Artificial Intelligence for Predicting HER2 Status of Gastric Cancer Based on Whole‐Slide Histopathology Images: A Retrospective Multicenter Study

• DOI: 10.1002/advs.202408451
• Publication Date: 2025-01-10T19:10:31Z
• Authors: Yuhan Liao; Xinhua Chen; Shupeng Hu; Bing Chen; Xinghua Zhuo; Hao Xu; Xiaojin Wu; Xiaofeng Zeng; Huimin Zeng; Donghui Zhang; Yunfei Zhi; Liang Zhao
• Keywords: Histopathology; Training set; Identification

Abstract Human epidermal growth factor receptor 2 (HER2) positive gastric cancer (GC) shows a robust response to the combined therapy based HER2‐targeted therapy. The application of these therapies is highly dependent on evaluation tumor HER2 status. However, there are many risks and challenges in assessment GC. Therefore, an economically viable readily available instrument requisite for distinguishing status among patients diagnosed with study has innovatively developed deep learning model, HER2Net, which can predict by quantitatively calculating proportion high‐expression regions. HER2Net trained internal training set derived from 531 hematoxylin & eosin (H&E) whole slide images (WSI) 520 patients. Subsequently, performance validated test 115 H&E WSI 111 external multi‐center 102 101 achieves accuracy 0.9043 set, 0.8922 multiple institutes. This discovery indicates that potentially offer novel methodology identification HER2‐positive