ZZZhijing Zhu

Zhijing Zhu

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Zhijing Zhu has a total of 27 co-authors. Here are the top co-authors by number of shared publications:

  • Ruiwen Wang

    Key Laboratory of Novel Targets and Drug Study for Neural Repair of Zhejiang Province, Brain and Cognition Research Institute, School of Medicine, Hangzhou City University, Hangzhou, China

    4 shared publications
  • Bobo Cai

    Zhejiang Hospital Hangzhou China

    3 shared publications
  • Shuyue Wang

    Key Laboratory of Novel Targets and Drug Study for Neural Repair of Zhejiang Province, Brain and Cognition Research Institute, School of Medicine, Hangzhou City University, Hangzhou, China

    3 shared publications
  • Bingzhi Lin

    College of Energy and Power Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing, China

    2 shared publications
  • Diming Zhang

    Research Center for Intelligent Sensing Systems, Zhejiang Laboratory, Hangzhou, China

    2 shared publications
  • Gelang Hu

    Key Laboratory of Novel Targets and Drug Study for Neural Repair of Zhejiang Province, Brain and Cognition Research Institute, School of Medicine, Hangzhou City University, Hangzhou, China

    2 shared publications

Latest Publications

  • Open Access

    DeepMoCo: graph neural network-based adaptive correction for motion artifacts in fluorescence microscopy

    by Shuyue Wang, Ruiwen Wang, Gelang Hu, Zhijing Zhu

    Fluorescence microscopy has emerged as an indispensable tool in neuroscience research, enabling subcellular-resolution imaging of neural circuits and monitoring of brain function in vivo. However, physiological motion artifacts which include both rigid displacements and non-rigid deformations induced by respiration and tissue dynamics, significantly compromise imaging fidelity and quantitative analysis reliability. In this paper, we propose DeepMoCo, an innovative deep learning-based framework that integrates a graph neural network (GNN) to establish spatiotemporal feature correlations between biological motion patterns and imaging artifacts. This architecture enables rapid tracking and adaptive correction of complex motion patterns, achieving 95% improvement in correlation coefficients and 77% enhancement in peak signal-to-noise ratio compared to existing methods. The proposed method presents a robust and universal solution for high-precision in vivo fluorescence imaging studies.

  • Data‐Driven Computational Super‐Resolution Fluorescence Microscopy for Live‐Cell Imaging

    by Xinyu Lu, Ruiwen Wang, Gelang Hu, Zhijing Zhu, Bobo Cai

    Fluorescence microscopy has emerged as an essential tool for dynamic high‐resolution imaging in life science, owing to its capacity for in vivo observation, precise molecular labeling, and subcellular resolution. This technique encounters two primary challenges in super‐resolution imaging: overcoming the optical diffraction limit in spatial domain to discern intricate biological structures while enhancing imaging speed in temporal domain to capture millisecond‐scale dynamic processes; adapting to complex biological environments for stable observations while minimizing phototoxicity to preserve sample viability. Despite traditional super‐resolution microscopy surpassing the diffraction limit, its reliance on intricate optical systems and high‐power excitation light leads to constrained imaging speed and considerable photodamage, hindering the capacity for prolonged dynamic observation of living samples, thereby rendering deep learning a potential solution. Here, this review systematically assesses the advancements in deep learning enabled super‐resolution fluorescence microscopy, including learning‐strategy (supervised, unsupervised, and zero‐shot) and integration‐strategy based methods. The performance and technical advantages of diverse methods in fluorescence microscopy are thoroughly investigated. This work offers valuable insights for researchers seeking to implement deep learning‐enhanced imaging techniques, potentially accelerating breakthroughs in bioimaging processing and neuroscience research.

  • Open Access

    Deep Learning Unlocks Behavioral Prediction and Neurobehavioral Decoding

    by Tianzhe Han, Yunhua Jia, Xiaomin Hu, Shuyue Wang, Ruiwen Wang, Zhijing Zhu

    Forecasting behavior and uncovering its underlying neuronal mechanisms are two pivotal pillars in understanding biological behaviors. The former focuses on predicting future actions and patterns through observational data, whereas the latter delves into the neural circuits and computations that give rise to behavior. The integration of behavioral characterization with cerebral exploration lays a robust foundation, both theoretical and methodological. It enables the investigation of decision‐making processes, as well as the advancement of brain‐computer interfaces (BCIs). However, conventional approaches are plagued by inherent limitations: manually crafted features lack sufficient representational power, linear temporal models fail to capture complex dynamic patterns, and inflexible physical priors restrict generalizability. These drawbacks severely undermine predictive accuracy and real‐time performance, rendering traditional methodologies incapable of meeting the evolving demands of modern neuroscience and artificial intelligence. Deep learning has emerged as a revolutionary framework, excelling in high‐dimensional nonlinear modeling, capturing complex spatiotemporal dependencies, integrating multimodal information, and enabling cross‐scale representation learning through sophisticated hierarchical feature extraction, scalable training paradigms, and the integration of interpretable AI methodologies. This study systematically integrates philosophical and technological advancements in deep learning. The aim is twofold: to predict biological behavior, and to decode neural‐behavioral correlations. We conduct a comprehensive analysis of its applications in real‐time action forecasting, trajectory prediction in both closed and open‐world scenarios, and the elucidation of multi‐scale brain mechanisms. Furthermore, we evaluate the efficacy, benefits, and ongoing obstacles of example methodologies, offering insights to guide future multidisciplinary advancements at the convergence of neuroscience, artificial intelligence, and computational behavioral science.


  • Open Access

    Deep learning‐driven methods for fluorescence imaging denoising

    by Xinyu Lu, Zixu Zhan, Bin Tan, Ruiwen Wang, Shuyue Wang, Diming Zhang, Bobo Cai, Zhijing Zhu

    Fluorescence imaging, serving as the primary imaging modality in modern life science research, faces a fundamental challenge in achieving high‐sensitivity imaging: optimizing the signal‐to‐noise ratio (SNR) under dynamic and complex experimental conditions. Due to autofluorescence, shot noise, and tissue scattering, this SNR deficiency disrupts subcellular morphometry, restricts recording reliability, and ultimately propagates artifacts in subsequent analysis. This review evaluates data‐driven deep learning denoising methods that overcome conventional limitations through effective feature extraction and nonlinear modeling. Focusing on fluorescence imaging acquisition under photon‐limited conditions, we delineate cutting‐edge architectures, including supervised learning, unsupervised learning, zero‐shot learning, and hybrid approaches. By producing higher‐fidelity image data, these denoising methods enhance the reliability of live‐cell imaging and the accuracy of neural mechanism analysis. This advancement provides a stronger foundation for elucidating dynamic biological processes and accelerating precision medicine.


  • Open Access

    Reveal the mechanism of brain function with fluorescence microscopy at single-cell resolution: from neural decoding to encoding

    by Kangchen Li, Huanwei Liang, Jialing Qiu, Xulan Zhang, Bobo Cai, Depeng Wang, Diming Zhang, Bingzhi Lin, Haijun Han, Geng Yang, et al.

    As a key pathway for understanding behavior, cognition, and emotion, neural decoding and encoding provide effective tools to bridge the gap between neural mechanisms and imaging recordings, especially at single-cell resolution. While neural decoding aims to establish an interpretable theory of how complex biological behaviors are represented in neural activities, neural encoding focuses on manipulating behaviors through the stimulation of specific neurons. We thoroughly analyze the application of fluorescence imaging techniques, particularly two-photon fluorescence imaging, in decoding neural activities, showcasing the theoretical analysis and technological advancements from imaging recording to behavioral manipulation. For decoding models, we compared linear and nonlinear methods, including independent component analysis, random forests, and support vector machines, highlighting their capabilities to reveal the intricate mapping between neural activity and behavior. By employing synthetic stimuli via optogenetics, fundamental principles of neural encoding are further explored. We elucidate various encoding types based on different stimulus paradigms—quantity encoding, spatial encoding, temporal encoding, and frequency encoding—enhancing our understanding of how the brain represents and processes information. We believe that fluorescence imaging-based neural decoding and encoding techniques have deepened our understanding of the brain, and hold great potential in paving the way for future neuroscience research and clinical applications.


  • Open Access

    Real-time and universal network for volumetric imaging from microscale to macroscale at high resolution

    by Bingzhi Lin, Feng Xing, Liwei Su, Kekuan Wang, Yulan Liu, Diming Zhang, Xusan Yang, Huijun Tan, Zhijing Zhu, Depeng Wang

    Light-field imaging has wide applications in various domains, including microscale life science imaging, mesoscale neuroimaging, and macroscale fluid dynamics imaging. The development of deep learning-based reconstruction methods has greatly facilitated high-resolution light-field image processing, however, current deep learning-based light-field reconstruction methods have predominantly concentrated on the microscale. Considering the multiscale imaging capacity of light-field technique, a network that can work over variant scales of light-field image reconstruction will significantly benefit the development of volumetric imaging. Unfortunately, to our knowledge, no one has reported a universal high-resolution light-field image reconstruction algorithm that is compatible with microscale, mesoscale, and macroscale. To fill this gap, we present a real-time and universal network (RTU-Net) to reconstruct high-resolution light-field images at any scale. RTU-Net, as the first network that works over multiscale light-field image reconstruction, employs an adaptive loss function based on generative adversarial theory and consequently exhibits strong generalization capability. We comprehensively assessed the performance of RTU-Net through the reconstruction of multiscale light-field images, including microscale tubulin and mitochondrion dataset, mesoscale synthetic mouse neuro dataset, and macroscale light-field particle imaging velocimetry dataset. The results indicated that RTU-Net has achieved real-time and high-resolution light-field image reconstruction for volume sizes ranging from 300 μm × 300 μm × 12 μm to 25 mm × 25 mm × 25 mm, and demonstrated higher resolution when compared with recently reported light-field reconstruction networks. The high-resolution, strong robustness, high efficiency, and especially the general applicability of RTU-Net will significantly deepen our insight into high-resolution and volumetric imaging.