Publications
Overview
2023
Journal
- Handling Open-set Noise and Novel Target Recognition in Domain Adaptive Semantic Segmentation
Xiaoqing Guo, Jie Liu, Tongliang Liu, Yixuan Yuan.
IEEE Transactions on Pattern Analysis and Machine Intelligence (IEEE TPAMI), 2023 [paper]bibtex
@article{guo2023handling,
title={Handling Open-set Noise and Novel Target Recognition in Domain Adaptive Semantic Segmentation},
author={Guo, Xiaoqing and Liu, Jie and Liu, Tongliang and Yuan, Yixuan},
journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},
year={2023},
publisher={IEEE}
} - GRAB-Net: Graph-based Boundary-aware Network for Medical Point Cloud Segmentation
Yifan Liu, Wuyang Li, Jie Liu, Hui Chen, Yixuan Yuan.
IEEE Transactions on Medical Imaging (IEEE TMI), 2023. [paper]bibtex
@article{liu2023grab,
title={GRAB-Net: Graph-based Boundary-aware Network for Medical Point Cloud Segmentation},
author={Liu, Yifan and Li, Wuyang and Liu, Jie and Chen, Hui and Yuan, Yixuan},
journal={IEEE Transactions on Medical Imaging},
year={2023},
publisher={IEEE}
}
Conference
- CLIP-Driven Universal Model for Organ Segmentation and Tumor Detection
Jie Liu, Yixiao Zhang, Jie-Neng Chen, Junfei Xiao, Yongyi Lu, Yixuan Yuan, Alan Yuille, Yucheng Tang, Zongwei Zhou
IEEE International Conference on Computer Vision (ICCV), 2023, Paris, France. [paper] [code] [press] [Rank 1st in MSD]
Medical dataset is usually limited in size, partially labeled, and focuses narrowly on specific cancer types, constraining the AI model to detect only particular cancer types and may not be easily extended to new cancers. Thus, we develop an assembly of 14 datasets and propose the CLIP-Driven Universal Model.bibtex
Stay Tuned! - Large Language-Image Model for Multi-Organ Segmentation and Cancer Detection from Computed Tomography
Jie Liu, Yixiao Zhang, Jie-Neng Chen, Junfei Xiao, Yongyi Lu, Yixuan Yuan, Alan Yuille, Yucheng Tang, Zongwei Zhou
Radiological Society of North America (RSNA), 2023, Chicago, Illinois. [Oral Presentation]
[paper]
We propose a large language-image model, which is computationally more efficient compared with dataset-specific models, generalized better to CT scans from varying sites, and shows stronger transfer learning performance on novel tasks.bibtex
Stay Tuned! - Adjustment and Alignment for Unbiased Open Set Domain Adaptation
Wuyang Li, Jie Liu, Bo Han, Yixuan Yuan
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2023, Vancouver, Canada.
[paper] [code]
We decoupled images into base-class and novel-class regions and addreseed the biased learning in the source domain and biased cross-domain transfer with causal theory.bibtex
@inproceedings{li2023adjustment,
title={Adjustment and Alignment for Unbiased Open Set Domain Adaptation},
author={Li, Wuyang and Liu, Jie and Han, Bo and Yuan, Yixuan},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
pages={24110--24119},
year={2023}
}
2022
Journal
- Instance Importance-Aware Graph Convolutional Network for 3D Medical Diagnosis
Zhen Chen, Jie Liu, Meilu Zhu, Peter Y. M. Woo, Yixuan Yuan. Medical Image Analysis (MedIA), 2022. [paper] [Code]
We propose a Multi-Instance Learning framework named Instance Importance-aware Graph Convolutional Network (I2GCN), which measures the instance importance and explores the complementary information among instances.bibtex
@article{chen2022instance,
title={Instance importance-Aware graph convolutional network for 3D medical diagnosis},
author={Chen, Zhen and Liu, Jie and Zhu, Meilu and Woo, Peter YM and Yuan, Yixuan},
journal={Medical Image Analysis},
volume={78},
pages={102421},
year={2022},
publisher={Elsevier}
}
Conference
- Unknown-Oriented Learning for Open Set Domain Adaptation
Jie Liu, Xiaoqing Guo, Yixuan Yuan.
The 17th European Conference on Computer Vision (ECCV), 2022, Tel-Aviv. [paper]
In open set domain adaptation (OSDA) problem, we dig out the valuable information in unknown category via multi-unknown detector and label graph propagation.bibtex
@inproceedings{liu2022unknown,
title={Unknown-Oriented Learning for Open Set Domain Adaptation},
author={Liu, Jie and Guo, Xiaoqing and Yuan, Yixuan},
booktitle={Computer Vision--ECCV 2022: 17th European Conference, Tel Aviv, Israel, October 23--27, 2022, Proceedings, Part XXXIII},
pages={334--350},
year={2022},
organization={Springer}
} - Edge-oriented Point-cloud Transformer for 3D Intracranial Aneurysm Segmentation
Yifan Liu, Jie Liu, Yixuan Yuan.
International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI), 2022, Singapore. [Early Accept] [paper] [code]
To achieve accurate surgery procedure, we focus on edge modeling in 3D point data and propose an Edge-oriented Point-cloud Transformer Network (EPT-Net) with transformer and graph techniques.bibtex
@inproceedings{liu2022edge,
title={Edge-Oriented Point-Cloud Transformer for 3D Intracranial Aneurysm Segmentation},
author={Liu, Yifan and Liu, Jie and Yuan, Yixuan},
booktitle={Medical Image Computing and Computer Assisted Intervention--MICCAI 2022: 25th International Conference, Singapore, September 18--22, 2022, Proceedings, Part V},
pages={97--106},
year={2022},
organization={Springer}
} - SimT: Handling Open-set Noise for Domain Adaptive Semantic Segmentation
Xiaoqing Guo, Jie Liu, Tongliang Liu, Yixuan Yuan.
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2022, New Orleans, Louisiana. [paper] [Code] [Zhihu]
We present a general DA semantic segmentation framework named SIMplex noise Transition matrix (SimT), which utilizes computational geometry analysis to model the closed-set and open-set noise distributions of target pseudo labels.bibtex
@inproceedings{guo2022simt,
title={SimT: Handling Open-set Noise for Domain Adaptive Semantic Segmentation},
author={Guo, Xiaoqing and Liu, Jie and Liu, Tongliang and Yuan, Yixuan},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
pages={7032--7041},
year={2022}
}
2021
Journal
- Graph-based Surgical Instrument Adaptive Segmentation via Domain-Common Knowledge
Jie Liu, Xiaoqing Guo, Yixuan Yuan.
IEEE Transactions on Medical Imaging (IEEE TMI), 2021. [paper] [Code]
We study the domain adaptation in the surgical scenario and present the IGNet that obtains common knowledge via fusing prototypes of two domains to provide supplementary information for the target domain.bibtex
@article{liu2021graph,
title={Graph-based Surgical Instrument Adaptive Segmentation via Domain-Common Knowledge},
author={Liu, Jie and Guo, Xiaoqing and Yuan, Yixuan},
journal={IEEE Transactions on Medical Imaging},
year={2021},
publisher={IEEE}
} - Semantic-oriented Labeled-to-unlabeled Distribution Translation for Image Segmentation
Xiaoqing Guo, Jie Liu, Yixuan Yuan.
IEEE Transactions on Medical Imaging (IEEE TMI), 2021. [paper] [Code]
To overcome the scare annotation problem in the medical domain, a novel Label-to-unlabeled Distribution Translation (L2uDT) framework is proposed to minimize a theoretical upper bound derived by the first-order Taylor expansion.bibtex
@article{guo2021semantic,
title={Semantic-oriented Labeled-to-unlabeled Distribution Translation for Image Segmentation},
author={Guo, Xiaoqing and Liu, Jie and Yuan, Yixuan},
journal={IEEE Transactions on Medical Imaging},
year={2021},
publisher={IEEE}
}
Conference
- Prototypical Interaction Graph for Unsupervised Domain Adaptation in Surgical Instrument Segmentation
Jie Liu, Xiaoqing Guo, Yixuan Yuan.
International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI), 2021, Strasbourg, France. [Early Accept] [paper] [Code] [Video]
To establish the inter-category relationship, we propose a prototypical inner-interaction graph and cross-interaction graph with the nodes being means of Gaussian Mixture Model.bibtex
@inproceedings{liu2021prototypical,
title={Prototypical Interaction Graph for Unsupervised Domain Adaptation in Surgical Instrument Segmentation},
author={Liu, Jie and Guo, Xiaoqing and Yuan, Yixuan},
booktitle={International Conference on Medical Image Computing and Computer-Assisted Intervention},
pages={272--281},
year={2021},
organization={Springer}
} - COINet: Adaptive Segmentation with Co-Interactive Network for Autonomous Driving
Jie Liu, Xiaoqing Guo, Baopu Li, Yixuan Yuan.
IEEE International Conference on Intelligent Robots and Systems (IROS), 2021, Prague, Czech Republic. [paper] [slides]
COINet builds a mutual interaction between the adversarial branch and segmentation branch in unsupervised domain adaptation task.bibtex
@inproceedings{liu2021coinet,
title={COINet: Adaptive Segmentation with Co-Interactive Network for Autonomous Driving},
author={Liu, Jie and Guo, Xiaoqing and Li, Baopu and Yuan, Yixuan},
booktitle={2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
pages={4800--4806},
year={2021},
organization={IEEE}
} - Joint Polyp Detection and Segmentation with Heterogeneous Endoscopic Data
Wuyang Li, Chen Yang, Jie Liu, Xinyu Liu, Xiaoqing Guo, Yixuan Yuan.
IEEE International Symposium on Biomedical Imaging Endoscopy Workshop (EndoCV 2021), 2021. [paper]
1st place for polyp detection and 3rd place for polyp segmentation in the EndoCV 2021 Challenge.bibtex
@inproceedings{wuyang2021joint,
title={Joint Polyp Detection and Segmentation with Heterogeneous Endoscopic Data},
author={Wuyang, LI and Chen, YANG and Jie, LIU and Xinyu, LIU and Xiaoqing, GUO and Yixuan, YUAN},
booktitle={3rd International Workshop and Challenge on Computer Vision in Endoscopy (EndoCV 2021): co-located with with the 17th IEEE International Symposium on Biomedical Imaging (ISBI 2021)},
pages={69--79},
year={2021},
organization={CEUR Workshop Proceedings}
}