Detection and Classification of Malaria Parasite Using Python Software

Abstract

Around 228 million Malaria instances have been expected in 89 countries worldwide as per data given by the World Malaria Report of 2019. India has made a vision of a malaria-free country by 2027 and complete malaria elimination by 2030. The Plasmodium parasites are responsible for malaria disease. Malaria is a blood disorder caused due to bite of the female Anopheles mosquito. The disease malaria is of five different types depending upon the parasite species: Plasmodium vivax, Plasmodium ovale, Plasmodium malariae, Plasmodium Knowlesi, Plasmodium falciparum. The general method used to detect Plasmodium using blood films is Malaria microscopy. The WHO approved this microscopic malaria detection method. In this method, blood has been collected through the finger, pricked and spread on the clean grease-free plate glass, and left to be airdried. Thin blood smears have been utilized for parasite identification in the microscope. If the presence of parasites is low, thick blood smears are used. The accuracy of work depends on the skills and experience of the technicians. The work presents an algorithm to detect the presence of malaria parasites by counting the infected blood cells. The obtained images are Giemsa-stained blood smear images, and these colored images are firstly transformed into gray-scale images. Then basic image enhancement techniques have been employed to minimize noise and enhance the contrast of these images. The cells' segmentation has been done, followed by classification using python Jupyter notebook to detect the parasites efficiently. A Convolutional Neural Network classifier has been used to differentiate positive and negative cases of malaria. This classifier gives an accuracy of about 97%, which is quite good. It can be employed in laboratories to save quality time and for accurate determination.