2026

BiFlow-GAN: Integrating Bidirectional Flows and Cross-Modal Enhancement for Multi-Modal Spatial Transcriptomics

Jingxuan Wang, Rui Guan, Kun Liang, Shiyu Wang, Xin Yang, Xiaodan Cui, Di Hu

IEEE Transactions on Multimedia (Under Review) 2026 JCR Q1

We propose BiFlow-GAN, a bidirectional-flow generative adversarial framework for robust tissue domain identification in spatial transcriptomics. We design a bidirectional cross-modal attention flow (Bi-AF) mechanism with reciprocal Image-to-Gene and Gene-to-Image pathways for dynamic feature refinement. Benchmarked on diverse spatial transcriptomics datasets, significantly outperforming state-of-the-art methods.

BiFlow-GAN: Integrating Bidirectional Flows and Cross-Modal Enhancement for Multi-Modal Spatial Transcriptomics

Jingxuan Wang, Rui Guan, Kun Liang, Shiyu Wang, Xin Yang, Xiaodan Cui, Di Hu

IEEE Transactions on Multimedia (Under Review) 2026 JCR Q1

We propose BiFlow-GAN, a bidirectional-flow generative adversarial framework for robust tissue domain identification in spatial transcriptomics. We design a bidirectional cross-modal attention flow (Bi-AF) mechanism with reciprocal Image-to-Gene and Gene-to-Image pathways for dynamic feature refinement. Benchmarked on diverse spatial transcriptomics datasets, significantly outperforming state-of-the-art methods.

2025

DGAN-MPCC: A Novel Dual-GAN Enhanced Multi-Positive Contrastive Clustering Method for Omics Data

Jingxuan Wang, Jie Yang, M. A. Khan, Poh Loong Yee, Jian Baili, Di Hu

IEEE Journal of Biomedical and Health Informatics 2025 JCR Q1

We propose a dual-GAN framework to enhance both input and latent representations for noisy single-cell data, and design a multi-positive contrastive clustering strategy to model continuous cell-state transitions. Evaluated on multiple real-world single-cell datasets, achieving state-of-the-art clustering performance.

DGAN-MPCC: A Novel Dual-GAN Enhanced Multi-Positive Contrastive Clustering Method for Omics Data

Jingxuan Wang, Jie Yang, M. A. Khan, Poh Loong Yee, Jian Baili, Di Hu

IEEE Journal of Biomedical and Health Informatics 2025 JCR Q1

We propose a dual-GAN framework to enhance both input and latent representations for noisy single-cell data, and design a multi-positive contrastive clustering strategy to model continuous cell-state transitions. Evaluated on multiple real-world single-cell datasets, achieving state-of-the-art clustering performance.

Enhancing Study-Abroad English Learning with AI-Powered Scenario Simulation

Shengtao Zhu, Jingxuan Wang, Xiaogang Hu, Jian Tao

Proceedings of ICFULL 2025 & GLoCALL 2025 2025 Oral / Poster

We conducted an undergraduate innovation research project focused on AI-driven scenario simulation for study-abroad English learning, resulting in a peer-reviewed conference paper and presentation.

Enhancing Study-Abroad English Learning with AI-Powered Scenario Simulation

Shengtao Zhu, Jingxuan Wang, Xiaogang Hu, Jian Tao

Proceedings of ICFULL 2025 & GLoCALL 2025 2025 Oral / Poster

We conducted an undergraduate innovation research project focused on AI-driven scenario simulation for study-abroad English learning, resulting in a peer-reviewed conference paper and presentation.