Integrative Explanations Across Imaging and Clinical Data: Hierarchical XAI for Personalized Diagnostics

Current XAI work in imaging (e.g., Elayavalli et al., 2025; Elsharkawy et al., 2024) often isolates imaging from other clinical data, despite the known importance of patient history, labs, and genomic markers. This research would design an XAI system that fuses explanations across modalities: for example, it could show that “the model’s high suspicion of glioma is due to both the non-enhancing region on FLAIR and a history of IDH mutation.” Explanations would be structured hierarchically — imaging cues at the first level, integrated with clinical or molecular context at higher levels — and evaluated with both clinicians and patients for interpretability and clinical actionability. This advances the field by moving beyond unimodal, “flat” explanations and has the potential to make AI support both more accurate and more trusted at the point of care.

References:

1. A Clinically Explainable AI-Based Grading System for Age-Related Macular Degeneration Using Optical Coherence Tomography. M. Elsharkawy, A. Sharafeldeen, F. Khalifa, A. Soliman, A. Elnakib, M. Ghazal, A. Sewelam, Aristomenis Thanos, Harpal S. Sandhu, A. El-Baz (2024). IEEE journal of biomedical and health informatics.
2. Non-Invasive Molecular Classification of Gliomas Using Explainable ai and Radiomic Features from Multi-Sequence MRI. S. Elayavalli, Ipsa Yadav, Grant T. Cookson, Elias Sharifi, Tanya Tvrdic, Anant Madabhushi, Brent D. Weinberg, M. Abedalthagafi, S. Khalighi (2025). IEEE International Symposium on Biomedical Imaging.