Intelligent Systems for Early Malaria Disease Detection in Patient Cells Using Transfer Learning Approaches

Abstract

Malaria is an infectious disease spread by mosquitoes that effect humans and other animals. It is a massive threat to humanity, with instances growing each year. It is essential to prevent and diagnose malaria immediately and efficiently. For the time being, conventional methods are used for diagnosing malaria in which the patient's blood sample is examined by microscope or by using malaria RTD kits. This approach has several limitations because it requires medical expertise, is expensive, takes a long time, and the results are unsatisfactory. Artificial intelligence-based systems can prevent and help in diagnose of this infectious disease. Because of these limitations, the proposed work has proposed an AI-based diagnosis system that can detect malaria parasites immediately and efficiently. In the proposed experiment, we have applied four different pre-trained deep learning models on the image dataset with some preprocessing and optimization techniques for malaria parasite detection. After investigations, evaluation matrices such as precision, Recall, F1-score, sensitivity, and specificity are used to measure the performance of the proposed models. The Inception-Resnet outperformed by achieving 95% accuracy, VGG16 achieved 92% accuracy, inception achieved 93% accuracy, and VGG19 achieved 91% accuracy. The positive outcomes of this study show that this approach performs much better than the approaches currently used. Furthermore, the proposed method is relevant to health experts for screening purposes.