Classification of Monkeypox Images Using LIME-Enabled Investigation of Deep Convolutional Neural Network.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computational Biology

Background:

• Monkeypox diagnosis is challenging due to symptom overlap with chickenpox and measles.
• Manual image analysis for monkeypox is time-consuming and error-prone.
• AI, particularly deep learning, shows promise for accurate disease detection from medical images.

Purpose of the Study:

• To develop and evaluate a Deep Convolutional Neural Network (CNN) based classification model for monkeypox detection.
• To differentiate monkeypox from similar conditions like chickenpox and measles using digital skin images.
• To improve the speed and accuracy of monkeypox diagnosis through an automated process.

Main Methods:

• Utilized several pre-trained CNN models (VGG16, VGG19, ResNet50, ResNet101, DenseNet201, AlexNet) for image classification.
• Employed an adaptive k-means clustering technique for precise image segmentation.
• Applied Local Interpretable Model-Agnostic Explanations (LIME) for feature extraction and model interpretability.

Main Results:

• The ResNet101 model achieved the highest classification accuracy of 94.25% with an Area Under the Curve (AUC) of 98.59%.
• The proposed model demonstrated effective differentiation between monkeypox and similar skin conditions.
• The AI-driven approach offers a reliable automated method for identifying monkeypox.

Conclusions:

• Deep Convolutional Neural Networks provide a robust framework for accurate monkeypox detection from skin lesions.
• Automated AI-based diagnosis can significantly aid clinicians in early and precise identification of monkeypox.
• Further research utilizing AI can enhance diagnostic capabilities for infectious diseases.