Image Analysis for Detecting Malaria Cell Using Otsu Thresholding and Machine Learning Models

Abstract

Motivation : Malaria is a dangerous disease that affects thousands of individuals each year all around the world. It can be fatal if not treated promptly. According to the most recent World Malaria Report from the World Health Organization, there would be 241 million malaria cases and 627 000 malaria deaths globally in 2020. Despite recent advances in malaria diagnosis, the microscopy approach remains the most widely used. Moreover, the efficiency of microscopic diagnosis is dependent on the expertise of the microscopist, which restricts malaria throughput. Distinguishing parasite development phases remains a very challenging task. Goal: The main aim is to develop a system to identify malaria stages in blood smears using machine learning models. This paper proposes a study of seven machine learning models and one ensemble model to foresee which model will better predict the malaria stage. Results: To avoid a large number of individuals from being infected with malaria, an early and precise diagnosis is essential. A web-based application is developed for the end-user using a flask, where the user can upload the sample images of the multi-stage malaria parasite and recognize the cell image. This will help the doctors to take the necessary steps to prevent the disease and choose the appropriate solution.