skip to main content


This content will become publicly available on December 1, 2024

Title: Classifying stages in the gonotrophic cycle of mosquitoes from images using computer vision techniques
Abstract

The ability to distinguish between the abdominal conditions of adult female mosquitoes has important utility for the surveillance and control of mosquito-borne diseases. However, doing so requires entomological training and time-consuming manual effort. Here, we design computer vision techniques to determine stages in the gonotrophic cycle of female mosquitoes from images. Our dataset was collected from 139 adult female mosquitoes across three medically important species—Aedes aegypti,Anopheles stephensi, andCulex quinquefasciatus—and all four gonotrophic stages of the cycle (unfed, fully fed, semi-gravid, and gravid). From these mosquitoes and stages, a total of 1959 images were captured on a plain background via multiple smartphones. Subsequently, we trained four distinct AI model architectures (ResNet50,MobileNetV2,EfficientNet-B0, andConvNeXtTiny), validated them using unseen data, and compared their overall classification accuracies. Additionally, we analyzed t-SNE plots to visualize the formation of decision boundaries in a lower-dimensional space. Notably,ResNet50andEfficientNet-B0demonstrated outstanding performance with an overall accuracy of 97.44% and 93.59%, respectively.EfficientNet-B0demonstrated the best overall performance considering computational efficiency, model size, training speed, and t-SNE decision boundaries. We also assessed the explainability of thisEfficientNet-B0model, by implementing Grad-CAMs—a technique that highlights pixels in an image that were prioritized for classification. We observed that the highest weight was for those pixels representing the mosquito abdomen, demonstrating that our AI model has indeed learned correctly. Our work has significant practical impact. First, image datasets for gonotrophic stages of mosquitoes are not yet available. Second, our algorithms can be integrated with existing citizen science platforms that enable the public to record and upload biological observations. With such integration, our algorithms will enable the public to contribute to mosquito surveillance and gonotrophic stage identification. Finally, we are aware of work today that uses computer vision techniques for automated mosquito species identification, and our algorithms in this paper can augment these efforts by enabling the automated detection of gonotrophic stages of mosquitoes as well.

 
more » « less
Award ID(s):
2014547
NSF-PAR ID:
10545412
Author(s) / Creator(s):
; ; ; ; ; ; ;
Publisher / Repository:
Springer Nature
Date Published:
Journal Name:
Scientific Reports
Volume:
13
Issue:
1
ISSN:
2045-2322
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. Abstract Background

    Culexmosquitoes are the primary vectors of West Nile virus (WNV) across the USA. Understanding when these vectors are active indicates times when WNV transmission can occur. This study determined the proportion of femaleCulexmosquitoes that were in diapause during the fall and winter and when they terminated diapause and began blood feeding in the spring.

    Methods

    Mosquitoes were collected from parks using various traps and/or aspirated from culverts in Franklin County, Ohio, from October to mid-May from 2019 to 2022.Culexmosquitoes were morphologically identified to species, and the ovaries of females were dissected to determine their diapause and parity statuses.

    Results

    By early October 2021, roughly 95% ofCulex pipienscollected in culverts were in diapause and 98% ofCx. erraticuswere in diapause. Furthermore, gravid and blood-fedCulex salinarius,Cx. pipiens, andCx. restuanswere collected in late November in 2019 and 2021 in standard mosquito traps. In the winter of 2021, the proportions of non-diapausingCulexdecreased within culverts. The last non-diapausingCx. erraticuswas collected in late December 2021 while the final non-diapausingCx. pipienswas collected in mid-January 2022, both in culverts. Roughly 50% ofCx. pipiensterminated diapause by mid-March 2022, further supported by our collections of gravid females in late March in all 3 years of mosquito collection. In fact, male mosquitoes ofCx. pipiens,Cx. restuans, andCx. territanswere collected by the 1st week of May in 2022, indicating that multiple species ofCulexproduced a second generation that reached adulthood by this time.

    Conclusions

    We collected blood-fed and gravidCulexfemales into late November in 2 of the 3 years of our collections, indicating that it might be possible for WNV transmission to occur in late fall in temperate climates like Ohio. The persistence of non-diapausingCx. pipiensandCx. erraticusthroughout December has important implications for the winter survival of WNV vectors and our overall understanding of diapause. Finally, determining whenCulexterminate diapause in the spring may allow us to optimize mosquito management programs and reduce the spread of WNV before it is transmitted to humans.

    Graphical Abstract 
    more » « less
  2. Abstract Background

    West Nile virus (WNV), primarily vectored by mosquitoes of the genusCulex, is the most important mosquito-borne pathogen in North America, having infected thousands of humans and countless wildlife since its arrival in the USA in 1999. In locations with dedicated mosquito control programs, surveillance methods often rely on frequent testing of mosquitoes collected in a network of gravid traps (GTs) and CO2-baited light traps (LTs). Traps specifically targeting oviposition-seeking (e.g. GTs) and host-seeking (e.g. LTs) mosquitoes are vulnerable to trap bias, and captured specimens are often damaged, making morphological identification difficult.

    Methods

    This study leverages an alternative mosquito collection method, the human landing catch (HLC), as a means to compare sampling of potential WNV vectors to traditional trapping methods. Human collectors exposed one limb for 15 min at crepuscular periods (5:00–8:30 am and 6:00–9:30 pm daily, the time whenCulexspecies are most actively host-seeking) at each of 55 study sites in suburban Chicago, Illinois, for two summers (2018 and 2019).

    Results

    A total of 223 human-seeking mosquitoes were caught by HLC, of which 46 (20.6%) were mosquitoes of genusCulex. Of these 46 collectedCulexspecimens, 34 (73.9%) wereCx. salinarius, a potential WNV vector species not thought to be highly abundant in upper Midwest USA. Per trapping effort, GTs and LTs collected > 7.5-fold the number of individualCulexspecimens than HLC efforts.

    Conclusions

    The less commonly used HLC method provides important insight into the complement of human-biting mosquitoes in a region with consistent WNV epidemics. This study underscores the value of the HLC collection method as a complementary tool for surveillance to aid in WNV vector species characterization. However, given the added risk to the collector, novel mitigation methods or alternative approaches must be explored to incorporate HLC collections safely and strategically into control programs.

    Graphical Abstract 
    more » « less
  3. Christofferson, Rebecca C (Ed.)
    Background

    Anopheles stephensiis an invasive malaria vector in Africa that threatens to put an additional 126 million people at risk of malaria if it continues to spread. The island nation of Mauritius is highly connected to Asia and Africa and is at risk of introduction due to this connectivity. For early detection ofAn.stephensi, the Vector Biology and Control Division under the Ministry of Health in Mauritius, leveraged a well-establishedAedesprogram, asAn.stephensiis known to shareAedeshabitats. These efforts triggered multisectoral coordination and cascading benefits of integrated vector and One Health approaches.

    Methods

    Beginning June 2021, entomological surveys were conducted at points of entry (seaport, airport) and on ships transporting livestock in collaboration with the Civil Aviation Department, the Mauritian Port Authority and National Veterinary Services.

    A total of 18, 39, 723 mosquito larval surveys were respectively conducted in the airport, seaport, and other localities in Mauritius while two, 20, and 26 adult mosquito surveys were respectively conducted in the airport, seaport, and twenty-six animal assembly points. Alongside adult mosquito surveys, surveillance of vectors of veterinary importance (e.g.-Culicoidesspp.) was also carried out in collaboration with National Parks and Conservation Service and land owners.

    Results

    A total of 8,428 adult mosquitoes were collected and 1,844 larval habitats were positive for mosquitoes. All collected mosquitoes were morphologically identified and 151Anophelesand 339Aedesmosquitoes were also molecularly characterized. Mosquito species detected wereAedes albopictus,Anopheles arabiensis,An.coustani,An.merus,Culex quinquefasciatus,Cx.thalassiusandLutzia tigripes.Anopheles stephensiwas not detected. The One Health approach was shared with the French Agricultural Research Centre for International Development (CIRAD), strengthening collaboration between Mauritius and Réunion Island on vector surveillance at entry points and insecticide resistance monitoring. The Indian Ocean Commission (IOC) was also alerted to the risk ofAn.stephensi, leading to regional efforts supporting trainings and development of a response strategy toAn.stephensibringing together stakeholders from Comoros, Madagascar, Mauritius, Réunion Island and Seychelles.

    Conclusions

    Mauritius is a model system showing how existing public health entomology capabilities can be used to enhance vector surveillance and control and create multisectoral networks to respond to any emerging public and veterinary health vector-borne disease threat.

     
    more » « less
  4. Perkins, Alex (Ed.)
    Mosquitoes vector harmful pathogens that infect millions of people every year, and developing approaches to effectively control mosquitoes is a topic of great interest. However, the success of many control measures is highly dependent upon ecological, physiological, and life history traits of mosquito species. The behavior of mosquitoes and their potential to vector pathogens can also be impacted by these traits. One trait of interest is mosquito body mass, which depends upon many factors associated with the environment in which juvenile mosquitoes develop. Our experiments examined the impact of larval density on the body mass of Aedes aegypti mosquitoes, which are important vectors of dengue, Zika, yellow fever, and other pathogens. To investigate the interactions between the larval environment and mosquito body mass, we built a discrete time mathematical model that incorporates body mass, larval density, and food availability and fit the model to our experimental data. We considered three categories of model complexity informed by data, and selected the best model within each category using Akaike’s Information Criterion. We found that the larval environment is an important determinant of the body mass of mosquitoes upon emergence. Furthermore, we found that larval density has greater impact on body mass of adults at emergence than on development time, and that inclusion of density dependence in the survival of female aquatic stages in models is important. We discuss the implications of our results for the control of Aedes mosquitoes and on their potential to spread disease. 
    more » « less
  5. ABSTRACT Mosquito surveillance is critical to reduce the risk of West Nile virus (WNV) transmission to humans. In response to surveillance indicators such as elevated mosquito abundance or increased WNV levels, many mosquito control programs will perform truck-mounted ultra-low volume (ULV) adulticide application to reduce the number of mosquitoes and associated virus transmission. Despite the common use of truck-based ULV adulticiding as a public health measure to reduce WNV prevalence, limited evidence exists to support a role in reducing viral transmission to humans. We use a generalized additive and fused ridge regression model to quantify the location-specific impact of truck-mounted ULV adulticide spray efforts from 2010 to 2018 in the North Shore Mosquito Abatement District (NSMAD) in metropolitan Chicago, IL, on commonly assessed risk factors from NSMAD surveillance gravid traps: Culex abundance, infection rate, and vector index. Our model also takes into account environmental variables commonly associated with WNV, including temperature, precipitation, wind speed, location, and week of year. Since it is unlikely ULV adulticide spraying will have the same impact at each trap location, we use a spatially varying spray effect with a fused ridge penalty to determine how the effect varies by trap location. We found that ULV adulticide spraying has an immediate temporary reduction in abundance followed by an increase after 5 days. It is estimated that mosquito abundance increased more in sprayed areas than if left unsprayed in all but 3 trap locations. The impact on infection rate and vector index were inconclusive due to the large error associated with estimating trap-specific infection rates. 
    more » « less