Automated Malaria Detection Using Deep Learning and the Lacuna Dataset

Malaria is one of the primary health issues in resource-poor environments, and treatment for it is very time-sensitive and requires high accuracy in diagnosis. Toward this end, this paper relies on the Lacuna Malaria Detection Dataset of microscopic blood smear images annotated for the presence of malaria parasites to develop a convolutional neural network-based automatic malaria parasite detection framework. The dataset is a collection of thick and thin blood smear images, thus enabling the development of highly predictive models. It makes use of state-of-the-art techniques in image preprocessing, optimized CNN architectures, and rigorous evaluation metrics to ensure that the model developed will attain a high degree of accuracy and generalizability. The developed model yielded approximately 98% accuracy in detection and may potentially be useful for the accurate identification of the parasites. This study has placed a lot of attention on the practical application of AI diagnostic tools within health care, especially toward the under-resourced regions.