News

Generalizable Cervical Cancer Screening via Large-scale Pretraining and Test-Time Adaptation

See our latest work, a new paradigm for cervical cancer screening that leverages large-scale pretraining and test-time adaptation to achieve robust generalization across diverse clinical settings. Our Smart-CCS framework aims to overcome longstanding challenges in cytology-based cancer screening.

Last updated on 2025/02/17

[MIA2025] MedlAnomaly: A Comparative Study of Anomaly Detection in Medical Images

A new survey from HKUST Smartlab presents a comprehensive benchmark for medical anomaly detection, evaluating 30 representative methods across seven diverse datasets.

Last updated on 2025/02/15

[ICLR 2025] GameGen-X: Open-World Game Generation with Diffusion Transformer and Interactive Control

HKUST Smart Lab and USTC present GameGen-X, the first open-world game generation model with interactive control, at ICLR 2025!

Last updated on 2025/02/09

[Press Release] HKUST SmartLab’s AI Medical Models Named Among Top Innovations

HKUST SmartLab's four AI medical models have been selected as one of Hong Kong's Top 10 Innovation and Technology News, providing diagnostic support for 30 types of cancers and diseases.

Last updated on 2025/02/05

[IEEE-TMI] Boosting Convolution with Efficient MLP-Permutation for Volumetric Medical Image Segmentation

A new study on computational pathology from HKUST Smart Lab on permutable hybrid network for 3D volumetric medical image segmentation has been accepted by IEEE Transactions on Medical Imaging.

Last updated on 2025/01/15