Funded by a grant from the University of Kansas, The Virtual Vector Laboratory set out to accomplish the ambitious goal of developing a system involving automated classification software that could process photographs of bugs that transmit diseases and determine which species of bug was in the photo.

 

Abstract

Identification of arthropods important in disease transmission is a crucial, yet difficult, task that can demand considerable training and experience. An important case in point is that of the 150+ species of Triatominae, vectors of Trypanosoma cruzi, causative agent of Chagas disease across the Americas.

We've created a fully automated prototype system that is able to identify Triatomine bugs from Mexico and Brazil with accuracy consistently above 80% (with several species at or near 100%), and with considerable potential for further improvement. The system instantly processes digital photographs submitted from the field into landmarks and uses ratios of measurements among those landmarks as the basis for classification. 

Species identification is important for generating accurate maps that can pinpoint with fine resolution where various species live, where they may be moving next and which human populations are most at risk of infection. Armed with this knowledge, customized strategies can be implemented by public health workers that are tailored to the specific species, environment and population in any given circumstance.

Our portable specimen photo devices are being used in the field by researchers across the Americas to upload images for instant processing by VVL software.

Our portable specimen photo devices are being used in the field by researchers across the Americas to upload images for instant processing by VVL software.

What We've Achieved

  • We designed an apparatus that permits taking consistent, high-quality, repeatable photographs of Triatomines. 
  • Initial photo devices were distributed to researchers in Brazil and Mexico to amass a large reference collection of Triatomine bug photographs. Following that success more are now being delivered throughout Latin America and the U.S.
  • We developed our prototype classification software capable of automatically distinguishing which species is represented in the photographs with an average accuracy rate above 80% with several species at or near 100%. The accuracy continues to increase as we get more samples.
  • We conducted field research interviews of at risk populations in Mexico to determine awareness of Triatomine bugs and their potential transmission of disease.
  •  Documentary videos are being produced for dispersal throughout the Americas to help raise awareness of our project and educate people about Chagas disease.
  • We've strategized next steps that would see our software developed into an application for mobile devices and thereby allow VVL to directly connect with millions of potential collaborators across the Americas allowing them to contribute images to the growing map as well as connect to health resources.
The greatest value of our work will be realized when we can help reduce human misery caused by vector born disease through better mapping and increased awareness and education of at risk populations.
— Town Peterson, Biodiversity Institute, University of Kansas