Malaria Parasite Detection and Species Identification on Thin Blood Smears Using a Convolutional Neural Network

To aid efforts for the total elimination of malaria, effective and fast diagnosis of cases must be done. The gold standard for malaria diagnosis is microscopy. This process becomes problematic when cases are in far-flung rural areas as experts may not be present in these areas to make such diagnosis. Automation of the diagnostic process with the use of an intelligent system that would recognize malaria parasites could solve this problem. This study proposes such an intelligent system, detecting malaria parasites through images of thin blood smears. We used a Convolutional Neural Network in this research where we obtained an accuracy of 92.4% and sensitivity of 95.2% for malaria parasite detection, and an accuracy of 87.9% for identifying the two species Plasmodium falciparum and Plasmodium vivax.