Classification System for Leukemia Cell Images based on Hu Moment Invariants and Support Vector Machines

Abstract

Leukemia is cancer that attacks the tissues of white blood cells. It occurs when the body produces abnormal blood cells exceeding normal limits; thus, causing them not to function properly. It has a huge effect on the immune system of humans. Medical personnel currently need a long time to recognize leukemia and distinguish acute leukemia cells from normal cells. This study aims to build a classification system of white blood cell images using a feature extraction technique with Hu moment invariants and Support Vector Machine (SVM) classification methods. In this study, the data of 800 blood image samples were divided into two classes, acute and normal, with each class having 400 sample images. The calculation of the average accuracy and average time value on the system obtained the accuracy value of 88% and the required time of 3.73 seconds. The highest accuracy values for the testing data is 95% with duration time 0.89 seconds. The system could classify the leukemia images using Hu moment invariants and SVM.