NZNan Zhang

Nan Zhang

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

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Nan Zhang has a total of 29 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • DEER: A Foundation Model with Dual-Branch CT-Enhanced Embeddings for Thoracoabdominal Radiographs

    by Yihua Sun, Jia Guo, Qijing Wang, Jingzhou Ouyang, Ge Li, Nan Zhang, Fang Chen, Hongen Liao

  • Monocular endoscopic light-field visualization via spatiotemporally consistent 3D Gaussian Splatting

    by Yuanqing Yang, Yuwang Wang, Duo Chen, Nan Zhang, Hui Li, Tianqi Huang, Hongen Liao

  • Open Access

    AR-guided emergency surgical navigation via aligned parameterized human model

    by Nan Zhang, Tianqi Huang, Yutong Wu, Hongen Liao, Hongbin Liu

    We propose a lightweight framework that integrates the parametric human model (SMPL) with anatomical priors. The system estimates body shape and poses from a single image, deforms predefined intracranial organ models (e.g., brain tissue and intracranial vasculature) through joint-driven and shape-adaptive transformations, and renders them in an augmented reality environment without requiring external trackers or pre-operative scans. Operating at over 60 FPS, the system provides a real-time AR overlay that remains spatially consistent across varying body shapes and viewing angles. These results demonstrate that parameterized model alignment offers a viable, hardware-light alternative for emergency surgical guidance in resource-limited environments.

  • Comparison of augmented reality visualization approaches in minimally invasive neurosurgery guidance: 2D, tablet, HMD and autostereoscopic displays

    by Nan Zhang, Wentao Zhao, Tianqi Huang, Ming Feng, Hongen Liao, Hongbin Liu

    Minimally invasive neurosurgery presents specific challenges due to the limited operative space and complex cranial anatomy, requiring highly precise and safe surgical guidance. Augmented Reality (AR) technology offers the potential to improve surgical accuracy and safety by overlaying critical digital information onto real-world surgical environments. In this study, we present a study that aims to compare four AR visualization methods—2D flat display, smart tablet, head-mounted display (HMD), and 3D autostereoscopic display—in guiding minimally invasive neurosurgical procedures, specifically focusing on ventriculocentesis. The effectiveness of the AR methods was evaluated through comprehensive user studies involving 32 participants (including 11 experienced surgeons), with assessment focused on critical performance metrics including accuracy, completion time, usability, and cognitive workload during simulated surgical procedures. Results demonstrated that 3D visualization methods significantly outperformed traditional 2D approaches in terms of puncture accuracy and angular precision. Specifically, surgeons showed a statistically significant improvement in localization accuracy, with mean error reduced from 2.69 mm to 1.67 mm, and angular deviation from 5.62° to 1.54°. In comparing the two 3D visualization systems, the HMD exhibited superior task completion efficiency, while the 3D autostereoscopic display demonstrated higher usability scores and lower perceived workload ratings. Notably, the 3D systems effectively reduced the performance disparity between novice and experienced practitioners, suggesting their potential to accelerate the learning curve for less experienced users. We conclude that AR holds significant potential to enhance performance and decision-making in minimally invasive neurosurgical guidance. Graphical Abstract


  • ARMedicalSketch: Exploring 3D Sketching for Medical Image Using True 2D-3D Interlinked Visualization and Interaction

    by Nan Zhang, Tianqi Huang, Hongen Liao

    In traditional clinical practice, doctors often have to deal with 3D information based on 2D-displayed medical images. There is a considerable mismatch between the 2D and 3D dimensions in image interaction during clinical diagnosis, making image manipulation challenging and time-consuming. In this study, we explored 3D sketching for medical images using true 2D-3D interlinked visualization and interaction, presenting a novel AR environment named ARMedicalSketch. It supports image display enhancement preprocessing and 3D interaction tasks for original 3D medical images. Our interaction interface, based on 3D autostereoscopic display technology, provides both floating 3D display and 2D tablet display while enabling glasses-free visualization. We presented a method of 2D-3D interlinked visualization and interaction, employing synchronized projection visualization and a virtual synchronized interactive plane to establish an integrated relationship between 2D and 3D displays. Additionally, we utilized gesture sensors and a 2D touch tablet to capture the user's hand information for convenient interaction. We constructed the prototype and conducted a user study involving 23 students and 2 clinical experts. The controlled study compared our proposed system with a 2D display prototype, showing enhanced efficiency in interacting with medical images while maintaining 2D interaction accuracy, particularly in tasks involving strong 3D spatial correlation. In the future, we aim to further enhance the interaction precision and application scenarios of ARMedicalSketch.


  • A VR Environment for Human Anatomical Variation Education: Modeling, Visualization and Interaction

    by Nan Zhang, Hongkai Wang, Tianqi Huang, Xinran Zhang, Hongen Liao

    Trunk anatomy education is critical in the training of surgeon. Most trunk anatomy education systems use a personal or synthetic anatomical model. It remains difficult to obtain appropriate population anatomy information and create individualized anatomy customization based on a quantity of diverse data. Furthermore, the naked-eye virtual reality modality provides advantages in immersive learning and meta-information transmission. In this study, we present a novel VR torso anatomy teaching environment with naked-eye 3D visualization and free-hand interaction. This work allows for comprehensive modeling, visualization and interaction of torso anatomy for human anatomical variation education. In particular, we derive genuine shape of Chinese male and female torso organs based on the statistical shape model generated from 129 healthy Chinese individuals. The following visual information is given with interaction: (1) Anatomy for human torso organs from CT imaging: geometric modeling, visualization and interaction of male and female torso organs, including skin, muscle, torso skeleton, heart, lungs, liver, spleen and kidneys; (2) Morphological differences of trunk organs from meta-information: motion deformation modeling, visualization and interaction of male and female deformable torso model; (3) Instant personalized customization: real-time interaction and visualization for personalized customization with shape mode coefficients and soft body simulation.


  • Omnidirectional 3D autostereoscopic aerial display with continuous parallax

    by Nan Zhang, Tianqi Huang, Xinran Zhang, Chengquan Hu, Hongen Liao

    We present anl omnidirectional 3D autostereoscopic aerial display with continuous parallax. Integral photography (IP) combined with polyhedron-shaped aerial imaging plates (AIPs) is utilized to achieve an extended view angle of 3D aerial images. With optical theoretical analysis and an aerial in situ rotation design, a 3D aerial display with an enlarged viewing angle is realized. In particular, the proposed 3D aerial display can realize any assigned angle within 360 deg. We also optimize the aerial display with artifact image removal and floating image brightness analysis. Experiments are performed to prove the 3D aerial display with full-motion parallax, continuous viewpoints, and multiplayer interaction. The proposed system is an attractive prospect of non-contact interaction and multi-person collaboration.


  • Research of Webpage Complexity Influence on Search Behavior Based on Eye Tracking Experiment

    by Fan Zhang, Nan Zhang, Na Shan, Feng Gao

    With the development of the Internet, users use electronic devices to browse information continuously for a long time every day. Excessive information brings a lot of cognitive load to users. Many studies have shown that the complexity of the webpage design will affect the search efficiency. For e-commerce webpages, the search efficiency is directly related to the purchase rate. Therefore, through eye tracking experiments, this paper studies the influence of density factors of e-commerce webpage complexity on user search behavior performance and subjective perception, establishes the relationship between webpage visual design and user search usability, and comprehensively constructs evaluation indexes from behavior data and perceptual data. The research results of this paper provide a reference for designers how to arrange information density when designing webpages.

  • Deep Learning Driven Wireless Real-time Human Activity Recognition

    by Hanqing Guo, Nan Zhang, Shaoen Wu, Qing Yang

    Human activity recognition based on wireless sensing is advantageous at various features such as privacy preservation, but also very challenging due to the instability of wireless signals. This paper proposes a deep learning driven wireless human activity recognition solution based on Multiple-Input-Multiple-Output (MIMO) radar sensing. User activities are first sensed by a low-power Frequency-Modulated Continuous Wave (FMCW) MIMO radar. Then a sequence of 3D images are generated out of the reflected signal strength. Next, deep neural networks (DNNs) are designed to analyze the correlation among the sequential 3D images to recognize various types of human activities. This work has developed: 1) a large dataset containing over 1,500 training videos of six different types of indoor activities, 2) a customized deep learning video data-loader to select proper training data in each training epoch, 3) a deep recurrent neural network (RNN) model to recognize human activities based on radar imaging results. This solution has been extensively evaluated in a research lab room. The results show that the solution is able to generate wireless imaging frame-by-frame, and it can achieve over 86.7% accuracy in recognizing six different types of human activities.