A novel multimodal deep learning architecture integrating a multi-scale gated residual block and soft attention for skin lesion classification

Key Points

• Improved skin lesion classification accuracy indicates the model's potential effectiveness in clinical settings.
• An accuracy rate of 95% for skin lesion classification was achieved using the proposed model.
• Multimodal deep learning architecture with multi-scale gated residual blocks was employed for classifications.
• The findings highlight the potential for advanced neural networks to aid in dermatological diagnostics.