Explainable AI: A Combined XAI Framework for Explaining Brain Tumour Detection Models

Multiple explanation methods provide more complete understanding: The study combined three different XAI techniques (GRAD-CAM, LRP, and SHAP) rather than relying on a single method, with each technique offering different insights—from broad spatial regions to pixel-level details to quantified feature contributions. This layered approach revealed information that individual methods alone would have missed.

High accuracy alone isn't enough for medical AI: Although the model achieved 91.24% accuracy in detecting brain tumours, the researchers emphasized that understanding how the model makes decisions is equally important for clinical trust and adoption. Explainability is essential for doctors to validate whether the AI is reasoning correctly.

XAI can handle complex, ambiguous cases: The integrated approach successfully explained predictions even in challenging scenarios with partial tumour visibility, suggesting that explainable AI techniques can help clinicians understand model behavior in real-world situations where medical images may be unclear or incomplete.

Transparency builds reliability in high-stakes healthcare applications: By making the model's decision-making process visible and interpretable, this research demonstrates how XAI can increase clinician confidence in AI-driven medical imaging systems, which is critical for adoption in life-or-death applications like cancer detection.