DGG-XNet: A Hybrid Deep Learning Framework for Multi-Class Brain Disease Classification with Explainable AI

AI can improve brain disease diagnosis accuracy: DGG-XNet achieved 91.33% test accuracy in classifying multiple brain disorders from MRI scans, suggesting AI tools could help reduce errors in medical imaging analysis that currently relies on manual review.

Explainability matters in medical AI: The model uses Grad-CAM visualization to show doctors which regions of brain scans the AI focused on when making diagnoses, addressing a key concern that AI systems often function as "black boxes" in healthcare settings.

Combining different AI approaches yields better results: By merging two different neural network architectures (VGG16 and DenseNet121), the researchers achieved more reliable classifications than either approach alone, demonstrating that hybrid models may be more robust for medical applications.

High performance across multiple metrics indicates balanced reliability: The model maintained precision, recall, and F1-scores all above 91%, meaning it performs well at both correctly identifying diseases and avoiding false alarms—both critical for clinical use.

AI-assisted diagnosis could support (not replace) doctors: As a computer-aided diagnosis tool, DGG-XNet is positioned to assist radiologists rather than make autonomous decisions, which aligns with responsible AI practices that keep human experts in the decision-making loop for high-stakes medical applications.