CSE Researchers Design AI Algorithms to Automate Vector Surveillance
August 3, 2020
CSE doctoral student Mona Minakshi and her advisor Sriram Chellappan co-authored a paper that is accepted for publication in Nature Scientific Reports. The paper is titled - A Framework based on Deep Neural Networks to Extract Anatomy of Mosquitoes from Images.
The broad goal of the study is to design AI algorithms that can process images of mosquito vectors (those that can spread diseases), and extract their thorax, abdomen, wings and legs separately. Their AI algorithms are based on state of the art deep neural networks. As of today, across the globe, countless hours are being spent by expert taxonomists in identifying vectors from non-vectors. “The AI algorithms designed in our paper will enable the vector surveillance process to be automated, and can could usher in faster and low-cost risk prediction models” says the authors. A salient feature of their dataset is that it contains hundreds of images of Aedes aegypti mosquitoes – a particularly deadly vector for many diseases including dengue, chikungunya and Zika fever.
Mona Minakshi also said that “This project was an eye opening experience for me, considering the global impact of identifying mosquito vectors, and associated challenges”. She also added that “my favorite part of participation in this effort was the significant number of researchers I interacted with across a diverse spectrum including, entomology, epidemiology and public health”. She and her advisor appreciated support provided by Hillsborough county mosquito control district in helping them trap and manually identify mosquito specimens.