The Predictive Value Of CT Imaging Features And Texture Analysis In The Risk Stratifications Of Gastrointestinal Stromal Tumors

Part one: the predictive value of CT imaging features in the risk stratifications of gastrointestinal stromal tumorsObjective To explore the predictive value of CT imaging features in predicting the risk of gastrointestinal stromal tumors(GISTs)before operation.Methods CT images,clinical and pathological data of 81 patients with primary single GISTs were analyzed retrospectively.The correlation between the patients' sex,age,unenhanced and contrast-enhanced CT features and the risk classification of GISTs was analyzed.SPSS 23.0 was used for data analysis,chi square test or ANOVA was used for group comparisons,ROC curve was used for effectiveness evaluation.Results Differences in age,location,size,growth mode,shape,necrosis,gas,enhancement mode,tumor vessels,CT value of arterial phase and enhanced peak were found among different risk levels GISTs(all P<0.05).There are no differences in the gender,calcification,CT value of venous and delayed phase(P>0.05).Conclusion Because of the different CT findings in different risk stratifications GISTs,unenhanced and contrast-enhanced CT scans is helpful to evaluate the risk levels of GISTs before operation.Part two: the predictive value of CT texture analysis in the risk stratifications of gastrointestinal stromal tumorsObjective To explore the predictive value of contrast-enhanced CT texture analysis on the risk stratifications of GISTs before operation.Methods The contrast-enhanced CT images,clinical and pathological data of 81 patients with primary single GISTs confirmed by pathology were analyzed retrospectively.SPSS 23.0 was used for statistical analysis,ANOVA and chi square test were used for comparisons between groups,and ROC curve was used for evaluation of diagnostic efficacy.Results Differences in minimum and entropy were found among GISTs with different risk levels(P<0.05),and there are no differences were found in maximum,average,median,standard deviation,skewness,kurtosis,nonuniformity and percentile(all P>0.05).Minimum,entropy,CT value of arterial phase,enhancement peak,combination of entropy and enhancement peak,and Ki-67 were analysis in differentiating the extremely low/low-risk and medium/high-risk GISTs(AUC=0.643,0.701,0.509,0.661,0.829,0.924,respectively).The correlation coefficients between entropy,enhancement peak and Ki-67 with GISTs risk levels were analysis(r = 0.406,0.341,0.708,respectively).Ki-67 correlated significantly with risk levels,with a cut-off value of 7.5%,and the corresponding sensitivity and specificity were 91.1% and 88.0%,respectively.Conclusion CT texture analysis combined with imaging features has a better evaluation value for the risk assessment of GISTs before operation.