Detailed Briefing Document: Application of Wing Interference Patterns (WIPs) and Deep Learning (DL) for Culex spp. Classification

Application of wings interferential patterns (WIPs) and deep learning (DL) to classify some Culex. spp (Culicidae) of medical or veterinary importance
Arnaud Cannet, Camille Simon Chane, Aymeric Histace, Mohammad Akhoundi, Olivier Romain, Pierre Jacob, Darian Sereno, Marc Souchaud, Philippe Bousses & Denis Sereno
Scientific Reports volume 15, Article number: 21548 (2025)
Source: https://doi.org/10.1038/s41598-025-08667-y
Received – 28 November 2024 | Accepted – 23 June 2025 | Published – 01 July 2025

This briefing document reviews a study that successfully demonstrates the utility of combining Wing Interference Patterns (WIPs) with deep learning (DL) models for the accurate identification of Culex mosquito species. Culex mosquitoes are significant vectors for numerous arboviruses and parasites of medical and veterinary importance, including West Nile virus, Japanese encephalitis, Saint Louis encephalitis, and lymphatic filariasis. Traditional morphological identification methods are labor-intensive, prone to errors due to cryptic species or damaged samples, and often yield variable accuracy (e.g., ~64% average species-level accuracy in external assessments).

The research team developed a method leveraging the unique, stable interference patterns visible on transparent insect wing membranes (WIPs) as species-specific morphological markers. By integrating these WIPs with Convolutional Neural Networks (CNNs), the study achieved over 95% genus-level accuracy for Culex and up to 100% species-level accuracy for certain species. While challenges remain with underrepresented species in the dataset, this approach presents a scalable, cost-effective, and robust alternative or complement to traditional identification methods, with significant potential for enhancing vector surveillance and global health initiatives.

Key Themes and Important Ideas/Facts

1. The Challenge of Mosquito Identification and its Importance

• Global Health Threat: Arthropod-transmitted pathogens, including viruses, bacteria, and parasites, are “among the most destructive infectious agents globally.”
• Vector Role of Culex: The Culex genus, comprising over 783 recognized species and 55 subspecies, “are recognized vectors of significant diseases, such as West Nile virus fever, Japanese encephalitis, Saint Louis encephalitis, or lymphatic filariasis.”
• Difficulty of Traditional Methods: “Traditional morphological identification is labor-intensive and relies on diagnostic features and determination keys.” This method is “often challenged by cryptic species, overlapping morphological traits, and damaged specimens.”
• Need for Innovation: These limitations “emphasize the need for innovative identification methods to enhance entomological surveys.”

2. Wing Interference Patterns (WIPs) as Species-Specific Markers

• Nature of WIPs: WIPs are “visible color patterns caused by thin-film interference” on the thin, transparent wing membranes of insects, particularly smaller species. They become visible when wings are “illuminated in a dark, light-absorbing setting.”
• Species-Specific Consistency: “These Wing Interference Patterns (WIPs) show substantial variation between different species, while remaining relatively consistent within a species or between sexes.”
• Stability of WIPs: Unlike conventional iridescence, the “microstructure of insect wings functions as a dioptric system that stabilizes the interference pattern, making WIPs largely insensitive to viewing angle.”
• Potential as Morphological Markers: Due to their “species-specific consistency and interspecific variability, WIPs hold strong potential as morphological markers for insect classification, offering a promising alternative or complement to traditional taxonomic traits.”

3. Integration of WIPs with Deep Learning (DL) for Classification

• Previous Successes: WIPs and DL have previously “successfully demonstrated their utility in identifying Anopheles, Aedes, sandflies, and tsetse flies.” This study extends the approach to Culex.
• Methodology: The study applied “WIPs, generated by thin-film interference on wing membranes, in combination with convolutional neural networks (CNNs) for species classification.”
• CNN Advantages: Deep Convolutional Neural Networks (CNNs) are “most effective for image classification” and “automatically selects the optimal features during the learning process, making it particularly suitable for WIP classification tasks.”
• Dataset: The study used a refined dataset of “553 images representing WIPs from 7 species” for training, with a larger database including “572 images of 12 species across 5 subgenera” for general classification and 4,944 images of non-Culex Diptera as negative controls.

4. Classification Performance and Results

• High Genus-Level Accuracy: The CNN achieved “genus-level classification accuracy exceeding 95.00%.”
• Variable Species-Level Accuracy: “At the species level, performance varied, with perfect accuracy (100.00%) for Cx. neavei and high accuracy (75.00% to 94.00%) for Cx. insignis, Cx. quinquefasciatus, and Cx. tritaeniorhynchus.”
• Comparison to Traditional Methods: The CNN-based method’s species-level accuracy “ranging from 40 to 100%” can “surpass the performance of morphological identification reported in that assessment,” where trained entomologists achieved “an average species-level accuracy of 64%.”
• Challenges with Underrepresented Species: “Misclassification occurred for Cx. thalassius and (accuracy 0.00%), while low 40.00% or moderate accuracy (50.00%) were recorded for Cx. univittatus and Cx. nebulosus respectively.” This variation is attributed to “dataset limitations, particularly for poorly represented species.”

5. Future Directions and Implications

• Enhancing Robustness: “Expanding the dataset to include more specimens and diverse conditions–such as age, preservation state, and environmental origin–could improve accuracy.”
• Complementary Techniques: “Integrating complementary techniques like molecular barcoding or protein profiling can enhance dataset robustness and address cryptic species identification.”
• Standardization: “Standardizing imaging protocols is essential to minimize variability and ensure consistent image quality.”
• Evaluation Criteria: “Establishment of standardized criteria for evaluating the accuracy of AI/ML-based mosquito identification systems” is crucial for reliable benchmarking.
• Scalability and Cost-Effectiveness: The “scalable and cost-effective nature of WIP imaging makes it suitable for large-scale biodiversity monitoring and entomological surveys.”
• Transformative Potential: The method “has the potential to become a transformative tool for vector surveillance and biodiversity research, advancing global health and ecological conservation efforts.”
• Broader Applicability: This approach aims to be “generalized to survey a broader range of Dipteran insects of major relevance to human health,” supporting “more efficient species identification at broader scales.”

Conclusion

The study successfully demonstrates that the fusion of Wing Interference Patterns (WIPs) and deep learning provides a promising and accurate tool for identifying Culex mosquitoes, a critical step in controlling vector-borne diseases. Despite current lim…