Gawon Lee, PhD Candidate at Hankuk University of Foreign Studies, Publishes Paper in 'NeuroImage'

Artificial Intelligence-Based MRI Harmonization Technology
Developed by Gawon Lee Published in NeuroImage

Hankuk University of Foreign Studies announced on July 18 that an artificial intelligence-based MRI harmonization technology developed by Gawon Lee, a PhD candidate in the Department of Biomedical Engineering, has been published in the world-renowned neuroimaging journal NeuroImage.

Gawon Lee, PhD Candidate, Hankuk University of Foreign Studies

MRI images often appear different even for the same person, depending on the type of machine and imaging conditions at each hospital. This variation is a major factor that undermines analytical accuracy in multi-institutional studies. As a result, the need for harmonization technology to address this issue has been continuously raised.

The core of this research is the development of an AI algorithm based on the physical principles of MRI, called PhyCHarm (Physics-Constrained Harmonization).

This technology integrates the physical equations (Bloch equations) that generate MRI signals into the AI model, allowing it to adjust images by taking into account the actual biological information represented in the scans.

The research team validated the technology using brain imaging data acquired from various MRI devices at institutions including Seoul National University Hospital. As a result, the technology maintained image quality more consistently than existing harmonization models and succeeded in precisely distinguishing between gray matter and white matter in the brain.

The SSIM metric, which indicates structural similarity of brain images, reached as high as 0.9579. The Dice score for brain tissue segmentation also achieved top-level results for both gray matter and white matter.

Professor Sehong Oh, the supervising professor, stated, "This original approach, which combines physical modeling and artificial intelligence, could set a new standard for medical image harmonization technology in the future."

Meanwhile, Gawon Lee is the first student admitted to the Department of Biomedical Engineering at Hankuk University of Foreign Studies and is expected to receive a PhD in February 2026. To date, Lee has led various research projects, including multi-institutional brain imaging analysis and the development of image-based AI algorithms.