Dr. Yohan Jun’s research focuses on developing novel MR acquisition and reconstruction techniques using MR physics and machine/deep-learning-based algorithms to accelerate MRI scans while achieving high-fidelity images. He works on the following research topics:
i. Rapid high-resolution quantitative imaging: Zero-shot self-supervised learning combined with subspace reconstruction technique (Zero-DeepSub) enables the acquisition of quantitative T1, T2, and proton density maps with high-fidelity using 3D-QALAS sequence. The self-supervised-learning-based mapping technique (SSL-QALAS) can also accelerate the quantitative mapping process without any external dataset or explicit dictionary.
ii. Highly accelerated distortion-free diffusion imaging: Phase Reversed Interleaved Multi-Echo acquisition (PRIME) enables distortion-free diffusion MRI by inserting an additional echo without prolonging TR, when generalized slice dithered enhanced resolution (gSlider) RF encoding is used for volumetric acquisition.
iii. Acceleration of structural MRI: Deep model-based MR image reconstruction algorithms (Joint-ICNet, DPI-net), which combine MR physical and deep-learning models, allow fast MRI up to 8-fold for TOF MRA, T1/T2-weighted, FLAIR, and post-contrast imaging.
iv. Automatic diagnosis of brain disorders: Artificial intelligence algorithms allow us to support the diagnosis of brain disorders, including brain metastases, meningioma, and glioblastoma, by using deep-learning-based diagnosis algorithms for automatic detection, segmentation, and grading.
Education
PhD in Electrical and Electronic Engineering, Yonsei University, South Korea
Select Publications
1. Jun, Y., Arefeen, Y., Cho, J., Fujita, S., Wang, X., Grant, P.E., Gagoski, B., Jaimes, C., Gee, M.S. and Bilgic, B., 2024. Zero‐DeepSub: Zero‐shot deep subspace reconstruction for rapid multiparametric quantitative MRI using 3D‐QALAS. Magnetic Resonance in Medicine, 91(6), pp.2459-2482.
2. Jun, Y., Cho, J., Wang, X., Gee, M., Grant, P.E., Bilgic, B. and Gagoski, B., 2023. SSL‐QALAS: Self‐Supervised Learning for rapid multiparameter estimation in quantitative MRI using 3D‐QALAS. Magnetic Resonance in Medicine, 90(5), pp.2019-2032.
3. Jun, Y., Shin, H., Eo, T., Kim, T. and Hwang, D., 2021. Deep model-based magnetic resonance parameter mapping network (DOPAMINE) for fast T1 mapping using variable flip angle method. Medical Image Analysis, 70, p.102017.
Highlights
2024 – ISMRM Junior Fellow
2024 – NIBIB R21 Trailblazer Award
2017-2024 – ISMRM 4 Summa Cum Laude & 2 Magna Cum Laude Merit Awards