of cells, i.e., red blood cells and platelets. These cells are known as blasts cells which overcrowd other cells rendering those cells as inefficient in their functions and are are themselves non-functional. This paper presents a comparative study of four different segmentation techniques on the images of peripheral blood smear and the classification of these images into diseased and healthy cells using the SVM classifier. The best result was obtained by a custom threshold method of segmentation with a classification accuracy of 96.89%.