Malaria Detection from Blood-Cell Images using MobileNet Model: A Study
S. Audithan, K. Vijayakumar, P. Illavarason, S. Prabha
Abstract
Recently, the frequency of the disease's occurrence has been gradually increasing due to variety of factors. Malaria is one of the most prevalent vector-borne diseases that affect a huge number of people. The diagnosis of malaria is typically made using a clinical level examination that includes an appraisal of the symptoms, and blood test (blood-cell) to determine the extent of its severity. It is imperative that an individual receives proper therapy with anti-malarial medication in order to recover from the disease. In order to ensure the prompt detection of malaria, it is needed to automatically examine blood cells. As a result, this work proposes the development of a deep-learning tool for the detection of healthy blood cells and blood cells that are infected with the disease. After resizing the blood cell photos to the necessary dimensions, this operation took them into consideration for the study. The proposed project involved the implementation of a pre-trained model-based analysis that involved the examination of malaria from blood cell images by employing an individual-feature vector (IFV) and a fused-feature vector (FFV). The suggested method achieved an accuracy of >97% when using FFV and a SoftMax classifier.