Analysis Of Pancreatic Neuroendocrine Tumors Based On Texture Analysis Of CT Images

Part ⅠThe value of MDCT image texture analysis in differential diagnosis of pancreatic neuroendocrine tumors and pancreatic cancerObjective:To analyze and compare the enhanced CT texture features of pancreatic neuroendocrine tumors(pNENs)with pancreatic cancer To explore the diagnostic value of MDCT texture parameters in the identification of pNENs and pancreatic cancer.Methods:A retrospective analysis of 34 patients with pancreatic neuroendocrine tumors,34 patients with pancreatic cancer information.All the 68 patients underwent enhanced CT scan.Texture analysis of pancreatic phase images,Apply Histgram Parameters and gray level co-occurrence matrix to record all texture analysis parameters.The differences in the texture features of the 41 images of the pancreas phase and the two tumor markers CA199 and CEA between the two groups were analyzed by MDCT.Results:Pancreatic neuroendocrine tumors and tumor markers of pancreatic cancer(CA199,CEA),Max Intensity of the pancreatic phase texture analysis,Max Intensity,Mean Value,Mean Value,Voxel Value Sum,mean Deviation,Energy,Correlation,and Run Length Non-uniformity.The difference between the two groups was statistically significant(P<0.05).Logistic multivariate regression analysis obtained tumor marker CEA and Mean Value as independent predictors of pancreatic neuroendocrine tumors and pancreatic cancer.The area under the ROC curve of tumor markers CEA and Mean Value for pancreatic cancer and pancreatic neuroendocrine tumors was 0.778(95%CI:0.660-0.896).When the cut-off value was 1.42,the diagnostic sensitivity was 73.3%and the specificity was 75.9%.Conclusions:The CT texture parameters are helpful for the differential diagnosis of neuroendocrine tumors and pancreatic cancer.The tumor markers CEA and Mean Value are independent predictors.Part ⅡThe Value of MDCT Image Texture analysis in the Grading Diagnosis of Pancreatic Neuroendocrine NeoplasmsObjective:To investigate the value of enhanced MDCT image texture analysis in the grading diagnosis of pancreatic neuroendocrine neoplasms(pNENs).Methods:34 cases of pathologically-proven pancreatic neuroendocrine neoplasms(pNENs)between January 2010 and March 2017 were retrospectively analyzed for texture features and clinical features in the arterial phase and the pancreatic phase of contrast-enhanced MDCT.The tumors were classified as G1,G2 and G3 according to the 2010 WHO classification.41 texture features and 2 tumor markers(CA199、CEA)were independently evaluated in different groups.Results:There is no statistic difference in the arterial phase texture features and tumor markers between the G1,G2 and G3 groups and within the groups(P values ranged from 0.130 to 0.939 and P ’values ranged from 0.062 to 1).Texture characteristics and tumor markers were not statistically significant at G1/2 and G3 levels in the arterial phase(P value 0.095-1).Texture characteristics and tumor markers were also not statistically significant at G1 and G2/3 in the arterial phase(P value 0.148-1).The texture features and tumor markers between the G1,G2 and G3 groups and within the groups were no statistic difference in the pancreatic phase(P values ranged from 0.095 to 0.90 and P ’values ranged from 0.032 to 0.949)..The Median Intensity,Meanvalue and Mean Deviation of the tumors were significant variables between the group of G1-2 and G3 in the pancreatic phase(P =0.036,0.034,0.034).The area under the ROC curve of three texture features in identifying G1-2 and G3 are 0.726,0.740,0.740.The sensitivity and specificity of diagnosis were 80.8%and 62.5%respectively with the Median Intensity of 71.48 as the threshold value.The sensitivity and specificity of diagnosis were 76.9%and 62.5%respectively,with an Meanvalue of 71.94 as the threshold value.The Mean deviation of 180.06 was Threshold,diagnostic sensitivity was 62.5%,specificity was 76.9%.Conclusion:The CT texture analysis have some value in the diagnosis of pNENs,mainly in the identification of G1-2 and G3 group in the pancreatic phase.