PICTURE: Representative sample image of Anopheles stephensi from the Mosquito Image Database. The picture shows a mosquito that was frozen at -80 ° C. view More
Photo credit: Couret, J. et al. 2020 (CC-BY 2.0)
Quickly and accurately identifying mosquitoes carrying human pathogens such as malaria is an essential part of monitoring mosquito-borne diseases. Researchers reporting in PLOS Neglected Tropical Diseases have now shown the effectiveness of an artificial intelligence system – known as the Convoluted Neural Network – for classifying the gender, genus, species and strain of mosquitoes.
Human malaria is an ongoing public health crisis that affects multiple continents, with most cases and people at risk occurring in sub-Saharan Africa. Identifying mosquitoes that transmit malaria can be difficult, however – some species are difficult to distinguish, even for trained taxonomists.
In the new work, Jannelle Couret of the University of Rhode Island, USA, and colleagues applied a Convoluted Neural Network (CNN) to a library of 1,709 two-dimensional images of adult mosquitoes. The mosquitoes were collected from 16 colonies in five geographic regions and included a species that was not easily identified by trained medical entomologists. This included mosquitos stored in two different ways – by flash freezing or as dried samples.
Using the library of identified species, researchers trained the CNN to differentiate Anopheles from other mosquito genera, identify species and sex within Anopheles, and identify two phyla within a single species. They found a prediction accuracy of 99.96% for the class and an accuracy of 98.48% for the gender.
“These results show that image classification using deep learning can be a useful method for identifying malaria mosquitoes, even in species with cryptic morphological variation,” the researchers say. “Developing an independent and accurate method of identifying species can potentially improve mosquito surveillance practices.”
Peer review; Simulation / modeling
Use this URL in your reporting to get access to the freely available paper: http://journals.plos.org/plosntds/article?id=10.1371/journal.pntd.0008904
Citation: Couret J., Moreira DC, Bernier D., Loberti AM, Dotson EM, Alvarez M. (2020) Delimitation of the cryptic morphological variation between human malaria vector species using convolutional neural networks. PLoS Negl Trop Dis 14 (12): e0008904. https://doi.org/10.1371/journal.pntd.0008904
Funding: This work was supported by USDA National Food and Agriculture Institute, Hatch Regional Project 1021058. Funders had no role in the design of the study, data collection and analysis, the decision to publish, or the preparation of the manuscript.
Competing interests: The authors have stated that there are no competing interests.
Disclaimer: AAAS and EurekAlert! are not responsible for the correctness of the press releases published on EurekAlert! by contributing institutions or to use information through the EurekAlert system.