Automated Malaria Detection using EfficientNetB3 in Deep Learning Frameworks
Goldy Verma · 2024
Particularly in tropical and subtropical areas, malaria is a major worldwide health concern that causes great mortality. Good disease control and prevention depend on accurate and prompt diagnosis. Historically, the diagnosis of malaria has depended on hand microscopic inspection of blood smears, a procedure prone to human error as well as time-consuming and labour-intensive nature. By means of a refined EfficientNetB3 model, this work intends to solve these difficulties by building an automated malaria diagnosis system. Renowned for their balanced scaling of depth, width, and resolution, the EfficientNetB3 architecture was refined to classify blood smear images into parasitized and uninfected groups. With an F1-score of 0.97 for both classes, the model showed great general accuracy of 97%, therefore demonstrating its dependability and resilience. Particularly in resource-limited environments, these findings show the promise of the customized EfficientNetB3 model to greatly increase malaria diagnosis, hence improving patient outcomes and supporting more efficient disease management.