Interpretable deep learning for multicenter gastric cancer T staging from CT images - Summary - MDSpire

Interpretable deep learning for multicenter gastric cancer T staging from CT images

  • By

  • Guoliang Zheng

  • Huan Wang

  • Xiaomiao Chai

  • Xin Xin

  • Fuze Li

  • Hongfei Li

  • Yaoyang Ban

  • Jinshi Wang

  • Xinhui Qi

  • Yingjie Li

  • Zishuo Yan

  • Fangning Guo

  • Zhixue Jiang

  • Dantong Zhu

  • Yanqiang Zhang

  • Zhendong Zheng

  • Xin Zhang

  • Jing Zhang

  • Yan Zhao

  • December 20, 2025

  • 0 min

Share

Objective:

To develop and evaluate GTRNet, an interpretable deep-learning framework specifically designed for accurate T staging of gastric cancer using routine CT imaging.

Key Findings:
  • GTRNet achieved high discrimination (AUC 0.86–0.95) and accuracy (81–85%) in internal and external tests.
  • The model outperformed radiologists in T staging accuracy.
  • Grad-CAM heatmaps effectively localized tumor invasion areas.
Interpretation:

GTRNet provides a reliable and interpretable method for T staging of gastric cancer, potentially improving clinical decision-making and treatment planning significantly.

Limitations:
  • The study is retrospective and may be subject to selection bias.
  • Performance may vary with different imaging protocols or populations.
  • Further validation in diverse populations is needed.
Conclusion:

GTRNet represents a significant advancement in automated and interpretable T staging of gastric cancer, supporting standardized preoperative assessments.

Original Source(s)

Interpretable deep learning for multicenter gastric cancer T staging from CT images

  • by Guoliang Zheng, Huan Wang, Xiaomiao Chai, Xin Xin, Fuze Li, Hongfei Li, Yaoyang Ban, Jinshi Wang, Xinhui Qi, Yingjie Li, Zishuo Yan, Fangning Guo, Zhixue Jiang, Dantong Zhu, Yanqiang Zhang, Zhendong Zheng, Xin Zhang, Jing Zhang, Yan Zhao

  • December 20, 2025

Related Content

Deep learning model for assessing survival benefits in hepatocellular carcinoma patients undergoing intra-arterial therapies based on proliferative subtype

Anatomically-guided Masked Autoencoder with Domain-Adaptive Prompting (AMAP) for multimodal cerebral aneurysm detection and segmentation

Automated real-time assessment of intracranial hemorrhage detection AI using an ensembled monitoring model (EMM)