A Deep Learning Based Framework for Malaria Diagnosis on High Variation Data Set

Malaria is a globally widespread disease caused by parasitic protozoa transmitted by infected female Anopheles mosquitoes. It is caused in humans only by the parasite Plasmodium, further classified into four different species. Identifying malaria parasites is possible by analyzing digital microscopic blood smears, which is tedious, time-consuming, and error-prone. Therefore, automation of the process has assumed great importance as it helps the laborious manual process of review and diagnosis. This work proposes a real-time malaria parasite detector and classification system after studying the YOLOv5 detector and comparing different off-the-shelf convolutional neural network architectures for four-class classification on Plasmodium Falciparum life stages. The results show that the use of the networks YOLOv5 and DarkNet-53 reaches great accuracy in detecting and classifying the life stages of Plasmodium Falciparum, achieving an accuracy of 95.2% and 96.02%, respectively, and outperforming the state-of-art. The obtained results enable broad improvements geared explicitly towards recognizing types and life stages of less common species of malaria parasites, even in mobile environments.