Copyright: ©Author(s) 2026. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution-NonCommercial (CC BY-NC 4.0) license. No commercial re-use. See permissions. Published by Baishideng Publishing Group Inc.
World J Gastroenterol. Nov 7, 2026; 32(41): 120899
Published online Nov 7, 2026. doi: 10.3748/wjg.120899
Published online Nov 7, 2026. doi: 10.3748/wjg.120899
Endoscopic ultrasound-based deep learning for predicting chemotherapy response in unresectable pancreatic ductal adenocarcinoma
Ze-Hua Li, Jun Weng, Shi-Yong Lin, Shuo Li, Kun-Hao Bai, Guo-Liang Xu, Department of Endoscopy, Sun Yat-sen University Cancer Center, State Key Laboratory of Oncology in South China, Guangdong Provincial Clinical Research Center for Cancer, Guangzhou 510060, Guangdong Province, China
Yu-Hong Zeng, Department of Medical Engineering, Zhujiang Hospital of Southern Medical University, Guangzhou 510280, Guangdong Province, China
Shuo Li, State Key Laboratory of Liver Research, The University of Hong Kong, Pokfulam, Hong Kong 999077, China
Co-first authors: Ze-Hua Li and Jun Weng.
Co-corresponding authors: Kun-Hao Bai and Guo-Liang Xu.
Author contributions: Li ZH and Weng J conceived and designed the study; they contributed equally to this work and are the co-first authors; Li ZH led the project implementation, performed the data analysis, developed the deep learning workflow, interpreted the results, and drafted the manuscript; Weng J and Zeng YH coordinated patient recruitment, acquired clinical data, completed tumor image annotation, collected follow-up information, and verified the relevant data; Lin SY assisted with model construction, computational and statistical validation, and figure preparation; Li S was responsible for data curation, image preprocessing, database management, and manuscript editing; Bai KH and Xu GL supervised the study, provided intellectual input, acquired resources and funding support, and critically revised the manuscript, they contributed equally to this article and are the co-corresponding authors; all authors read and approved the final manuscript.
AI contribution statement: The authors used artificial intelligence tools, including ChatGPT, only for language polishing, translation assistance, and writing assistance during manuscript preparation and revision. No artificial intelligence tool was used for study design, data analysis, image generation, interpretation of results, or generation of scientific conclusions. The scientific content, responses to reviewers, and final manuscript were prepared, reviewed, and approved by the authors.
Supported by the National Natural Science Foundation of China (General Program), No. 82403973, No. 82200442, and No. 82373118; Guangdong Basic and Applied Basic Research Foundation, No. 2023A1515010828; Science and Technology Program of Guangzhou, No. 2025A04J3768; Guangdong Medical Equipment Association Research Fund, No. YZXH2025KT07; and Hong Kong Scholar, Hong Kong Scholar, No. XJWQ2025016.
Institutional review board statement: This study was approved by the Medical Ethics Committee of Sun Yat-sen University Cancer Center (Approval No. SL-G2023-244-01).
Informed consent statement: The requirement for informed consent was waived by the Institutional Review Board.
Conflict-of-interest statement: All the authors report no relevant conflicts of interest for this article.
Data sharing statement: The data that support the findings of this study are available from the corresponding author upon reasonable request. Owing to institutional regulations and patient privacy considerations, the data are not publicly available.
Corresponding author: Guo-Liang Xu, MD, PhD, Professor, Chief Physician, Department of Endoscopy, Sun Yat-sen University Cancer Center, State Key Laboratory of Oncology in South China, Guangdong Provincial Clinical Research Center for Cancer, No. 651 Dongfeng East Road, Yuexiu District, Guangzhou 510060, Guangdong Province, China. xugl@sysucc.org.cn
Received: March 12, 2026
Revised: April 20, 2026
Accepted: June 3, 2026
Published online: November 7, 2026
Processing time: 187 Days and 13.9 Hours
Revised: April 20, 2026
Accepted: June 3, 2026
Published online: November 7, 2026
Processing time: 187 Days and 13.9 Hours
Core Tip
Core Tip: Pancreatic ductal adenocarcinoma (PDAC) has a poor prognosis, and predicting chemotherapy response remains challenging. In this study, we developed deep learning models based on pretreatment endoscopic ultrasound images to predict chemotherapy responses in patients with PDAC. Four convolutional neural network architectures were evaluated, with ResNeXt50 demonstrating the best performance in the independent test cohort. The model also enabled effective risk stratification for overall survival, outperforming the conventional serum biomarker carbohydrate antigen 19-9. These findings suggest that endoscopic ultrasound-based convolutional neural network models may provide a noninvasive tool to support individualized treatment planning in PDAC.