Prediction of Peritoneal Metastasis for Gastric Cancer Based on Radiomics

Peritoneal metastasis of gastric cancer is difficult to be detected in time, thus delaying treatment. Based on the conventional CT images of gastric cancer, this study plans to develop, improve and validate an intelligent analysis system based on radiomics. By extracting and combining the radiomics features related to peritoneal metastasis of gastric cancer, the intelligent analysis system could predict the risk of peritoneal metastasis, and provide personalized decision suggestions for the treatment of gastric cancer.

Inclusion Criteria:(1) diagnosed advanced gastric cancer (≥cT3) by endoscopy-biopsy pathology, combined with CT and/or endoscopic ultrasound;(2) with both enhanced CT and laparoscopy;(3) without typical peritoneal metastasis indications in CT (diffuse omental nodules or omental cake, large amount of ascites, obvious irregular thickening with high peritoneal enhancement);(4) without other evidence of distant metastasis, and no stage IV features on CT.Exclusion Criteria:(1) previous abdominal surgery;(2) previous abdominal malignancies or inflammatory diseases;(3) time intervals between CT and laparoscopy longer than 2 weeks;(4) CT image artifacts that undermine peritoneal lesion assessment.