Deep Learning Approach for Malaria Parasite Detection in Thick Blood Smear Images

Abstract

Malaria is caused by a bite of female anopheles mosquitos transmitting the parasite Plasmodium into human bodies. Malaria is a common disease in tropical and subtropical regions and is also a severe public health problem due to its risk. Early diagnosis is required to avoid the hazard of death from malaria. Microscopic analysis of blood smears remains a standard method for malaria analysis. However, manual microscopic observation is laborious, and the results have a heavy dependence on the examiner’s skill. To alleviate this problem, this study proposed a deep learning method for detecting malaria automatically malaria parasite on thick blood smear microscopic images. The proposed approach achieved the fastest examination at 0.25 sec/image or more than 20 times faster compared to that of previous with mAP, sensitivity, and a precision score of 72, 78.4, and 83.2 %, respectively. These performances indicated that the proposed approach can be a promising alternative to CAD systems for fast parasite detection.