Blood smear imagery dataset for malaria parasite detection: A case of Tanzania.

Abstract

Malaria is a major public health issue in many regions of Africa, including Tanzania. The Tanzania Malaria National Strategic Plan (2021-2025) emphasizes on high-quality testing services availability, high coverage of timely diagnosis of malaria, and availability of innovative diagnostic systems for effective detection, treatment and control of malaria. This would be achieved by employing state of the art technologies like Machine learning. However, Machine learning requires diverse dataset to work effectively and efficiently. Therefore, this paper presents blood smear imagery dataset that can be used by researchers to develop computer vision systems for malaria parasite detection. The imagery dataset were acquired by setting up a 40X-2500X Real 4 K compound microscope with a 4k SONY IMX334 sensor camera mounted to it in five health centres of Tanga region. Blood samples taken according to normal routine of diagnosing patients in health care, were stained using Giemsa reagent and examined under microscope. Following these procedures, the study collected and annotated Thick infected blood smear images ( $n=1139)$ ; Thick uninfected blood smear images ( $n=1071$ ); Thin uninfected blood smear images ( $n=270$ ); and Thin infected blood smear images ( $n=1064$ ). Furthermore, the curated dataset have been uploaded in a public Harvard data verse repository. In summary, the dataset aims to support the creation of diagnostic tools that improve malaria detection, thereby advancing health outcomes and aiding malaria control initiatives in Tanzania and other regions impacted by the disease.