A novel multimodal combining radiomics and tumor-stroma ratio (TSR) improves diagnosis of gastric cancer peritoneal metastasis.

Peritoneal metastasis (PM) is the most common form of metastasis in gastric cancer (GC), frequently leading to severe complications and a significantly poor prognosis. Prompt and early diagnosis of PM in GC is crucial. However, diagnostic laparoscopy and CT scans, while being the primary methods for identifying PM in GC, have notable limitations, such as being invasive and having low sensitivity. Therefore, developing a diagnostic model for PM in GC based on routine examination results holds substantial importance. In this retrospective study, we enrolled 813 patients from two medical centers and developed a robust diagnostic model by integrating various routine examination results, including CT scans, endoscopy, pathology, and hematological tests. To further explore the potential significance of various examination results, we conducted radiomic analysis of CT images, analyzed histopathological results via the Tumor-Stroma Ratio (TSR), and examined hematological results through parameters such as the Prognostic Nutritional Index (PNI), Neutrophil to Lymphocyte Ratio (NLR), and Albumin to Globulin Ratio (AGR). A novel diagnostic model, incorporating CA125, CA724, Borrmann classification, radiomics features, and the TSR, was successfully constructed. This model demonstrated strong performance in diagnosing synchronous PM (AUC = 0.874) and predicting metachronous (AUC = 0.784) PM in GC. To facilitate clinical application, a nomogram was constructed. Consequently, the study presents a novel and comprehensive diagnostic model for PM in GC patients, offering significant promise for clinical applicability based on routine examination results.