Diagnostic support of parasitic infections with an AI-powered microscope

Purpose: The unplanned expansions of urban and rural areas result in several environmental and population issues. Due to the lack of adequate sanitation infrastructure, these places’ population suffers from diseases caused by enteric parasites. Among these parasites, helminthiasis infections are the most common ones. The primary form of diagnosis of these diseases is a faecal examination of parasites using optical microscopy. In this sense, a fast, accurate, and affordable diagnosis is essential to treating patients with these infections.

Methods: A low-cost automated light microscope was produced using 3D-printed parts and off-shelf components capable of automatically scanning the microscopy slide. The images acquired are sent to a web server that applies a PSO-optimised AdaBoost classifier to identify, count, and classify the presence of Schistosoma mansoni eggs and three types of helminths in the images.

Results: The proposed solution was validated through experiments conducted in a laboratory under the supervision of specialised technicians and achieved 87% of global accuracy for detecting and counting parasites. While in similar experiments, technicians’ global accuracy was around 70%.

Conclusion: Compared to similar devices in the literature, the proposed solution has a reduced production cost compared to similar products and achieved satisfactory results. These characteristics are vital to implementing this technology, particularly among economically developing nations.