Purpose To explore the value of model based on clinical data,spectral CT parameters and radiomics features in predicting perineural invasion of gastric cancer before operation.Materials and Methods A total of 80 patients with gastric cancer who underwent preoperative spectral CT examination in the First Affiliated Hospital of Xinxiang Medical University from January 2021 to August 2022 were retrospectively analyzed.They were divided into perineural invasion positive group and perineural invasion negative group according to the pathological results.The clinicopathological data of patients were collected and the spectral CT parameters of primary lesions of gastric cancer were measured for univariate analysis.214 radiomics features were extracted from biphasic mixed energy images and screened by univariate analysis and support vector machine.Statistically significant variables were included in multivariate Logistic regression analysis to construct a prediction model.The receiver operating characteristic curve was used to evaluate the performance of the model.Results In the clinical data,there were significant differences in gender,CA199,diameter,Lauren type and Borrmann classification between the two groups(all P<0.05).In spectral CT parameters,there were significant differences in CT60 keV-CT110 keV monoenergetic CT values in arterial phase,CT values,iodine concentration,normalized iodine concentration and other monoenergetic CT values except CT80 keV in portal vein phase between the two groups(all P<0.05).The radiomics analysis showed that the support vector machine model with the largest area under curve was chosen,and its area under curve,sensitivity,specificity,accuracy,P-value,and parameters were 0.843,0.923,0.714,0.925,<0.001 and c∶g 2.64∶10.56,respectively.Finally,based on Logistic regression algorithm,clinical model,spectral CT model,radiomics model,clinical+ spectral model,clinical+radiomics model,spectral+radiomics model and clinical+spectral+radiomics model were established to predict the risk of gastric cancer perineural invasion.The diagnostic efficacy of clinical+spectral+radiomics model was the best,and its area under curve,optimal threshold,Youden index,sensitivity and specificity were 0.927(95%CI 0.850-1.000),0.879,0.778,0.778 and 1.000,respectively.Conclusion The combined model based on clinical features,spectral CT parameters and radiomics features is of good value in predicting perineural invasion of gastric cancer before operation.
NETL
NSTL
万方数据
