Work

DMLMD I dataset for machine learning based malaria detection using blood smears

Year: 2026

Type: article

Abstract: The fight against malaria, a life-threatening disease transmitted by Anopheles mosquitoes, relies heavily on accurate and timely diagnosis. Therefore, it is imperative to develop and augment a thin blood smear dataset for malaria detection using deep learning (DL), aiming to diagnose the malaria and reduce the burden on healthcare systems. The original dataset (primary dataset) comprised 631 high-resolution blood smear images representing five stages of Plasmodium parasite infection: Multi Ring Stage (M-Rg), Multi Trophozoite Stage (M-Trp), Ring Stage (Rg), Schizont Stage (Sz), and Trophozoite Stage (Trp). Image augmentation techniques were applied to expand the dataset, increasing it from 631 to 9314 images (reproducible). The enhanced dataset can be used for training AI models for bounding-box object detection of intraerythrocytic Plasmodium stages i.e.M-Rg, M-Trp, Rg, Sz,Trp, not binary cell-level classification (infected vs. uninfected). The augmented dataset provides a robust resource for training DL models. These models will demonstrate the ability to automatically detect malaria parasites stages by identifying subtle patterns in blood smears, thus can reduce human error and improving diagnostic precision. The automated system will be capable of offering scalable diagnostic solutions for resource-limited settings.The augmented malaria dataset will facilitate the development of AI-powered, real-time diagnostic tools, that will enhance the accuracy and accessibility of malaria detection in under-resourced regions. This contributes to more efficient disease management and control.

Source: Discover Informatics

DOI: 10.1007/s44564-026-00036-9