Ju-Hyeon Nam

Ju-Hyeon Nam is a Ph.D. student in the Department of Computer Engineering at Inha University, South Korea. He is currently conducting his doctoral research under the supervision of Professor Sang-Chul Lee in the Computer Vision and Artificial Intelligence (CVAI) Laboratory at Inha University. His research interests primarily involve developing reliable deep learning models robust against various corruptions, perturbations, and adversarial attacks using data augmentation techniques. Additionally, he actively researches modality-agnostic and domain-generalizable models for medical image segmentation applications across multiple imaging modalities, including colonoscopy, microscopy, dermoscopy, ultrasound, and radiology.

Ju-Hyeon has made significant contributions to medical image processing, frequency-domain analysis, and multi-modality fusion. He has introduced novel methods such as M3FPolypSegNet and MADGNet, which are accepted in ICIP2023 and CVPR2024 for medical image segmentation.

He actively contributes as a reviewer for top-tier conferences, including CVPR (2021-2024), ICCV (2021, 2023), ICLR (2022), and AAAI (2021-2025), and has also reviewed manuscripts for prestigious journals such as Computer Vision and Image Understanding (Elsevier), Pattern Recognition (Elsevier), Journal of Medical Systems (Springer), and IEEE Access.

Ju-Hyeon earned his bachelor’s degree in Mathematics from Inha University in 2021 (Summa Cum Laude) and his master’s degree in Computer Engineering from the same institution in 2023.

For more details about his research and publications, please visit his publications page or contact him via email at jhnam0514@inha.edu.

news

Feb 24, 2025	Ju-Hyeon Nam was honored with the *BK21 Top Researcher Award* in Inha University.
Aug 25, 2024	His third author paper is accepted in Elsevier Engineering Applications of Artificial Intelligence (IF: 7.5/2024, Q1), Park, S. H., Syazwany, N. S., *Nam, J. H.*, & Lee, S. C. (2024). Integrating multimodal contrastive learning with prototypical domain alignment for unsupervised domain adaptation of time series. Engineering Applications of Artificial Intelligence, 137, 109205.
Aug 22, 2024	He presented at the *2024 Digital Innovation and Convergence Talent Symposium* as a representative of Inha University.
Jun 07, 2024	His *first author paper* is accepted in ICIP2024, *Nam, J. H.*, Park, S. H., Kim, S. J., & Lee, S. C. (2024, October). Vizecgnet: Visual ECG Image Network for Cardiovascular Diseases Classification With Multi-Modal Training and Knowledge Distillation. In 2024 IEEE International Conference on Image Processing (ICIP) (pp. 3219-3223). IEEE.
Feb 27, 2024	His *first author paper* is accepted in CVPR2024 (BK21 IF:4), *Nam, J. H.*, Syazwany, N. S., Kim, S. J., & Lee, S. C. (2024). Modality-agnostic domain generalizable medical image segmentation by multi-frequency in multi-scale attention. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 11480-11491).

selected publications

Modality-agnostic domain generalizable medical image segmentation by multi-frequency in multi-scale attention

Ju-Hyeon Nam, Nur Suriza Syazwany, Su Jung Kim, and Sang-Chul Lee

In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2024

Abs DOI Bib HTML Video

Generalizability in deep neural networks plays a pivotal role in medical image segmentation. However deep learning-based medical image analyses tend to overlook the importance of frequency variance which is critical element for achieving a model that is both modality-agnostic and domain-generalizable. Additionally various models fail to account for the potential information loss that can arise from multi-task learning under deep supervision a factor that can impair the model’s representation ability. To address these challenges we propose a Modality-agnostic Domain Generalizable Network (MADGNet) for medical image segmentation which comprises two key components: a Multi-Frequency in Multi-Scale Attention (MFMSA) block and Ensemble Sub-Decoding Module (E-SDM). The MFMSA block refines the process of spatial feature extraction particularly in capturing boundary features by incorporating multi-frequency and multi-scale features thereby offering informative cues for tissue outline and anatomical structures. Moreover we propose E-SDM to mitigate information loss in multi-task learning with deep supervision especially during substantial upsampling from low resolution. We evaluate the segmentation performance of MADGNet across six modalities and fifteen datasets. Through extensive experiments we demonstrate that MADGNet consistently outperforms state-of-the-art models across various modalities showcasing superior segmentation performance. This affirms MADGNet as a robust solution for medical image segmentation that excels in diverse imaging scenarios. Our MADGNet code is available in GitHub Link.
@inproceedings{nam2024modality, title = {Modality-agnostic domain generalizable medical image segmentation by multi-frequency in multi-scale attention}, author = {Nam, Ju-Hyeon and Syazwany, Nur Suriza and Kim, Su Jung and Lee, Sang-Chul}, booktitle = {Proceedings of the IEEE/CVF conference on computer vision and pattern recognition}, pages = {11480--11491}, year = {2024}, doi = {10.1109/CVPR52733.2024.01091}, }