AUTOMATED MALARIA PARASITE QUANTIFICATION OF THIN BLOOD SMEAR

Malaria is a common epidemic which usually strikes the tropical and subtropical regions like the South East Asia and Africa. It is a disease which is caused by a parasitic protozoan which is called Plasmodium. There are five types of Plasmodium parasites detected and there are four among them are found in human bodies, namely, Plasmodium Falciparum (PF), and Plasmodium Vivax (PV), Plasmodium Ovale (PO), and Plasmodium Malariae (PM). Among the four, PF and PV are more lethal which contribute to more than half of death cases around the world. The disease is transmitted by a vector called anopheles mosquito which is infected by the Plasmodium parasites. There are two ways of diagnosis for malaria, which is either manual or automated. Manual diagnosis requires more time as compared to the automated one and specialists with high expertise and experience in microscopic and pathological field for execution are needed. The result obtained from manual count is always debatable as the result is always inconsistent and subjective. Automated Diagnosis which saves time, cost and experienced specialists which are hardly available in the market can solve all the issues at once. Till date, it is more prevalent for researchers to focus on using a thick blood smear in counting the parasitemia. This study aims to develop an algorithm which is able to do counting parasitemia with thin blood smear that can help in integrating both detecting types and counting of parasites with just a thin blood smear. Besides, this study aims to automatically classify the level of severity of PV infection based on the parasitemia counted. The algorithm is divided into two parts to detect the parasitemia. The first part is to detect and count the PV infected RBC from S-channel. The second part is processed in grayscale with some morphological operators. The study is able to give a mean accuracy of 74.06% out of 50 samples of thin blood images being tested. The algorithm of this study successfully computes the parasitemia of a thin blood smear. This study shows that this algorithm could add more value to the function of thin blood smear by eliminating the separated processes of detecting and counting parasitemia using two types of blood smears. Other than conventional parasite species determination, this algorithm allows thin blood smear to also be used for counting parasitemia and classifying severity of infection simultaneously.
vi