A multidisciplinary team involving the Microbiology Service of the Vall d’Hebron University Hospital, the Microbiology research group of the Vall d’Hebron Research Institute (VHIR), the Universitat Politècnica de Catalunya - BarcelonaTech (UPC) and the Probitas Foundation introduced a novel AI-based diagnostic method for malaria.

The research was primarily conducted at the microbiology laboratory of the Drassanes Vall d’Hebron Centre for International Health, in collaboration with the UPC’s Computational Biology and Complex Systems Group (BIOCOM-UPC), Image and Video Processing Group (GPI) and Database Technologies and Information Management Group (). It is a system based on artificial intelligence that combines a mobile application with a low-cost robotised microscope. Its design aims to provide a practical and effective method, particularly suitable for countries with limited resources where the disease is endemic. The results of the first iMAGING prototype have been published in the Frontiers in Microbiology journal. The system has demonstrated a reliability of over 90% in the laboratory. The next step will be to test it on site.

Malaria is an infectious disease caused by parasites of the genus Plasmodium that are transmitted through mosquito bites. According to the World Health Organization (WHO) report, in 2022 there were an estimated 249 million malaria cases globally, with the African region being home to 93% cases and 95% of deaths. The report also warned that climate change and globalisation are contributing to the spread of mosquitoes into new regions with insufficient preparedness and resources to deal with it. The gold-standard method for malaria diagnosis is still the microscopic visualisation by an expert of parasites in blood samples. It is a manual, long and repetitive procedure, which, coupled with the lack of laboratory staff and instruments, leads to significant underdiagnosis. So far, any attempt to automate the process increased its cost exponentially, making it unaffordable for countries with few healthcare resources.

An automatic microscope controlled via Bluetooth
The solution proposed is iMAGING, a mobile application that uses artificial intelligence to process digital images of blood samples to determine the presence or absence of infection. If infection is confirmed, it also determines its density and stage. To capture images, a robotised microscope has been created from a regular optical microscope with 3D printed parts, significantly reducing its cost.

The app connects via Bluetooth to the microscope and controls its movements and focusing to automatically analyse the sample and obtain the images needed for diagnosis. Technical staff only need to prepare the samples, which greatly reduces their workload and the possibility of errors.

The prototype was trained on more than 2,500 images and achieved a reliability of over 96% on high-density samples and 94% on low-density samples. False positives and negatives did not exceed 5% in any case. However, Dr Joan Joseph i Munné, the principal investigator and a researcher at the VHIR’s Microbiology group, explains that “the acid test will be operation on site, but if successful, it can pave the way for adaptation to other neglected tropical diseases.”

The project is part of the science and technology work for human development driven by the UPC’s Centre for Development Cooperation (CCD). BIOCOM-UPC researcher Daniel López Codina explains that “we need to continue working to develop quality and low-cost tools to improve the health of people living in countries with a low or very low human development index. We are very pleased with the results achieved so far and we are confident that we can adapt the tools developed for other neglected tropical diseases.”

Now, the plan is to continue training the artificial intelligence to introduce improvements in other areas, for example, to differentiate between the five different species of parasites causing the illness. This will allow for much more personalised treatment, enhancing its effectiveness. The project is supported by WHO as part of its initiative for diagnosis through digital image of hemoparasites in low- and middle-income countries.

In addition to López Codina, other UPC researchers participating in the iMAGING project are Allisson Dantas de Oliveira and Clara Prats from the BIOCOM-UPC; Sergi Nadal, Besim Bilalli and Alberto Abelló from the DTIM research group; and first author Carles Rubio from the VHIR. The project also featured Elisa Sayrol, a researcher from the GPI research group at the time of the project.