Paul Allen School of Computer Science and Engineering, University of WashingtonI am Hanwen Xu, a Ph.D. student at the University of Washington, working at the intersection of gigascale AI and precision medicine, advised by Prof. Sheng Wang. My research builds large-scale multimodal foundation models that integrate pathology, radiology, and clinical records to decipher the complex cancer patient journey. I led/co-led projects such as GigaPath, GigaTIME, LLaVA-Rad and BiomedCLIP, aiming to advance multimodal AI for precision medicine and real-world deployments. Before my PhD, I earned a BS and MS in Department of Automation in Tsinghua University.
Warning
Problem: The current name of your GitHub Pages repository ("Solution: Please consider renaming the repository to "
http://".
However, if the current repository name is intended, you can ignore this message by removing "{% include widgets/debug_repo_name.html %}" in index.html.
Action required
Problem: The current root path of this site is "baseurl ("_config.yml.
Solution: Please set the
baseurl in _config.yml to "Education
-
University of WashingtonPaul Allen School of Computer Science and Engineering, University of Washington
Ph.D. StudentSep. 2022 - present -
Bioinformatics Division, Tsinghua UniversityM.S. in AutomationSep. 2020 - Jul. 2022 -
Department of Automation, Tsinghua UniversityB.S. in AutomationSep. 2016 - Jul. 2020
Honors & Awards
-
Weil Family Endowed Fellowship in Computer Science and Engineering2023
-
The Scholarship for Comprehensive Outstanding Performance of Graduate Students, Tsinghua University2021
-
Merit Student, Tsinghua University2020
-
National Scholarship of China2019
-
Tsinghua Innovation Award of Science and Technology2019
News
Selected Publications (view all )
Multimodal AI generates virtual population for tumor microenvironment modeling
Jeya Maria Jose Valanarasu*, Hanwen Xu*, Naoto Usuyama*, Chanwoo Kim*, Cliff Wong, Peniel Argaw, Racheli Ben Shimol, Angela Crabtree, Kevin Matlock, Alexandra Q. Bartlett, Jaspreet Bagga, Yu Gu, Sheng Zhang, Tristan Naumann, Bernard A. Fox, Bill Wright, Ari Robicsek, Brian Piening, Carlo Bifulco, Sheng Wang, Hoifung Poon (* equal contribution)
Cell 2025
ChatGPT said:The tumor immune microenvironment (TIME) shapes cancer progression and immunotherapy response, but mIF imaging is too expensive and slow for large populations. GigaTIME is a multimodal AI system that generates virtual mIF from routine H&E slides, trained on 40 million cells across 21 proteins. We applied it to 14,256 patients from 51 hospitals, producing 299k virtual mIF slides across 24 cancers and revealing 1,234 significant clinical associations previously impossible to detect. Independent validation on 10,200 TCGA patients confirmed these insights.
Multimodal AI generates virtual population for tumor microenvironment modeling
Jeya Maria Jose Valanarasu*, Hanwen Xu*, Naoto Usuyama*, Chanwoo Kim*, Cliff Wong, Peniel Argaw, Racheli Ben Shimol, Angela Crabtree, Kevin Matlock, Alexandra Q. Bartlett, Jaspreet Bagga, Yu Gu, Sheng Zhang, Tristan Naumann, Bernard A. Fox, Bill Wright, Ari Robicsek, Brian Piening, Carlo Bifulco, Sheng Wang, Hoifung Poon (* equal contribution)
Cell 2025
ChatGPT said:The tumor immune microenvironment (TIME) shapes cancer progression and immunotherapy response, but mIF imaging is too expensive and slow for large populations. GigaTIME is a multimodal AI system that generates virtual mIF from routine H&E slides, trained on 40 million cells across 21 proteins. We applied it to 14,256 patients from 51 hospitals, producing 299k virtual mIF slides across 24 cancers and revealing 1,234 significant clinical associations previously impossible to detect. Independent validation on 10,200 TCGA patients confirmed these insights.
A whole-slide foundation model for digital pathology from real-world data
Hanwen Xu*, Naoto Usuyama*, Jaspreet Bagga, Sheng Zhang, Rajesh Rao, Tristan Naumann, Cliff Wong, Zelalem Gero, Javier González, Yu Gu, Yanbo Xu, Mu Wei, Wenhui Wang, Shuming Ma, Furu Wei, Jianwei Yang, Chunyuan Li, Jianfeng Gao, Jaylen Rosemon, Tucker Bower, Soohee Lee, Roshanthi Weerasinghe, Bill J. Wright, Ari Robicsek, Brian Piening, Carlo Bifulco, Sheng Wang, Hoifung Poon (* equal contribution)
Nature 2024 2,000,000 downloads
Prov-GigaPath is a large-scale digital pathology foundation model pretrained on 1.3 billion image tiles from 171,189 whole-slide images across 30,000+ patients and 31 tissue types in the Providence network. Built on a new GigaPath architecture that adapts LongNet for slide-level learning, it enables ultra-large-context modeling of gigapixel pathology slides. Prov-GigaPath achieves state-of-the-art results on 25 of 26 benchmark tasks and demonstrates strong potential for vision–language pathology modeling using real-world data.
A whole-slide foundation model for digital pathology from real-world data
Hanwen Xu*, Naoto Usuyama*, Jaspreet Bagga, Sheng Zhang, Rajesh Rao, Tristan Naumann, Cliff Wong, Zelalem Gero, Javier González, Yu Gu, Yanbo Xu, Mu Wei, Wenhui Wang, Shuming Ma, Furu Wei, Jianwei Yang, Chunyuan Li, Jianfeng Gao, Jaylen Rosemon, Tucker Bower, Soohee Lee, Roshanthi Weerasinghe, Bill J. Wright, Ari Robicsek, Brian Piening, Carlo Bifulco, Sheng Wang, Hoifung Poon (* equal contribution)
Nature 2024 2,000,000 downloads
Prov-GigaPath is a large-scale digital pathology foundation model pretrained on 1.3 billion image tiles from 171,189 whole-slide images across 30,000+ patients and 31 tissue types in the Providence network. Built on a new GigaPath architecture that adapts LongNet for slide-level learning, it enables ultra-large-context modeling of gigapixel pathology slides. Prov-GigaPath achieves state-of-the-art results on 25 of 26 benchmark tasks and demonstrates strong potential for vision–language pathology modeling using real-world data.