One Stop Service Center

Sejong News

Sejong University MITI Lab Sweeps Three Awards at ICMRI·ASMRM 2025, Including Best Contribution Award

오세혁 2025-12-15

The MITI Lab (Medical Imaging & Technology Intelligence Laboratory), led by Professor Mohamed Almasni from the AI Data Science Department, won the Best Contribution Award, the Best Oral Presentation Silver Award, and the Best Digital Poster Gold Award at the international medical imaging conference, ICMRI & ASMRM 2025.

Central to these awards is the lab's recently published research, “Learning robust brain tumor segmentation under label corruption and data scarcity.” This latest study, which was published in Engineering Applications of Artificial Intelligence (EAAI), proposes a novel method that simultaneously addresses the problems of label errors and data scarcity that brain tumor MRI segmentation models often encounter in real-world clinical environments.

The research team developed the MGR-DAS (Meta-Gradient Reweighting via Direction-Aware Similarity) framework. This framework automatically identifies samples critical for learning by comparing the gradient direction between a small amount of highly reliable data and the entire dataset, thereby autonomously excluding less reliable data.

The MITI Lab research team achieved significant international recognition not only for the completeness of their research but also for their presentation skills and academic contribution, winning the Silver Award in the oral presentation category and the Gold Award in the digital poster category. This is a symbolic result demonstrating the world-class influence of Sejong University's medical AI research.

Professor Almasni stated, “These awards are very meaningful as they signify international recognition of MITI Lab’s research achievements, and they served as an opportunity to reconfirm the role medical AI can play in solving real clinical problems.” He added, “We will continue to develop robust and interpretable AI technologies that can be utilized even in environments with limited medical data, contributing to the forefront of global medical innovation.”

The MITI Lab plans to continue expanding international collaborations and developing AI-based diagnostic assistance technologies applicable in diverse medical settings. In particular, they will focus on solving practical issues such as data disparity, shortage of medical resources, and label quality variance to continuously develop medical AI technologies that can be immediately applied in clinical practice.

Reporter / Hyunseok Lee, PR Reporter (hslee901@naver.com)