Related Experiment Video
Updated: Jun 14, 2026
10:28
Gene Regulation and Targeted Therapy in Gastric Cancer Peritoneal Metastasis: Radiological Findings from Dual Energy CT and PET/CT
Published on: January 22, 2018
Habitat and Peritumoral Radiomics with Clinical Variables for Predicting Lymphovascular Invasion in Gastric Cancer: A
Yun Gong1, Qingyun Wang1, Jie Cheng2
1Department of Gastrointestinal Surgery, The Second Affiliated Hospital of Kunming Medical University, Kunming, Yunnan, China (Y.G., Q.W., L.W., Y.L., W.P., H.X., S.Z., F.S.).
Academic Radiology
|June 12, 2026
Summary
This study developed advanced radiomics models to predict lymphovascular invasion (LVI) in gastric cancer. Combining peritumoral radiomics features with clinical data achieved the highest prediction accuracy, improving diagnostic capabilities.
Area of Science:
- Oncology
- Radiology
- Medical Imaging Analysis
Background:
- Lymphovascular invasion (LVI) is a critical prognostic factor in gastric cancer.
- Accurate preoperative prediction of LVI is essential for treatment planning and patient management.
Purpose of the Study:
- To construct and compare radiomics models for preoperative LVI prediction in gastric cancer using intratumoral, peritumoral, and habitat-based strategies.
- To evaluate the incremental value of integrating radiomics models with clinical factors for enhanced prediction.
Main Methods:
- Analysis of 247 gastric adenocarcinoma patients' CT images.
- Development of 68 predictive models including intratumoral, peritumoral (2-mm, 3-mm), and habitat radiomics features.
- Feature-level fusion of intratumoral and peritumoral features, and integration with clinical variables to form a combined predictive model.
Main Results:
- The clinical model showed an AUC of 0.846.
- Peritumoral radiomics features improved prediction, with the 2-mm peritumoral model achieving an AUC of 0.738.
- The combined model integrating the optimal radiomics (IntraPeri feature-level fusion) and clinical variables achieved the highest validation AUC of 0.908.
Conclusions:
- Feature-level fusion of peritumoral 2-mm radiomics features with clinical variables provides the best preoperative prediction of LVI in gastric cancer.
- A nomogram based on this combined model can aid in predicting LVI risk.
