Vector mosquito image classification using novel RIFS feature selection and machine learning models for disease...

Area of Science:

• Medical Entomology
• Computer Science
• Epidemiology

Background:

• Mosquito-borne diseases cause approximately one million deaths annually.
• Preventing mosquito bites is crucial for disease control.
• Accurate identification of disease-vectoring mosquito species is essential for public health.

Purpose of the Study:

• To develop and evaluate a Machine Learning (ML) and Deep Learning (DL) system for classifying Aedes and Culex mosquitoes.
• To aid epidemiological efforts in risk assessment and policy design for mosquito-borne disease control.
• To introduce a novel feature selection method, RIFS, integrating ROI-based image filtering and FFS techniques.

Main Methods:

• Utilized ML and Convolutional Neural Network (CNN) models for mosquito classification.
• Introduced the RIFS (Region of Interest-based Image Filtering and Wrapper-based FFS) technique for feature selection.
• Performed comparative analysis of various ML and DL models based on performance metrics and computational efficiency.

Main Results:

• The ETC model achieved 0.992 accuracy among ML models.
• The VGG16 model achieved 0.986 accuracy among CNN models.
• The RIFS technique enhanced the classification performance.

Conclusions:

• ML and DL models, particularly ETC and VGG16, show high accuracy in classifying disease-carrying mosquito species.
• The proposed system and RIFS technique offer a valuable tool for epidemiological surveillance and disease prevention.
• Accurate mosquito classification supports evidence-based public health policies to mitigate mosquito-borne disease transmission.