Gen Shi

Gen Shi

Auto-generated profile, based on publicly available information

Gen Shi

Beihang University

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
20 July 2023 in IEEE Transactions on Medical Imaging
Verified
Journal Article
Open access
GSGen Shi
Gen Shi
LYLin Yin
Lin Yin
YAYu An
Yu An
GLGuanghui Li
Guanghui Li
LZLiwen Zhang
Liwen Zhang
JTJie Tian
Jie Tian
Magnetic particle imaging (MPI) is an emerging technique for determining magnetic nanoparticle distributions in biological tissues. Although system-matrix (SM)-based image reconstruction offers higher image quality than the X-space-based approach, the SM calibration measurement is time-consuming. Additionally, the SM should be recalibrated if the tracer’s characteristics or the magnetic field environment change, and repeated SM measurement further increase the required labor and time. Therefore, fast SM calibration is essential for MPI. Existing calibration methods commonly treat each row of the SM as independent of the others, but the rows are inherently related through the coil channel and frequency index. As these two elements can be regarded as additional multimodal information, we leverage the transformer architecture with a self-attention mechanism to encode them. Although the transformer has shown superiority in multimodal fusion learning across several fields, its high complexity may lead to overfitting when labeled data are scarce. Compared with labeled SM (i.e., full size), low-resolution SM data can be easily obtained, and fully using such data may alleviate overfitting. Accordingly, we propose a pseudo-label-based progressive pretraining strategy to leverage unlabeled data. Our method outperforms existing calibration methods on a public real-world OpenMPI dataset and simulation dataset. Moreover, our method improves the resolution of two in-house MPI scanners without requiring full-size SM measurements. Ablation studies confirm the contributions of modeling SM inter-row relations and the proposed pretraining strategy.

1 October 2021 in Medical Image Analysis
Verified
Journal Article
Closed access
Yifan ZhuYifan Zhu
Yifan Zhu
XLXuesong Li
Xuesong Li
YQYufei Qiao
Yufei Qiao
RSRuihong Shang
Ruihong Shang
GSGen Shi
Gen Shi
GHGuo Hua
Guo Hua

13 May 2025 in IEEE Transactions on Knowledge and Data Engineering
Verified
Journal Article
Closed access
GSGen Shi
Gen Shi
Yifan ZhuYifan Zhu
Yifan Zhu
WLWenjin Liu
Wenjin Liu
QYQuanming Yao
Quanming Yao
XLXuesong Li
Xuesong Li
Graph neural networks (GNNs) provide powerful insights into brain neuroimaging technology from the view of graphical networks. However, most existing GNN-based models treat the brain connectome, derived from neuroimaging, as a homogeneous graph characterized by uniform node and edge types. In fact, emerging studies have reported and emphasized the significance of heterogeneity among human brain activities, especially between the two cerebral hemispheres. Thus, homogeneous-structured brain network-based graph methods are insufficient for modeling complicated cerebral activity states. To overcome this problem, we introduce a novel heterogeneous graph neural network (HeBrainGNN) for multimodal brain neuroimaging fusion learning. HeBrainGNN first conceptualizes the brain network as a heterogeneous graph with multiple types of nodes (representing the left and right hemispheres) and edges (categorizing intra- and interhemispheric interactions). We further develop a self-supervised pretraining strategy for this heterogeneous network to address the potential overfitting problem caused by the conflict between a large parameter size and a small medical data sample size. Empirical results show the superiority of the proposed model over other existing methods in brain-related disease prediction tasks. Ablation experiments show that our heterogeneous graph-based model attaches more importance to hemispheric connections that may be neglected due to their low strength by previous homogeneous graph models. Additional experiments reveal that our pretraining strategy not only addresses the challenge of limited labeled data but also significantly enhances accuracy, affirming the potential of our approach in advancing neuroimaging analysis.

25 May 2023 in IEEE Transactions on Neural Networks and Learning Systems
Verified
Journal Article
Closed access
GSGen Shi
Gen Shi
Yifan ZhuYifan Zhu
Yifan Zhu
JLJian K. Liu
Jian K. Liu
XLXuesong Li
Xuesong Li
Representation learning in heterogeneous graphs with massive unlabeled data has aroused great interest. The heterogeneity of graphs not only contains rich information, but also raises difficult barriers to designing unsupervised or self-supervised learning (SSL) strategies. Existing methods such as random walk-based approaches are mainly dependent on the proximity information of neighbors and lack the ability to integrate node features into a higher-level representation. Furthermore, previous self-supervised or unsupervised frameworks are usually designed for node-level tasks, which are commonly short of capturing global graph properties and may not perform well in graph-level tasks. Therefore, a label-free framework that can better capture the global properties of heterogeneous graphs is urgently required. In this article, we propose a self-supervised heterogeneous graph neural network (GNN) based on cross-view contrastive learning (HeGCL). The HeGCL presents two views for encoding heterogeneous graphs: the meta-path view and the outline view. Compared with the meta-path view that provides semantic information, the outline view encodes the complex edge relations and captures graph-level properties by using a nonlocal block. Thus, the HeGCL learns node embeddings through maximizing mutual information (MI) between global and semantic representations coming from the outline and meta-path view, respectively. Experiments on both node-level and graph-level tasks show the superiority of the proposed model over other methods, and further exploration studies also show that the introduction of nonlocal block brings a significant contribution to graph-level tasks.

6 February 2025 in IEEE Transactions on Instrumentation and Measurement
Verified
Journal Article
Closed access
GSGen Shi
Gen Shi
LYLin Yin
Lin Yin
ZBZhongwei Bian
Zhongwei Bian
ZCZiwei Chen
Ziwei Chen
YAYu An
Yu An
JTJie Tian
Jie Tian
Magnetic particle imaging (MPI) has emerged as a promising medical imaging technique known for its high sensitivity and high imaging speed, making real-time, in vivo imaging feasible. However, existing MPI systems often require multiple repetition measurements for signal denoising. Few repetitions may result in low-quality images with increased noise, whereas many repetitions compromise temporal resolution and may introduce significant motion artifacts in dynamic imaging. Therefore, to fully exploit the advantages of MPI in real-time imaging, it is crucial to reduce the repetition number while maintaining high-quality images. In this study, we introduced a novel deep-learning (DL)-based approach, the content-aware distillation network (CAD-Net), for accelerated MPI. The method reconstructs high-quality images by denoising noisy images, typically acquired with a limited number of repetitions (tens of milliseconds). CAD-Net incorporates a proposed multiscale content-aware (MCA) block to accurately model noise distribution and enhance denoising performance. In addition, we proposed an activation-mask-based distillation strategy to reduce model processing time, particularly important for real-time imaging. Evaluation on a public real-world dataset, OpenMPI, and a simulation dataset, proved that CAD-Net outperformed existing methods in denoising performance and model efficiency. Compared to traditional methods based on multiple measurements, CAD-Net increased the frames per second (FPS) metric by approximately 70 times. Experiments on in-house data demonstrated the applicability of CAD-Net in MPI denoising in in vitro and in vivo imaging. CAD-Net improved image quality in real-time denoising with only a marginal increase in time cost. The code and data will be available at: https://github.com/shigen-StoneRoot/CAD-Net.git.

2 May 2025 in IEEE Transactions on Image Processing
Verified
Journal Article
Closed access
WZWenxuan Zou
Wenxuan Zou
GSGen Shi
Gen Shi
SLSiao Lei
Siao Lei
GLGuanghui Li
Guanghui Li
GZGuangxing Zhou
Guangxing Zhou
JTJie Tian
Jie Tian
Recently, Magnetic Particle Imaging, an emerging functional imaging modality, has exhibited outstanding spatial-temporal resolution and sensitivity. The general reconstruction pipeline of Magnetic Particle Imaging involves calibrating a System Matrix and then solving an ill-posed inverse problem combined with the measured particle signals. However, the introduction of noise during the System Matrix calibration procedure is inevitable, which degrades the detailed information in the reconstructed images. Therefore, frequency selection methods based on signal-to-noise ratio are commonly adopted. However, these methods lead to a decrease in the available high-frequency components, which damages the spatial resolution. To address this problem, we propose an unsupervised memory-guided denoising framework with unpaired noisy-clean System Matrix components, called U-N2C. Specifically, we design a Pattern Memory Block to memorize System Matrix patterns, directed by a position-aware frequency index embedding. Meanwhile, we devise a Noise Memory Block to implicitly approximate noise distributions. With the guidance of our dual memory blocks, we can disentangle the noise and content of the System Matrix in the latent space. Furthermore, benefiting from the ability to model complex noise, our method can generate pseudo but high-quality noisy-clean pairs and further enhance our denoising capability. Experiments on both synthetic and real noise demonstrate that our U-N2C achieves cutting-edge performance compared to other methods. Moreover, we conduct extensive qualitative and quantitative ablation studies to verify the effectiveness of our method. Our code has been available at U-N2C.