Automated digital microscopy using artificial intelligence for the point-of-care malaria diagnosis 

Besides the very frequent use of rapid diagnostic tests (RDTs) for malaria diagnosis in the field, expert microscopy is still used for reliable diagnosis. In contrast to microscopy, RDTs are unable to quantify parasitemia, identify gametocytes, and cannot distinguish all malaria species. Furthermore, the rapid spread of Plasmodium parasites having the hrp2 / hrp3 gene deleted compromises the reliable use of RDTs. Here we present an innovative platform that is designed to be employed in endemic areas that automatically generates microscopic blood smears termed miLab. These smears are liquid-free stained using a gel-based stamping technology coupled with automatic digital microscopy. All images captured are being analysed by a machine-learned algorithm to quantify and identify malaria parasites. We have validated the instrument on full human blood samples spiked with various numbers of synchronized and non-synchronized cultured P. falciparum parasites. In addition, we have used cultures and induced gametocytes for training the algorithm to also detect gametocytes. After training and validation the platform was employed in a clinical field study in Lilongwe, Malawi in which collected clinical samples were analysed. For further performance tests of the miLab platform, dried blood spots were collected on filter paper for subsequent quantitative PCR and slides were prepared for expert microscopy testing at Swiss TPH. We will present and describe the platform, its performance characteristics and initial results from the field study conducted in Malawi.