University of MacauPinxian Zeng received his B.S. and M.S. degrees from Sichuan University. He is currently a Ph.D. candidate in Computer Science at the Faculty of Science and Technology, University of Macau. His primary research interests focus on computer vision, medical image analysis, and medical/biometric image restoration. He is also interested in the trending fields like large vision-language model and world model.
Education
-
University of MacauPh.D. in Computer Science and Technology2025 - Present
-
Sichuan UniversityM.ENG. in Artificial Intelligence2022 - 2025 -
Sichuan UniversityB.S. in Computer Science and Technology2018 - 2022
Honors & Awards
-
Type-A Scholarship of University of Macau (full funding)2025
-
IEEE CIS/RAM Best Student Paper2024
-
Second-Prize Graduate Academic Scholarship (Sichuan University)2024
-
Second-Prize Graduate Academic Scholarship (Sichuan University)2022
-
MCM/ICM Finalist Award2021
-
Huawei Scholarship2023
News
Selected Publications (view all )
PalmIR: Blind Palmprint Image Restoration with Palm-oriented Mixture-of-Experts and Downstream Alignment for Robust Identification
Pinxian Zeng, Chenqin Wu, Bob Zhang
IEEE Transactions on Information Forensics and Security (TIFS) 2026
We propose PalmIR, a novel blind palmprint restoration framework featuring a Palm-oriented Mixture-of-Experts (PalmMoE) network that adaptively processes mixed degradations using grouped domain-specific experts, and a Restoration-Identification Alignment (RIA) module that preserves identity-critical features through multi-task learning.
PalmIR: Blind Palmprint Image Restoration with Palm-oriented Mixture-of-Experts and Downstream Alignment for Robust Identification
Pinxian Zeng, Chenqin Wu, Bob Zhang
IEEE Transactions on Information Forensics and Security (TIFS) 2026
We propose PalmIR, a novel blind palmprint restoration framework featuring a Palm-oriented Mixture-of-Experts (PalmMoE) network that adaptively processes mixed degradations using grouped domain-specific experts, and a Restoration-Identification Alignment (RIA) module that preserves identity-critical features through multi-task learning.
MPOTSurv: Modality-balancing Prototype Learning with Optimal Transport-based Integration for Multi-modal Survival Prediction
Pinxian Zeng, Bob Zhang
ACM Transactions on Intelligent Systems and Technology 2026
we propose MPOTSurv, a multi-modal framework that leverages Modality-Balancing Prototype Learning (MBPL) with Optimal Transport (OT) to map disparate modalities onto a symmetrical, unified latent manifold with more balanced importance for accurate cancer Survival prediction. Within the MBPL module, we introduce Sinkhorn-Knopp OT to achieve principled feature aggregation, enforcing strict marginal constraints that suppress morphological noise in WSI and tabular data.
MPOTSurv: Modality-balancing Prototype Learning with Optimal Transport-based Integration for Multi-modal Survival Prediction
Pinxian Zeng, Bob Zhang
ACM Transactions on Intelligent Systems and Technology 2026
we propose MPOTSurv, a multi-modal framework that leverages Modality-Balancing Prototype Learning (MBPL) with Optimal Transport (OT) to map disparate modalities onto a symmetrical, unified latent manifold with more balanced importance for accurate cancer Survival prediction. Within the MBPL module, we introduce Sinkhorn-Knopp OT to achieve principled feature aggregation, enforcing strict marginal constraints that suppress morphological noise in WSI and tabular data.
Dual-Path Latent Diffusion with Multi-Task Interaction for CT-to-PET Translation and Tumor Segmentation
Pinxian Zeng, Xinyi Zeng, Bob Zhang
ICASSP 2026 - 2026 IEEE International Conference on Acoustics, Speech and Signal Processing 2026
We propose a dual-path latent diffusion framework with multi-task interaction for CT-to-PET translation and tumor segmentation. Our method achieves state-of-the-art performance on both tasks.
Dual-Path Latent Diffusion with Multi-Task Interaction for CT-to-PET Translation and Tumor Segmentation
Pinxian Zeng, Xinyi Zeng, Bob Zhang
ICASSP 2026 - 2026 IEEE International Conference on Acoustics, Speech and Signal Processing 2026
We propose a dual-path latent diffusion framework with multi-task interaction for CT-to-PET translation and tumor segmentation. Our method achieves state-of-the-art performance on both tasks.
Plug-in-and-Play Dual-Domain Mixup Augmentation Network for PET Denoising at Multiple Dose Levels
Pinxian Zeng, Bob Zhang, Haozhuo Zhang
ICASSP 2026 - 2026 IEEE International Conference on Acoustics, Speech and Signal Processing 2026
we propose a Dual-Domain Mixup Augmentation Network (D2MA-Net) for robust multi-dose PET denoising using only training samples of a single dose level.
Plug-in-and-Play Dual-Domain Mixup Augmentation Network for PET Denoising at Multiple Dose Levels
Pinxian Zeng, Bob Zhang, Haozhuo Zhang
ICASSP 2026 - 2026 IEEE International Conference on Acoustics, Speech and Signal Processing 2026
we propose a Dual-Domain Mixup Augmentation Network (D2MA-Net) for robust multi-dose PET denoising using only training samples of a single dose level.
MAK-GAN: Multi-level Adaptive Convolutional Kernels for Asymmetric Multi-modal PET Reconstruction
Xinyi Zeng, Pinxian Zeng*, Yan Wang, Jiaqi Cui, Luping Zhou, Caiwen Jiang, Han Zhang, Dinggang Shen (* co-first)
MICCAI 2025
We propose MAK-GAN, a multi-level adaptive convolutional kernel framework for asymmetric multi-modal PET reconstruction.
MAK-GAN: Multi-level Adaptive Convolutional Kernels for Asymmetric Multi-modal PET Reconstruction
Xinyi Zeng, Pinxian Zeng*, Yan Wang, Jiaqi Cui, Luping Zhou, Caiwen Jiang, Han Zhang, Dinggang Shen (* co-first)
MICCAI 2025
We propose MAK-GAN, a multi-level adaptive convolutional kernel framework for asymmetric multi-modal PET reconstruction.
Multi-Modal Long-Short Distance Attention-Based Transformer-GAN for PET Reconstruction With Auxiliary MRI
Pinxian Zeng, Xinyi Zeng, Luping Zhou, Chen Zu, Yan Wang
IEEE Transactions on Circuits and Systems for Video Technology (TCSVT) 2025
We propose a multi-modal Transformer-GAN with long-short distance attention mechanisms for PET reconstruction using auxiliary MRI.
Multi-Modal Long-Short Distance Attention-Based Transformer-GAN for PET Reconstruction With Auxiliary MRI
Pinxian Zeng, Xinyi Zeng, Luping Zhou, Chen Zu, Yan Wang
IEEE Transactions on Circuits and Systems for Video Technology (TCSVT) 2025
We propose a multi-modal Transformer-GAN with long-short distance attention mechanisms for PET reconstruction using auxiliary MRI.
S3PET: Semi-supervised Standard-dose PET Image Reconstruction via Dose-aware Token Swap
Jiaqi Cui, Pinxian Zeng*, Yinghui Xu, Xi Wu, Jiliu Zhou, Yan Wang (* co-first)
IEEE CIS/RAM 2024
We propose S3PET, a semi-supervised standard-dose PET image reconstruction method via dose-aware token swap.
S3PET: Semi-supervised Standard-dose PET Image Reconstruction via Dose-aware Token Swap
Jiaqi Cui, Pinxian Zeng*, Yinghui Xu, Xi Wu, Jiliu Zhou, Yan Wang (* co-first)
IEEE CIS/RAM 2024
We propose S3PET, a semi-supervised standard-dose PET image reconstruction method via dose-aware token swap.
3D CVT-GAN: A 3D Convolutional Vision Transformer-GAN for PET Reconstruction
Pinxian Zeng, Luping Zhou, Chen Zu, Xinyi Zeng, Zhengyang Jiao, Xi Wu, Jiliu Zhou, Dinggang Shen, Yan Wang
MICCAI 2022
We propose 3D CVT-GAN, a 3D convolutional vision transformer-GAN for PET reconstruction.
3D CVT-GAN: A 3D Convolutional Vision Transformer-GAN for PET Reconstruction
Pinxian Zeng, Luping Zhou, Chen Zu, Xinyi Zeng, Zhengyang Jiao, Xi Wu, Jiliu Zhou, Dinggang Shen, Yan Wang
MICCAI 2022
We propose 3D CVT-GAN, a 3D convolutional vision transformer-GAN for PET reconstruction.