Joohyeok Kim

I am a second-year master’s student at Yonsei University, advised by Prof. Seong Jae Hwang at MICV (Medical Imaging and Computer Vision) Lab.

My research interests lie in computer vision, medical image analysis, and efficient AI.

I am currently a research intern at LG AI Research’s Bio Intelligence Lab, where I work on computational pathology and spatial transcriptomics.

Contact: kyyle2114@yonsei.ac.kr

Selected Publications

Contrastive and Adaptive Multi-modal Masked Autoencoder for Spatial Transcriptomics

CAMMST, a contrastive multi-modal masked autoencoder that adaptively selects genetic anchors to predict whole-slide gene expression from partial transcriptomic coverage.

Joohyeok Kim, Taejin Jeong, Jinyeong Kim, Seong Jae Hwang

29th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI 2026)

arXiv Code

FEAST: Fully Connected Expressive Attention for Spatial Transcriptomics

FEAST, an attention-based framework that models the tissue as a fully connected graph, enabling the consideration of all pairwise interactions for spatial transcriptomics.

Taejin Jeong*, Joohyeok Kim*, Jinyeong Kim, Chanyoung Kim, Seong Jae Hwang

The IEEE/CVF Conference on Computer Vision and Pattern Recognition 2026 (CVPR 2026)

arXiv Code

Rethinking Glaucoma Calibration: Voting-Based Binocular and Metadata Integration

V-ViT, a novel framework that enhances calibration by incorporating glaucoma-specific characteristics.

Taejin Jeong, Joohyeok Kim, Jaehoon Joo, Seong Jae Hwang

2026 IEEE 23rd International Symposium on Biomedical Imaging (ISBI 2026 Oral)

arXiv Code

Interpreting vision transformers via residual replacement model

Systematically analyzing 6.6K ViT features across all layers, extracted via sparse autoencoders.

Jinyeong Kim, Junhyeok Kim, Yumin Shim, Joohyeok Kim, Sunyoung Jung, Seong Jae Hwang

39th Conference on Neural Information Processing Systems (NeurIPS 2025)

arXiv Code

EP-SAM: Weakly Supervised Histopathology Segmentation via Enhanced Prompt with Segment Anything

EP-SAM, a weakly supervised semantic segmentation framework by CAM and SAM-based pseudo labeling for histopathology images.

Joonhyeon Song, Seohwan Yun, Seongho Yoon, Joohyeok Kim, Sangmin Lee

arXiv Code