Review of Decision Tree-Based Binary Classification Framework Using Robust 3D Image and Feature Selection for Malaria-Infected Erythrocyte Detection

Abstract

We start with a famous proverb ‘health is wealth.’ Malaria is one of the most rapidly spreading and contagious diseases, mostly spread through microbes. Efficient treatment of the disease requires early and accurate estimation to ensure control from spreading and treatment in early phases. Accordingly, several studies have been put forward during the past decade. Analyzing the blood smear’s images is one of the prominent works proposed in this context. This manuscript attempts to automate the process of diagnosis through machine learning techniques. The algorithm trains the model through different selected features of the input images and thereby uses the learning experience to classify the blood smears as disease prone or healthy. The cuckoo search algorithm is used for designing a heuristic scale, which is further assessed through multiple experiments to evaluate its accuracy. Different performance evaluation measures like precision, sensitivity, specificity, and accuracy are used to assess the robustness of the model toward early identification of malaria in the premature stage.