Differentiation Of Solid Pseudopapillary Tumor Of Pancreas From Neuroendocrine Tumor Of Pancreas By CT,MRI And Imaging Omics

Part 1 Imaging features of solid pseudopapilloma of thepancreasObjective:To analyze the computed tomography(CT)and magnetic resonance imaging(MRI)features of solid pseudopapillary tumor of pancreas(SPTP),so as to improve the understanding and diagnosis of SPTP.Methods:The clinical data,and CT and MRI data of 103 patients with SPTP confirmed by postoperative pathology,who underwent surgical resection in our hospital,were analyzed retrospectively.The gender,age,tumor maximum diameter,tumor location,tumor shape,seudocapsule,cystic degeneration,intratumoral hemorrhage and calcification were analyzed,and the stratified analysis was carried out according to gender,age,tumor maximum diameter and tumor location.Results:A total of 104 lesions(25 lesions in males and 79 in females)of 103 SPTP patients(25 [ 24.3% ] males and 78 [ 75.7% ] females)were included.The average ages of the males and females were(40.4±11.3)years old and(32.9±12.0)years old,respectively,and the difference was significant(P = 0.004).The mean tumor maximum diameters of the males and females were(36.8±25.3)mm and(50.0±31.2)mm,respectively,and the difference was significant(P = 0.046).The results of stratified analysis showed that the females were more likely to have pseudocapsule(62 cases),cystic degeneration(41 cases)and intratumoral hemorrhage(37 cases)compared with the males(13 cases,3 cases and 6 cases,respectively),and the differences were significant(all P < 0.05).The patients aged ≤ 40 years old were more likely to have tumor cystic degeneration(37 cases)and intratumoral hemorrhage(37 cases)compared with the patients aged > 40 years old(7 cases and 6 cases,respectively),and the differences were significant(both P < 0.01).The tumors with maximum diameter >3 cm were mainly distributed in the tail of the pancreas(33 cases),and were more likely to have pseudocapsule(59 cases),cystic degeneration(40 cases)and intratumoral calcification(26 cases)compared with the tumors with maximum diameter ≤ 3 cm(14cases,16 cases,4cases and 5 cases,respectively),and the differences were significant(all P < 0.01).The tumors in the tail of the pancreas were more likely to have pseudocapsule(40 cases)and cystic degeneration(25 cases),followed by the tumors in the head of the pancreas(26 cases and 17 cases,respectively),and the differences were significant(both P < 0.01).Conclusions : Female patients are more likely to form pseudocapsule,cystic degeneration and intratumoral hemorrhage.The younger age group(age ≤ 40 years)was more likely to form cystic degeneration and intratumoral hemorrhage.Large tumors(maximum diameter > 3cm)are more likely to locate in the tail of pancreas,and more likely to form pseudocapsule,cystic degeneration and calcification.Pseudocapsule and cystic degeneration are more likely to form in pancreatic tail tumors.Part 2 CT,MRI and radiomics differentiation of G stage of pancreatic neuroendocrine tumorObjective:To identify the G1,G2 and G3 phases of pancreatic neuroendocrine tumors(p NETs)by CT,MRI and their iconography.Methods:CT and MRI findings of 200 patients with PNETs confirmed by pathology were analyzed retrospectively.138 cases were enhanced by CT and 125 cases by MRI.To evaluate the tumor location,shape,maximum diameter,solid capsule,clear boundary,capsule,hemorrhage,calcification,tumor CT or MRI plain scan / normal pancreatic CT or MRI plain scan,tumor CT or MRI arterial phase / normal pancreatic CT or MRI arterial phase,tumor CT or MRI portal phase / normal pancreatic CT or MRI portal phase,tumor CT or MRI delayed phase / normal pancreatic CT or MRI delayed phase,yes There was no dilatation of pancreatic duct,peripancreatic invasion,liver metastasis and tumor thrombus of splenic vein.The clinical examination indexes of tumor were collected: AFP,carcinoembryonic antigen,CA724,CA199.The radiologic features(G1,G2,G3)were extracted from the CT and MRI images(4 stages in total)of the three phases(arterial phase,portal vein phase and delayed phase)after plain scanning and enhancement,and the features were screened by mrmr and lasso models.ROC(receiver operating characteristic curve)analysis was used to evaluate the diagnostic efficacy of radiation features.Application of decision curve analysis in clinical application.Results:There were statistically significant differences in tumor location,shape,maximum diameter,tumor CT + 2 / Normal Pancreas CT + 2,tumor CT + 3 / Normal Pancreas CT + 3,peripancreatic invasion and liver metastasis in CT univariate analysis.The results of CT single factor analysis showed that: there was statistical significance in the difference of the maximum tumor size and liver metastasis between G1 and G2;there was statistical significance in the difference of tumor location,shape and maximum diameter between G1 and G3;there was statistical significance in the difference of tumor shape and peripancreatic invasion between G2 and G3.There were significant differences in the maximum diameter of tumor and CT + 3 of tumor / normal pancreas.There were significant differences in tumor location,shape,tumor maximum diameter and pancreatic duct dilation in single factor MRI analysis.The results of MRI single factor analysis showed that there was statistical significance in the difference between the maximum diameter of G1 phase and G2 phase,and there was statistical significance in the difference between G1 phase and G3 phase.There was significant difference in the diameter of tumor and the presence or absence of pancreatic duct dilation in the multi factor analysis of MRI.Results: in the verification group,the AUC was 0.883 in plain scan,0.867 in arterial phase,0.900 in venous phase,and 0.867 in delayed phase.In the verification group,the AUC was 0.858 in plain scan,0.854 in arterial phase,0.808 in venous phase,and 0.850 in delayed phase.Conclusions:The CT and MRI manifestations of G1,G2 and G3 tumors are different.The maximum diameter of tumor,CT + 3 of tumor / normal pancreas and the presence or absence of pancreatic duct dilation are the most significant for the differentiation of G1,G2 and G3 stages of pancreatic neuroendocrine tumors.CT and MRI plain scan and enhanced radiomics features can distinguish G1 G2 and G3 PNETs,and CT venous phase and MRI arterial phase have the best differentiation efficiency.Part 3 Differential diagnosis of solid pseudopapilloma of the pancreas and hypovascular neuroendocrine tumor by CT and MRIObjective:To identify the CT and MRI features of solid pseudopapilloma of the pancreas and p NETs.Methods:The CT and MRI findings of 101 SPTP patients and 26 p NETs patients were analyzed retrospectively.To evaluate the tumor location,shape,maximum diameter,solid capsule,clear boundary,capsule,hemorrhage,calcification,tumor CT or MRI plain scan / normal pancreatic CT or MRI plain scan,tumor CT or MRI arterial phase / normal pancreatic CT or MRI arterial phase,tumor CT or MRI portal phase / normal pancreatic CT or MRI portal phase,tumor CT or MRI delayed phase / normal pancreatic CT or MRI delayed phase,yes There was no dilatation of pancreatic duct,peripancreatic invasion,liver metastasis and tumor thrombus of splenic vein.The clinical examination indexes of tumor were collected: AFP,carcinoembryonic antigen,CA724,CA199.Results:There were significant differences in sex,age,CT + 1 / normal pancreas,CT + 2 / normal pancreas,liver metastasis and tumor thrombus of spleen vein in CT univariate analysis.There were significant differences in sex,age,hemorrhage,T2 WI signal intensity,tumor MRI + 1 / normal pancreas MRI + 1,peripancreatic invasion,liver metastasis and carcinoembryonic antigen in single factor MR analysis.There was significant difference between MRI+ 1 and normal pancreas.The AUC of ROC curve of MRI prediction model is 0.851,the sensitivity is 0.877 and the specificity is 0.773.According to nomogram map of Mr prediction model,when the risk corresponding to the total score is less than 0.4,the patient is more likely to have SPTP,and when the risk is more than 0.4,the patient is more likely to have p NETs.Conclusions:The young,female and intratumoral hemorrhage are helpful for the diagnosis of SPTP,and the incidence of peripancreatic invasion,liver metastasis and tumor thrombus of splenic vein is more likely to be diagnosed as p NETs.Under the condition that the enhancement degree of each phase is lower than that of normal pancreatic parenchyma,p NETs has a high enhancement rate of arterial phase.The nomogram map of MRI prediction model can help to differentiate SPTP from p NETs.Part 4 3D CT and MRI imaging omics different between SPTP and p NETsObjective:To differentiate SPTP and p NETs by CT and MRI imaging omics.Methods:The SPTP and p NETs confirmed by pathology from 2012 to 2018 were analyzed retrospectively.The radiologic features were extracted from CT and MRI images of three phases(arterial phase,portal vein phase,delayed phase)after plain scanning and enhancement(four phases in total).The features were screened by mrmr and LASSO models.The diagnostic effectiveness of radiation characteristics was evaluated by ROC(receiver operating characteristic curve)analysis.Application of decision curve analysis in clinical application.Results:The results of SPTP and pNETs were as follows: in the verification group,the AUC was 0.842 in plain scan,0.960 in arterial phase,0.906 in venous phase,and 0.874 in delayed phase.In the verification group,the AUC was 0.843 in plain scan,0.906 in arterial phase,0.854 in venous phase,and 0.835 in delayed phase.The results of differentiation between SPTP and p NETs in G1 phase: in the verification group,the AUC was 0.910 in plain scan,0.982 in arterial phase,0.990 in venous phase,and 0.935 in delayed phase.In the verification group,the AUC was 0.900 in plain scan,0.933 in arterial phase,0.917 in venous phase,and 0.900 in delayed phase.The results of p NETs differentiation between SPTP and g2g3: in the verification group,the AUC was 0.946 in plain scan,0.979 in arterial phase,0.971 in venous phase,and 0.929 in delayed phase.In the verification group,the AUC was 0.897 in plain scan,0.957 in arterial phase,0.917 in venous phase,and 0.937 in delayed phase.The results of SPTP and p NETs: the AUC of the validation group was 0.950 in CT plain scan,0.983 in CT arterial phase,0.950 in CT venous phase,and 0.900 in CT delayed phase.In the verification group,the AUC was 0.853 in plain scan,0.907 in arterial phase,0.787 in venous phase,and 0.773 in delayed phase.The AUC value of MRI in the training group and the verification group was the highest.The most characteristic parameters are Median 、 Skewness 、 Non Uniformity Normalized、Variance、Emphasis、10 Percentile、90 Percentile.The AUC value of CT and MRI in arterial phase was the highest in most(7 / 8)validation groups.The results of decision curve indicated that most(5 / 8)CT and MRI arterial phase had the greatest clinical value in the range of 0-1 threshold.Conclusions:SPTP and p NETs can be distinguished by CT and MRI plain and enhanced radiomics features,and the arterial phase is the best.Based on the enhanced results,plain scan can effectively identify SPTP and p NETs.