Grounded knowledge-enhanced medical vision-language pre-training for chest X-ray

Open Webpage

Qiao Deng, Zhongzhen Huang, Yunqi Wang, Zhichuan Wang, Zhao Wang, Xiaofan Zhang, Qi Dou, Yeung Yu Hui, Edward S. Hui

Published: 01 Jan 2026, Last Modified: 12 Oct 2025Biomed. Signal Process. Control. 2026EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•GK-MVLP used fine-grained visual–knowledge alignment for representation learning.•Knowledge prompts enhanced localization and prevented irrelevant information.•GK-MVLP exceeded SOTA in classification, localization, report generation, and VQA.