Malaria Parasite Classification using Deep Convolutional Neural Network

Abstract

Malaria is one of the life-threatening diseases which spread by the Plasmodium parasites. Traditionally, microscopists analyze the microscopic blood smear images but it is time consuming and may leads to false negatives. Automated detection of malaria from the thin blood smear slide images is a challenging task. However, in the domain of medical and healthcare applications, classification accuracy plays a vital role. The higher level of false negatives in medical diagnosis systems may raise the risk of the patients by not employing the required treatment they exactly need. In this article, we have developed three Convolution Neural Network (CNN) models for the prediction of malaria from the red blood cell images into infected parasite red blood cells and uninfected parasite red blood cells. Finally, out of the three setups, proposed CNN setup-1 with kernel size 3 x 3 and pool size of 2 x 2 achieved an accuracy of 96%.