Using novel deep learning models for rapid and efficient assistance in monkeypox screening from skin images

Area of Science:

• Medical Diagnostics
• Artificial Intelligence
• Virology

Background:

• Monkeypox is a viral disease transmitted via direct contact with infectious lesions.
• Current diagnostic methods are time-consuming, labor-intensive, and pose infection risks.
• There is a critical need for automated, efficient auxiliary diagnostic tools.

Purpose of the Study:

• To develop and evaluate a deep learning classification model for monkeypox diagnosis.
• To enhance diagnostic accuracy and model generalizability using advanced deep learning techniques.
• To provide an automated solution for monkeypox diagnosis, especially in resource-limited settings.

Main Methods:

• Developed a deep learning classification model incorporating self-attention mechanisms and feature pyramid integration.
• Utilized attentional strategies to combine image features across multiple scales.
• Integrated prior knowledge from the VGG model to capture salient features.
• Conducted ablation studies to assess the contribution of individual model components.

Main Results:

• The proposed deep learning model achieved the highest accuracy and precision compared to state-of-the-art models.
• Demonstrated a 6% improvement in accuracy and precision over the second-best model.
• Ablation experiments confirmed the effectiveness of each module in enhancing performance.

Conclusions:

• The developed deep learning model offers improved diagnostic precision for monkeypox.
• This automated approach can extend the reach of medical services in resource-constrained settings.
• The method shows significant potential as an auxiliary diagnostic tool for monkeypox.