From a graduation project at SBME to an international AI stage: students Youssef Afify, Alia Omar, Hamsa Saber, and Hamza Elghonemy have had their work on explainable AI for medical imaging accepted for oral presentation at EXPLIMED 2026, the Third Workshop on Explainable Artificial Intelligence for the Medical Domain, held alongside IJCAI-ECAI 2026 in Bremen, Germany.

Developed by SBME students Youssef Afify, Alia Omar, Hamsa Saber, and Hamza Elghonemy under the supervision of Dr. Tamer Basha, the project addresses one of the central challenges in medical artificial intelligence: making AI decisions transparent and clinically interpretable. The team also received technical guidance from Abdulrahman M. Selim, an SBME 2019 graduate, Omair Shahzad Bhatti, and Hasan Alam at the German Research Center for Artificial Intelligence (DFKI).

The team developed TRACE (Temporal Causal Reasoning with Acyclic Concept Explanation), an explainable AI framework for analyzing longitudinal medical images. Rather than producing a prediction directly from images as a black box, TRACE structures its reasoning through clinically meaningful concepts and rule-based pathways, allowing the factors behind a prediction to be examined and, when necessary, corrected by clinicians.

The project was extended into a web application that allows radiologists to explore longitudinal 3D MRI scans and segmentation overlays, inspect which clinical variables drove the model’s decision, and perform test-time interventions by correcting model-derived concepts and observing how those changes propagate to the final prediction.

EXPLIMED is a specialized international workshop dedicated to explainable artificial intelligence in medicine. Its 2026 edition is held in conjunction with IJCAI-ECAI 2026, bringing research on explainability, interpretability, and trustworthy AI into the broader international artificial intelligence research community. The department congratulates the students, their supervisor, and mentors on having their work selected for oral presentation at EXPLIMED 2026.

💻 Explore the Project, Poster & Presentation