Malaria Detection Enhanced by Deep Learning: Implementing the VGG16 Architecture

Abstract

Malaria remains one of the deadliest diseases globally, presenting significant challenges to public health departments. Traditional diagnostic methods rely on the manual examination of blood smears by trained laboratory technicians, which can be inefficient and subject to the examiner's expertise. Although deep learning algorithms have been previously applied to the diagnosis of malaria from blood smears, their practical performance has not yielded satisfactory results. This study proposes a robust machine-learning model utilizing a convolutional neural network (CNN) to enhance the automatic classification and prediction of infected cells in thin blood smears on standard microscope slides. We implemented a ten-fold cross-validation technique using a dataset of 27,558 single-cell images to analyze cell parameters effectively. Three different CNN models—Basic CNN, VGG-16 Frozen CNN, and another variant—were evaluated in terms of accuracy. Through comparative analysis, the model demonstrating the highest accuracy was identified. The findings underscore the potential of advanced CNN architectures in improving malaria diagnostics, potentially leading to more reliable and efficient screening processes