Sejong News
NOW Sejong #180: Meet Our Professor – An Interview with Professor Mugahed A. Al-antari from the Department of Artificial Intelligence and Data Science
오세혁
2026-04-20
Professor Mugahed A. Al-antari from the Department of Artificial Intelligence and Data Science conducts research on diagnostic systems aimed at improving the reliability of medical AI, while actively collaborating with the global medical AI community. We met with Professor Mugahed A. Al-antari to learn more about his research direction and the future of medical AI.
Q. Please introduce yourself.
A. I am Mugahed A. Al-antari, a professor in the Department of Artificial Intelligence and Data Science at Sejong University. My research focuses on AI-based healthcare and medical imaging, and I lead the AISSLab (Artificial Intelligence Systems and Science Lab), where we work on solving medical challenges and advancing AI technologies.
Q. We understand that the Department of Artificial Intelligence and Data Science was formed by merging the AI and Data Science departments.
A. The Department of Artificial Intelligence and Data Science was established in 2024 through the integration of the AI and Data Science departments, with the goal of combining AI technologies and data analytics in education. The curriculum covers key areas such as machine learning, deep learning, big data processing, statistical analysis, and data-driven decision-making, enabling students to develop problem-solving skills and apply them across various domains.
Q. Could you explain the AISSLab team you are currently leading?
A. AISSLab develops cutting-edge diagnostic frameworks in medical AI, aiming to automate and enhance clinical decision-making. We focus on efficiently processing complex 3D medical images and generating evidence-based clinical reports. Our key systems include MedXRay-CAD for respiratory disease diagnosis and AutoSpineAI for lumbar spinal stenosis diagnosis. The team consists of multinational researchers, which provides a strong global research perspective and is one of our core strengths.
Q. When do you feel most rewarded while mentoring students?
A. I feel most proud when I see our students’ research being recognized both domestically and internationally. One memorable example is when our AISSLab team received the Best Workshop Paper Award at MICCAI 2025 for MedXpert-CAD, a multimodal agent AI system integrating X-ray and MRI analysis. It was especially meaningful as I had been involved throughout the entire research process, and seeing their hard work turn into achievements was truly rewarding.
Q. Your recent papers in IEEE JBHI and IEEE-EMBS BHI applied advanced RAG techniques to medical imaging AI. How did you technically address AI hallucinations, one of the most critical challenges in clinical settings?
A. AI hallucination is one of the most critical issues in medical applications. We addressed this by employing LLM-based RAG techniques that ensure all outputs are grounded in verifiable clinical evidence.
The MedXRay-CAD system, published in IEEE JBHI, enhances diagnostic accuracy and reliability across various respiratory diseases by combining text-based reasoning with visual explanations. Meanwhile, the AutoSpineAI system, published in IEEE-EMBS BHI, introduces an innovative AgenticRAG framework using knowledge graph-based reasoning, ensuring that AI-generated reports align with quantitative measurements extracted from 3D MRI scans.
By tightly integrating data-driven insights with structured reasoning, we minimize hallucinations and enable the generation of clinically interpretable reports that medical professionals can confidently rely on.
Q. We heard that you are actively promoting open science by releasing large-scale lumbar MRI datasets and core codebases. What motivated this decision, and what impact do you expect it to have on the global medical AI ecosystem?
A. Embracing open science was a decision aimed at contributing to the advancement of the global medical AI community. By providing open access to key research assets, we aim to foster international collaboration, improve reproducibility, and accelerate innovation in spinal disease analysis and AI-based diagnostics. Ultimately, we expect this to contribute to a more efficient global research ecosystem and enable faster, more reliable development of practical AI solutions.
Q. As medical AI rapidly advances, ethical and regulatory guidelines are becoming increasingly important. What are the main barriers to integrating medical AI into real hospital systems, and what direction should academia and industry take to overcome them?
A. The biggest barriers are lack of interpretability and trust, especially due to the limitations of black-box models. Despite rapid technological progress, clinicians still face challenges in understanding and verifying AI decisions before relying on them in critical situations.
To address this, close collaboration with medical professionals is essential. In our research, we work with radiologists and neurosurgeons from hospitals in Korea and Türkiye to ensure clinical validity and practicality. We also provide interactive tools, such as heatmaps, that allow clinicians to directly interpret and verify results, thereby enhancing transparency and accountability. Ultimately, medical AI should not replace clinical expertise but serve as a reliable assistant that complements it.
Q. Lastly, do you have any advice for students interested in AI and data science?
A. I always encourage students to embrace open-source innovation. By sharing their achievements through open-source platforms, they can grow within their respective fields. It is also important to work across multiple modalities—such as vision and language—and to focus on real-world applications like healthcare. Through this, students can build groundbreaking frameworks and gain valuable opportunities to contribute to the global community.
Reported by Minseok Cho, PR Reporter (minseok020929@naver.com)