Monkeypox Detection Using CNN with Transfer Learning.

Area of Science:

• Medical Informatics
• Computer Science
• Virology

Background:

• Monkeypox virus (MPXV) poses a public health threat, particularly following the COVID-19 pandemic.
• The need for rapid and accurate diagnostic tools for emerging infectious diseases is critical.
• Skin lesions are a primary clinical manifestation of monkeypox.

Purpose of the Study:

• To develop and evaluate a deep learning model for the fast and safe detection of monkeypox from skin lesions.
• To assess the efficacy of various deep learning architectures in classifying monkeypox images.
• To create a reliable diagnostic aid for potential monkeypox outbreaks.

Main Methods:

• Utilized deep learning, specifically Convolutional Neural Networks (CNNs), with transfer learning.
• Customized hybrid models incorporating hyperparameter optimization.
• Evaluated models including MobileNetV3-s, EfficientNetV2, ResNET50, Vgg19, DenseNet121, and Xception.
• Performance metrics included AUC, accuracy, recall, loss, and F1-score.

Main Results:

• The optimized hybrid MobileNetV3-s model demonstrated superior performance.
• Achieved an average F1-score of 0.98, AUC of 0.99, accuracy of 0.96, and recall of 0.97.
• The custom CNN model showed successful classification and discrimination capabilities.

Conclusions:

• Deep learning methods, particularly customized CNNs with transfer learning and hyperparameter optimization, are highly effective for monkeypox detection.
• The proposed model offers a rapid, safe, and accurate approach for diagnosing monkeypox.
• This technology can be crucial for managing future viral disease outbreaks.