Kyungwon Kim

News

Feb 2026

Paper has been accepted to CVPR 2026 in Denver Colorado, United States — "MUST: Modality-Specific Representation-Aware Transformer for Diffusion-Enhanced Survival Prediction with Missing Modality"

Sep 2024

Completed a 4-month on-site research visit at Mila in Montréal — LLM-based clinical report generation from 3D CT

Jun 2024

Paper has been accepted to MICCAI 2024 in Marrakesh, Morocco — "LLM-guided Multi-modal Multiple Instance Learning for 5-year Overall Survival Prediction of Lung Cancer"

Oct 2023

3rd Place at MICCAI 2023 Challenge — SEG.A. Segmentation of the Aortic Vessel Tree

Publications

CVPR 2026 · h5-index 450

MUST: Modality-Specific Representation-Aware Transformer for Diffusion-Enhanced Survival Prediction with Missing Modality

Kyungwon Kim, and Dosik Hwang

URL GitHub

MICCAI 2024 · FWCI 5.09

LLM-guided Multi-modal Multiple Instance Learning for 5-year Overall Survival Prediction of Lung Cancer

Kyungwon Kim, Yongmoon Lee, Doohyun Park, Taejoon Eo, Daemyung Youn, Hyesang Lee, and Dosik Hwang

URL GitHub

MICCAI Challenge 2023

M3F: Multi-Field-of-View Feature Fusion Network for Aortic Vessel Tree Segmentation in CT Angiography

Yunsu Byeon, Hyeseong Kim, Kyungwon Kim, Doohyun Park, Euijoon Choi, and Dosik Hwang

URL

SPIE Photonics West 2022

Deep residual network with data consistency for subsampled Fourier Ptychographic Microscopy

Hyeongyu Kim, Kyungwon Kim, Kyungchul Lee, Taejoon Eo, Seungah Lee, and Dosik Hwang

URL

ISMRM & SMRT 2020

Are Two MR Images Enough to Generate the Third One Accurately? — Clinically Feasible Fat Suppression of Lumbar Spine MRI

Sewon Kim, Hanbyol Jang, Kyungwon Kim, Hyeon Gyu Kim, Young Han Lee, Sungjun Kim, and Dosik Hwang

URL

Research

Multimodal

Multi-modal Cancer Prognosis Prediction

Survival prediction for cancer patients using CT images, pathological images (WSI), and genomic data under missing-modality settings.

Clinical Report

Clinical Report Generation

LLM-based approaches for generating clinical reports including findings and impressions from 3D CT images.

3D CT

Segmentation & Classification

Automatic detection of cancer regions in radiological images.

X-ray

Self-supervised X-ray Denoising

Deep learning methods for denoising low-dose X-ray images without paired supervision.

Optical Microscopy

Fourier Ptychographic Microscopy

Deep learning-based reconstruction of high-resolution optical images from multiple low-resolution acquisitions.

3D CT

Landmark Detection

Detection of landmarks in 3D craniofacial CT for identifying congenital anomalies.

3D STL scan

Navigation Implant

Finding the optimal implant location using tooth scan data.

News

Publications

MUST: Modality-Specific Representation-Aware Transformer for Diffusion-Enhanced Survival Prediction with Missing Modality

LLM-guided Multi-modal Multiple Instance Learning for 5-year Overall Survival Prediction of Lung Cancer

M3F: Multi-Field-of-View Feature Fusion Network for Aortic Vessel Tree Segmentation in CT Angiography

Deep residual network with data consistency for subsampled Fourier Ptychographic Microscopy

Are Two MR Images Enough to Generate the Third One Accurately? — Clinically Feasible Fat Suppression of Lumbar Spine MRI

Research

Multi-modal Cancer Prognosis Prediction

Clinical Report Generation

Segmentation & Classification

Self-supervised X-ray Denoising

Fourier Ptychographic Microscopy

Landmark Detection

Navigation Implant

Awards

MICCAI 2023 SEG.A. Challenge

Yonsei Medical Center Big Data Challenge

Education

Curriculum Vitae

Contact

Email

Office

GitHub

LinkedIn