Qing Xia （夏 清）

Portrait of Qing Xia

I am an associate professor at the State Key Laboratory of Virtual Reality Technology and Systems, Beihang University. Prior to this, I served as a senior research manager at SenseTime Research, where I led an R&D team in developing AI-powered medical products. I received my Ph.D. in 2018 from Beihang University under the joint supervision of Prof. Aimin Hao and Prof. Hong Qin. From 2019 to 2021, I pursued postdoctoral research at the iMoon: Intelligent Media and Cognition Lab, Tsinghua University, working with Prof. Yue Gao. My research focuses on computer graphics, computer vision, and artificial intelligence, aiming to develop computational methods that address fundamental challenges in healthcare and biomedicine.

We are looking for self-motivated students with strong CV/CG backgrounds and an interest in healthcare. Please send your CV to xiaqing[AT]buaa[DOT]edu[DOT]cn.

News

[2026-03-30] One paper on foveated ray tracing for virtual reality rendering was published in IEEE TVCG.
[2025-05-20] Our work, ECGFM, a foundation model for ECG analysis, was published in Information Fusion.
[2025-05-13] I joined the School of Computer Science and Engineering, Beihang University as an associate professor.
[2024-12-12] Our work, Slice2Mesh, on 3D cardiac reconstruction from sparse slices, was published in IEEE TMI.
[2024-10-09] One paper on active learning for salient object detection was published in IEEE TPAMI.
[2024-07-30] One paper on coronary segmentation and vectorization was published in IEEE TMI.
[2023-11-30] One paper on a genome-wide association study of the left ventricle was published in Nat. Commun.
[2023-10-14] Our work, AVDNet, on topology-consistent coronary artery segmentation, was published in MedIA.
[2022-06-29] One paper on contrastive learning for medical image segmentation was published in IEEE TMI.
[2021-09-30] I completed my postdoc at the iMoon: Intelligent Media and Cognition Lab, Tsinghua University.
[2021-01-20] One paper on few-shot learning for whole-heart segmentation was published in IEEE TMI.
[2020-07-22] I was selected for the 2020 Beijing Science and Technology Rising Star Program (Talents under 35).
[2018-12-11] I joined SenseTime Research (Beijing) as a full-time research scientist specializing in computer vision.
[2018-09-16] We won first place among 27 participating teams in the Atrial Segmentation Challenge at MICCAI 2018.

Research

Our research integrates artificial intelligence and virtual reality, drawing on computer vision and geometric computing to model biological systems, analyze multimodal biomedical data, and understand complex anatomical structures and surgical environments. We aim to advance biomedical discovery, quantitative diagnosis, and intelligent intervention.

Three overlapping chest X-ray films showing the ribs, lungs, spine and cardiac silhouette.

Medical Image Analysis

We develop data-efficient methods for segmenting, registering, and reconstructing anatomical structures across imaging modalities. Geometric and topological constraints preserve anatomical consistency when annotations or observations are limited. Linking local image features with global organ structure, we recover patient-specific anatomy from incomplete data to support image-based quantitative phenotyping, longitudinal disease assessment, and individualized treatment planning.

Segmentation · Registration · Anatomical Reconstruction · Topological Constraints · Quantitative Phenotyping

The original blue-gray three-dimensional rabbit with sparse blue-gray triangular mesh lines.

3D Vision & Geometric Computing

We develop geometric representations and learning methods for reconstructing, analyzing, deforming, and generating 3D shapes and dynamic scenes. Combining geometric priors with data-driven models, we capture structural relationships while preserving shape fidelity and temporal coherence. We connect shape analysis and generative modeling with motion capture and animation, and explore efficient rendering for virtual reality and interactive simulation of complex dynamic scenes.

3D Reconstruction · Shape Analysis & Deformation · Generative Modeling · Motion & Animation · VR Simulation

Glandular tissue with epithelial folds and a magnifying glass showing cellular and nuclear morphology.

Computational Pathology

We develop multimodal, multiscale methods connecting tissue morphology with molecular profiles, spatial omics, and clinical data to study how cellular composition, tissue organization, and spatial interactions relate to disease progression and patient outcomes. Linking whole-slide images with molecular measurements, we aim to identify biomarkers and characterize tissue phenotypes for cancer diagnosis, prognostic assessment, and evidence-grounded pathology reporting.

Whole-Slide Analysis · Spatial Omics · Tissue Phenotyping · Biomarker Discovery · Cancer Diagnosis & Prognosis

Lung wedge resection with an open incision between two forceps gripping tissue folds and a diagonal linear stapler.

Surgical Intelligence & Intervention

We develop methods for perception, reconstruction, and reasoning in dynamic surgical environments. Combining scene understanding with 3D/4D reconstruction, we model patient anatomy, surgical instruments, and their interactions over time. We integrate spatiotemporal modeling with procedural understanding to relate visual observations to intervention progress, supporting virtual surgery, interactive training, intraoperative decision support, and autonomous surgical systems.

Scene Understanding · 3D/4D Reconstruction · Procedural Understanding · Virtual Surgery · Autonomous Surgery

A pastel cutaway heart with the ascending aorta clearly visible at the pulmonary artery crossing, blood-flow arrows and molecular motifs.

Virtual Organ Modeling

We develop multiscale computational models that integrate multimodal biological data from molecular to organ scales. Combining representation learning with mechanistic priors, we aim to link molecular regulation, cellular behavior, tissue organization, and organ function. We study how these models can predict genetic and pharmacological perturbation responses, uncover disease mechanisms, and support therapeutic target identification, drug discovery, and personalized treatment design.

Multiscale Modeling · Biological Data Integration · Perturbation Responses · Disease Mechanisms · Drug Discovery

Selected Publications

  1. CUFormer pipeline with modality-aware features, visibility-guided correspondences, and rigid registration.

    CUFormer: Modality-Aware CT-Ultrasound Point Cloud Registration for Orthopedic Navigation

    Lixing Tan, Hongye Hao, Tianrun Gu, Qing Xia*, Yuting Guo, Shuai Li, Aimin Hao

    arXiv preprint, arXiv:2611.01536, 2026.

  2. MoPort framework showing paired view alignment, camera-modulated encoding, and plug-and-play inference.

    MoPort: A View-Aware 2D Interface for Motion Language Models

    Lixing Tan, Qing Xia*, Yuting Guo, Shuai Li, Aimin Hao

    arXiv preprint, arXiv:2611.01541, 2026.

  3. STRIDE examples comparing temporal order, mirror reflection, and action identity in three motion sequences and their captions.

    Can Motion-Language Models Ground Structure? STRIDE for Evaluating the Evaluators

    Lixing Tan, Qing Xia*, Yuting Guo, Shuai Li, Aimin Hao

    arXiv preprint, arXiv:2609.32462, 2026.

  4. ECGFM framework with ECG pre-training, task-specific fine-tuning, and transfer to different ECG leads.

    ECGFM: A Foundation Model for ECG Analysis Trained on a Multi-Center Million-ECG Dataset

    Shaoting Zhang, Yishan Du, Wenji Wang, Xianying He, Fangfang Cui, Liang Zhao, Bei Wang, Zhiqiang Hu, Ziqiang Wang, Qing Xia, Tian Shen, Jie Zhao*

    Information Fusion, 124: 103363, 2025.

  5. Slice2Mesh architecture showing sparse MRI inputs, feature sampling, and iterative deformation of a left-ventricular template mesh.

    Slice2Mesh: 3D Surface Reconstruction From Sparse Slices of Images for the Left Ventricle

    Jia Xiao, Wen Zheng, Wenji Wang, Qing Xia*, Zhennan Yan, Qian Guo, Xiao Wang*, Shaoping Nie*, Shaoting Zhang

    IEEE Transactions on Medical Imaging, 44(3): 1541–1555, 2025.

  6. Geometry-based cascaded network with mesh deformation segmentation followed by mesh refinement segmentation.

    Segmentation and Vascular Vectorization for Coronary Artery by Geometry-Based Cascaded Neural Network

    Xiaoyu Yang, Lijian Xu*, Simon Yu, Qing Xia, Hongsheng Li, Shaoting Zhang

    IEEE Transactions on Medical Imaging, 44(1): 259–269, 2025.

  7. Three original Figure 1 panels stacked on the left: enlarged coronary anatomy, an easy case, and a hard case; the complete AVDNet two-stage workflow remains on the right.

    AVDNet: Joint Coronary Artery and Vein Segmentation With Topological Consistency

    Wenji Wang, Qing Xia*, Zhennan Yan, Zhiqiang Hu, Yinan Chen, Wen Zheng, Xiao Wang, Shaoping Nie, Dimitris Metaxas, Shaoting Zhang

    Medical Image Analysis, 91: 102999, 2024.

  8. Myocardial segmentation and measurement workflow from cardiac MRI to regional left-ventricular wall-thickness phenotypes.

    Genome-Wide Association Analysis of Left Ventricular Imaging-Derived Phenotypes Identifies 72 Risk Loci and Yields Genetic Insights Into Hypertrophic Cardiomyopathy

    Caibo Ning, Linyun Fan, …, Qing Xia, …, Shaoting Zhang*, Jianbo Tian*, Xiaoping Miao*

    Nature Communications, 14: 7900, 2023.

  9. Few-shot whole-heart segmentation framework with initial mean-teacher learning, two self-training stages, and shape-constrained pseudo-label assessment.

    Few-Shot Learning by a Cascaded Framework With Shape-Constrained Pseudo Label Assessment for Whole Heart Segmentation

    Wenji Wang, Qing Xia*, Zhiqiang Hu, Zhennan Yan, Zhuowei Li, Yang Wu, Ning Huang, Yue Gao, Dimitris Metaxas, Shaoting Zhang

    IEEE Transactions on Medical Imaging, 40(10): 2629–2641, 2021.

  10. Left-atrium benchmark workflow from clinical LGE-MRI and expert segmentation to challenge evaluation and clinical use.

    A Global Benchmark of Algorithms for Segmenting the Left Atrium From Late Gadolinium-Enhanced Cardiac Magnetic Resonance Imaging

    Zhaohan Xiong, Qing Xia, Zhiqiang Hu, Ning Huang, Cheng Bian, Yefeng Zheng, Sulaiman Vesal, Nishant Ravikumar, …, Nicolo Savioli, Pablo Lamata, Jichao Zhao*

    Medical Image Analysis, 67: 101832, 2021.

For a complete and up-to-date publication list, please visit Google Scholar or DBLP.

Last updated: 2026-11-15