AUTOMATIC CLASSIFICATION OF MALARIA PARASITE INFECTED BLOOD SAMPLES BASED ON THEIR SPECIES AND DEVELOPMENTAL STAGES USING IMAGE PROCESSING

Abstract

Malaria is one of the deadly diseases in sub-Saharan countries including Ethiopia. It is caused by Plasmodium parasite transmitted by mosquito bites. An infected female anopheles mosquito is mainly responsible for the transmission of the disease. In this study an automatic classification of malaria parasite infected blood samples based on their species and developmental stages was investigated using image processing. Database consisting of 763 images of RBC of malaria parasites were utilized. The data were analyzed by utilizing standard image processing tools such as histogram equalization, binarization, and Fourier descriptors (FDs) of one dimensionalcontour extracted from the binarized shape. One dimensional extracted contour, showing the boundary of the object, represented in complex coordinates was transformed using Fourier transformation. Fourier descriptors characterize objects’ shape in frequency domain. The descriptors were both positive and negative frequency axis and the classification between species and developmental stages of parasite infected images can be understood (classified) using the extracted features of erythrocytes in blood images using SVM classifiers. All the techniques were implemented using MATLAB (Version R2016b) platform. The results show an average classification accuracy rate of 97.1% in recognizing Plasmodium species and 95.85% in recognizing their developmental stages respectively.