Capsule network approach for monkeypox (CAPSMON) detection and subclassification in medical imaging system.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Dermatology

Background:

• Accurate detection of Monkeypox virus (MPXV) is critical.
• Traditional Machine Learning and Deep Learning models have limitations in classifying complex dermatological conditions.
• Distinguishing visually similar skin conditions requires advanced image analysis techniques.

Purpose of the Study:

• To introduce the Enhanced Spatial-Awareness Capsule Network (ESACN) for precise multi-class classification of dermatological images.
• To address the shortcomings of existing models in differentiating conditions like monkeypox, chickenpox, and measles.
• To leverage Capsule Networks' spatial hierarchy for improved diagnostic accuracy.

Main Methods:

• Developed an Enhanced Spatial-Awareness Capsule Network (ESACN) architecture.
• Utilized dynamic routing and spatial hierarchy inherent to Capsule Networks (CapsNets).
• Applied the ESACN model to a dataset of 659 dermatological images across four classes: Monkeypox, Chickenpox, Measles, and Normal skin.

Main Results:

• ESACN demonstrated superior performance in differentiating complex and visually similar skin conditions.
• Significant improvements in accuracy, precision, recall, and F1 score were observed, even with limited data.
• The model achieved robust and accurate classification for Monkeypox, Chickenpox, Measles, and Normal skin presentations.

Conclusions:

• ESACN shows potential as a reliable tool for enhancing diagnostic accuracy in medical settings.
• The model's ability to process spatial relationships aids in distinguishing dermatological conditions.
• This approach can greatly aid in early diagnosis and treatment planning for skin diseases like Monkeypox.