The Value Of Dual Energy CT Quantitative Parameters And CT-based Radiomics In Preoperative Prediction Of Lymph Node And Occult Peritoneal Metastasis In Gastric Cancer

Objective: To investigate the value of dual energy CT(DECT)quantitative parameters in predicting preoperative lymph node metastasis(LNM)in advanced gastric cancer(AGC).Methods: This study retrospectively collected the preoperative clinical data and DECT images of 180 patients with advanced gastric cancer in our hospital.According to the postoperative pathological results,all patients were divided into lymph node metastasis group(n=115)and non metastasis group(n=65).Through GSI viewer software,the DECT quantitative parameters were measured and the imaging features were evaluated at the same time.The independent risk predictors in the clinical-imaging variables and DECT quantitative parameters were screened out to construct the clinical model and the DECT parameters model respectively,and then integrating clinical model and DECT parameters model to build the combined model.The receiver operating characteristic curve(ROC)and area under the curve(AUC)were used to evaluate the diagnostic effectiveness of the model.Delong test was performed to compare whether the difference of the AUCs among models have statistical significance.Results: The clinical model included three variables: CEA,tumor thickness and fat infiltration,the DECT parameters model included one variable: standardized iodine concentration(n IC)of portal phase,the combined model included four variables:CEA,tumor thickness,fat infiltration and standardized iodine concentration(n IC)of portal phase.The AUCs of clinical model,DECT parameters model and combined model were 0.703,0.772,0.820,respectively,the sensitivity and specificity of clinical model,DECT parameters model and combined model were 62.6%,67.7%,75.7% and75.4%,87.7%,76.9%,respectively.The results indicated that the combined model had the highest diagnostic efficiency and the Delong test showed that there was significant difference in AUCs among three models(P<0.05).Conclusion: This study showed that DECT quantitative parameters can more accurately predict lymph node metastasis in advanced gastric cancer before surgery,providing more evidence for clinical decision-making.Objective: To investigate the value of CT-based radiomics in preoperative prediction of occult peritoneal metastasis in advanced gastric cancer(AGC).Methods: This study retrospectively collected the clinical and imaging data of 110 patients with advanced gastric cancer confirmed by postoperative pathology,all patients had negative preoperative CT diagnosis of peritoneal status,and the postoperative pathological results showed that 47 patients were with positive peritoneal metastasis.110 patients were randomly divided into a training group(n=77)and a validation group(n=33)in a ratio of 7:3.Using Darwin platform,manually draw the region of interest(ROI)on patients’ arterial and portal phase images to extract radiomics features and construct arterial phase radiomics model and portal phase radiomics model,respectively.Clinical model were constructed using independent risk predictors in patients’ clinical variables.Then combining clinical model and radiomics model to construct the combined model and visualizing it by drawing nomograph.The receiver operating characteristic curve(ROC),correction curve and decision curve analysis(DCA)were used to evaluate the diagnostic efficiency and clinical application value of the three models.Delong test was performed to compare whether the difference of the AUCs among models have statistical significance.Results: Compared with portal phase radiomics model,the arterial phase radiomics model showsed better predictive performance in both the training group and the validation group,so combining arterial phase radiomics model with clinical model to construct the combined model.In training group and validation group,the AUCs of clinical model,radiomics model and combined model was 0.780,0.844,0.921 and0.865,0.865,0.944,respectively.Meanwhile,Delong test showed there were significant differences in AUC among three models(P<0.05).The calibration curve and decision curve analysis(DCA)show that the combined model has good consistency between the predicted probability and the true probability,and has higher net benefit within a certain threshold,with the highest prediction efficiency.Conclusion: The CT-based radiomics can more accurately predict whether patients with advanced gastric cancer develop occult peritoneal metastasis before surgery,providing more basis for clinical decision-making.