Pengxiang Su

Dr. Pengxiang Su

School of Software, Nanchang University, Nanchang, Jiangxi, China
orcid

Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

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

Representative Publications

View All Publications
Verified
Journal Article
Open access
PDF file
XPXing Yue Peng
Xing Yue Peng
Pengxiang SuPengxiang Su
Pengxiang Su
YGYaxin Guo
Yaxin Guo
JZJing Zhang
Jing Zhang
LPLinghan Peng
Linghan Peng
RZRongrong Zhang
Rongrong Zhang
Cell-to-cell communication must occur through molecular transport in the intercellular fluid space. Nanoparticles, such as exosomes, diffuse or move more slowly in fluids than small molecules. To find a microfluidic technology for real-time exosome experiments on intercellular communication between living cells, we use the microfluidic culture dish’s quaternary ultra-slow microcirculation flow field to accumulate nanoparticles in a specific area. Taking stem cell–tumor cell interaction as an example, the ultra-slow microcirculatory flow field controls stem cell exosomes to interfere with tumor cells remotely. Under static coculture conditions (without microfluidics), the tumor cells near stem cells (<200 µm) show quick breaking through from its Matrigel drop to meet stem cells, but this ‘breaking through’ quickly disappears with increasing distance. In programmed ultra-slow microcirculation, stem cells induce tumor cells 5000 μm far at the site of exosome deposition (according to nanoparticle simulations). After 14 days of programmed coculture, the glomeration and migration of tumor cells were observed in the exosome deposition area. This example shows that the ultra-slow microcirculation of the microfluidic culture dish has good prospects in quantitative experiments to study exosome communication between living cells and drug development of cancer metastasis.

1 December 2025 in IEEE Internet of Things Journal
Verified
Journal Article
Closed access
GRGuoqiang Ren
Guoqiang Ren
QWQi Wang
Qi Wang
JTJieying Tu
Jieying Tu
Pengxiang SuPengxiang Su
Pengxiang Su
HLHua Li
Hua Li
SLShuxiao Li
Shuxiao Li
In IoT-enabled intelligent healthcare systems, medical images are frequently acquired in real time from heterogeneous imaging sensors such as dermoscopic devices and MRI scanners. In resource-limited or edge-deployed settings, achieving precise and rapid image segmentation plays a crucial role in facilitating early diagnosis and supporting clinical decision-making. This article proposes a medical image segmentation method based on a residual Mamba backbone network, combining a multiscale gated attention (MGA) module with a boundary enhancement (BE) module to effectively enhance the model’s feature representation and boundary localization capabilities. Specifically, the R-Mamba backbone network combines the advantages of state-space modeling and convolutional feature extraction, achieving efficient fusion of global context and local details. The MGA module dynamically captures multiscale semantic information through dilated convolutions and gating mechanisms, enhancing the model’s adaptability to targets of different scales and shapes. The BE module significantly strengthens boundary representation and fine-grained structural segmentation through multiscale convolutions and channel–spatial dual attention mechanisms. Additionally, this article designs a multiloss function joint optimization strategy to comprehensively constrain region overlap, pixel classification, and structural consistency. Experimental validation on ISIC skin lesion and LGG brain tumor datasets shows competitive performance compared to several mainstream models under the tested conditions.

24 November 2025 in IEEE Journal of Biomedical and Health Informatics
Verified
Journal Article
Closed access
ZLZhiwen Liao
Zhiwen Liao
QWQi Wang
Qi Wang
XTXinyi Tang
Xinyi Tang
HWHan Wang
Han Wang
JHJun Hu
Jun Hu
HLHua Li
Hua Li
Image segmentation is a critical technology to improve the accuracy of clinical decisions and treatments in computer-aided diagnostic systems. However, the diverse morphology and fuzzy boundaries of gastrointestinal tumors incur substantial challenges for existing segmentation models, leading to inaccurate feature capture and generating suboptimal results. For solving these problems, we design an edge-guided multi-scale frequency attention network for the gastrointestinal tumor segmentation task, termed EGMFA-Net, which consists of a Kernel Adaptive Enhancement Module (KAEM) and a Frequency-domain Self-attention Module (FDSA). Specifically, KAEM adaptively adjusts the feature extraction kernel based on the morphology of different lesion regions, which enhances the recognition of different morphology regions via a progressive optimization strategy of feature expression. Furthermore, FDSA effectively aggregates multi-scale features in the frequency domain to achieve global receptive fields while preserving more high-frequency details, thereby enhancing adaptability to complex pathological contexts. Extensive experiments on eight medical image benchmark datasets, including SEED, Kvasir, ClinicDB, ColonDB, ETIS, BKAI, CVC-300, and Synapse, show that EGMFA-Net attains state-of-the-art performance over existing methods. Our implementation is available at https://github.com/med-segment/egmfa-net.

15 December 2025 in IEEE Transactions on Consumer Electronics
Verified
Journal Article
Closed access
JYJing Yang
Jing Yang
JDJialin Dai
Jialin Dai
XTXinyi Tang
Xinyi Tang
MKMuhammad Attique Khan
Muhammad Attique Khan
MAManel Ayadi
Manel Ayadi
Lip Yee PorLip Yee Por
Lip Yee Por
Remote patient monitoring (RPM) via smart consumer electronics is revolutionizing post-discharge and chronic disease management. Yet, current AI-powered RPM systems often lack transparency, adaptability across diverse devices, and integration into clinical triage systems. To address this, we introduce TeleXAI-R, an explainable AI-driven risk stratification model that operates in real-time across IoT-integrated consumer devices such as pulse oximeters, blood pressure monitors, and smart rings. TeleXAI-R incorporates a transformer-based temporal encoder for dynamic vital sign prediction and an XAI attention module that visualizes feature importance over time. We develop a cloud-edge hybrid inference engine to enable on-device alerts and clinician dashboards. The model also includes a confidence-aware alerting system to reduce false positives in high-noise settings. Core contributions include: (1) a temporal interpretability framework tailored for RPM; (2) cross-device calibration and data standardization protocol; and (3) integration with FHIR-compatible clinical workflows. Experiments on a real-world dataset of 1,200 patients show 90.1% precision in early risk detection for cardiac events, while maintaining interpretability that clinicians rated 4.7/5 in a usability study. TeleXAI-R sets the foundation for trustworthy RPM in connected healthcare. Unlike prior temporal-attention pipelines, TeleXAI-R couples a rhythm-aware transformer with a calibrated, device-agnostic alignment layer and a confidence-adaptive alerting mechanism that jointly optimize latency, interpretability, and false alerts. Removing any of these core components leads to significant drops in precision and F1, underscoring their necessity.

22 August 2024 in Electronics
Verified
Journal Article
Open access
PDF file
WWWeifeng Wei
Weifeng Wei
JWJie Wang
Jie Wang
XXXie Xiaolong
Xie Xiaolong
JLJie Liu
Jie Liu
Pengxiang SuPengxiang Su
Pengxiang Su
Developing a high-quality, real-time, dense visual SLAM system poses a significant challenge in the field of computer vision. NeRF introduces neural implicit representation, marking a notable advancement in visual SLAM research. However, existing neural implicit SLAM methods suffer from long runtimes and face challenges when modeling complex structures in scenes. In this paper, we propose a neural implicit dense visual SLAM method that enables high-quality real-time reconstruction even on a desktop PC. Firstly, we propose a novel neural scene representation, encoding the geometry and appearance information of the scene as a combination of the basis and coefficient factors. This representation allows for efficient memory usage and the accurate modeling of high-frequency detail regions. Secondly, we introduce feature integration rendering to significantly improve rendering speed while maintaining the quality of color rendering. Extensive experiments on synthetic and real-world datasets demonstrate that our method achieves an average improvement of more than 60% for Depth L1 and ATE RMSE compared to existing state-of-the-art methods when running at 9.8 Hz on a desktop PC with a 3.20 GHz Intel Core i9-12900K CPU and a single NVIDIA RTX 3090 GPU. This remarkable advancement highlights the crucial importance of our approach in the field of dense visual SLAM.

1 January 2025 in Computers, Materials & Continua
Verified
Journal Article
Open access
PDF file
JGJiayi Geng
Jiayi Geng
YWYuxuan Wu
Yuxuan Wu
WLWenbo Lu
Wenbo Lu
Pengxiang SuPengxiang Su
Pengxiang Su
AKAmel Ksibi
Amel Ksibi
DGDi Gai
Di Gai
Predicting human motion based on historical motion sequences is a fundamental problem in computer vision, which is at the core of many applications. Existing approaches primarily focus on encoding spatial dependencies among human joints while ignoring the temporal cues and the complex relationships across non-consecutive frames. These limitations hinder the model’s ability to generate accurate predictions over longer time horizons and in scenarios with complex motion patterns. To address the above problems, we proposed a novel multi-level spatial and temporal learning model, which consists of a Cross Spatial Dependencies Encoding Module (CSM) and a Dynamic Temporal Connection Encoding Module (DTM). Specifically, the CSM is designed to capture complementary local and global spatial dependent information at both the joint level and the joint pair level. We further present DTM to encode diverse temporal evolution contexts and compress motion features to a deep level, enabling the model to capture both short-term and long-term dependencies efficiently. Extensive experiments conducted on the Human 3.6M and CMU Mocap datasets demonstrate that our model achieves state-of-the-art performance in both short-term and long-term predictions, outperforming existing methods by up to 20.3% in accuracy. Furthermore, ablation studies confirm the significant contributions of the CSM and DTM in enhancing prediction accuracy.