Efficient Classification Techniques in Classifying Human Intestinal Parasite Ova

Abstract

Helminth parasites live in the human body and can cause serious health problems that will lead to cancer and may cause death in patients. These parasitic helminths congregate in the intestines to mate and produce ova. Therefore, early identification screening is necessary to prevent the spread of helminth parasites throughout the body. A manual microscopic feces test is still the most often used approach for helminth detection. As a result, the purpose of this research is to investigate the effectiveness of three classifiers in classifying four types of human intestinal parasite ova. Three classifier techniques used are k-Nearest Neighbourhood (k-NN), Support Vector Machine (SVM), and Ensemble classifier. There are four types of helminth ova which are Ascaris Lumbricoides Ova (ALO), Enterobius Vermicularis Ova (EVO), Hookworm Ova (HWO), and Trichuris Trichiura Ova (TTO). A total of 664 helminth parasite ova images were analyzed, consisting of 166 images from each helminth species. The Linear kernel function from the SVM classifier has obtained the highest accuracy performance reaching 92.23%. Followed by Cityblock distance from the k-NN classifier with an accuracy of 91.16% and AdaBoostM2 from the Ensemble classifier with an accuracy of 89.94%.