Malaria Detection and Classification in Microscopic Images

Abstract

Malaria is a fatal disease, continues to be a major burden on global health. Malaria is caused by plasmodium parasites that were injected by a certain female mosquito. About half million deaths were caused by malaria every year. The way in which people are diagnosed for malaria is microscopic examination, by observing blood smears under a microscope. However, these techniques are accurate but they are very time consuming, in contrast, using computer vision and deep learning techniques, the proposed system is automated. In this work, we developed an automated and robust diagnosis system to detect malaria parasites. We used Region based Fully Convolutional Neural Network (RFCN) object detection model for detection and classification of malaria parasites. For this work we used labeled dataset with bounding boxes of approximately 1328 images and demonstrated that our work outperforms baseline methods.

Dataset

Results