Surgical Video Understanding
Online, causal recognition of surgical workflow — from single frames to multi-hour procedures. Long-video transformers, key-information compression and stable temporal inference.
01
About
I received my PhD from King's College London (School of Biomedical Engineering & Imaging Sciences) in January 2026, supervised by Prof. Sébastien Ourselin (primary supervisor), Prof. Prokar Dasgupta and Dr. Alejandro Granados. I received my Master's degree from Huazhong University of Science and Technology in 2021, advised by Prof. Xiang Bai.
My research builds AI systems that understand surgical video in real time — online phase recognition over multi-hour procedures, streaming vision-language models for the operating room, and label-efficient segmentation of instruments and anatomy. I also work on broader medical imaging problems (echocardiography, stroke lesion segmentation, structural MRI) and on vision-language learning with weak supervision.
From January to July 2025 I worked part-time as a Data Scientist at Proximie, bringing surgical-video models closer to clinical deployment.
0
Peer-reviewed papers & preprints
0
Top-tier venues
MICCAI · CVPR · ICCV · MedIA · WACV
MICCAI · CVPR · ICCV · MedIA · WACV
0st
MICCAI 2022 ATLAS Stroke Lesion Segmentation Challenge
0%
MICCAI 2026 early-accept tier (StableSPR)
02
Research
Vision-Language Models
Streaming VLMs for the operating room, weakly supervised referring comprehension & segmentation, and noise-injected cross-modal alignment for language-driven visual tasks.
Label-efficient Segmentation
Unsupervised surgical instrument segmentation from low-quality optical flow, fast boundary-to-pixel direction segmentation, and diffusion-based crack segmentation.
Medical Imaging
Training-free cardiac phase detection in echocardiography, ischemic stroke lesion segmentation, and gray-matter-guided attention for Alzheimer's diagnosis from structural MRI.
03
Publications
* equal contribution · † corresponding author
04
News
05
Experience & Education
2021 — 2026
King's College London
PhD · School of Biomedical Engineering & Imaging Sciences (awarded January 2026)
Surgical video understanding, streaming vision-language models, label-efficient medical segmentation. Primary supervisor: Prof. Sébastien Ourselin; co-supervised by Prof. Prokar Dasgupta and Dr. Alejandro Granados.
Jan — Jul 2025
Proximie
Data Scientist (part-time)
Surgical-video machine learning for a clinical telepresence and analytics platform.
2021
ByteDance
Computer Vision Algorithm Engineer (intern) · Education department
Applied computer vision for education products.
2019 — 2021
Huazhong University of Science and Technology
MSc · advised by Prof. Xiang Bai
Image segmentation and image restoration; publications at CVPR 2020 and WACV 2021.
06
Let's collaborate
Open to collaboration on surgical AI, medical video understanding and vision-language models.
visitors
—unique visitors
—page views