Multi-parameter spectral CT for differentiating grade G2-3 pancreatic neuroendocrine tumor and pancreatic neuroendocrine carcinoma
Objective To explore the value of multi-parameter spectral CT for differentiating grade G2-3 pancreatic neuroendocrine tumor(pNET)and pancreatic neuroendocrine carcinoma(pNEC).Methods Preoperative double-layer detector spectral CT(DLCT)data of 35 patients with pNET(pNET group,including 25 cases of G2 grade and 10 cases of G3 grade)and 17 patients with pNEC(pNEC group)were retrospectively analyzed.Conventional CT and spectral CT parameters were compared between groups,and those being significant different between groups according to univariate analysis were respectively incorporated into multivariate logistic regression to select the independent predictors for identifying grade G2-3 pNET and pNEC.Conventional CT model and spectral CT model were constructed,and the combined model was constructed based on the two.The efficacy of each model for distinguishing grade G2-3 pNET and pNEC was evaluated.Results CT values of lesions during venous phase(OR=0.939,P=0.025)and vascular invasion(OR=5.049,P=0.027)shown on conventional CT were both independent predictors,and conventional CT model was constructed,its area under the curve(AUC)for distinguishing grade G2-3 pNET and pNEC was 0.808.Normalized iodine concentration during venous phase(OR=0.603)and normalized effective atomic number during venous phase(OR=0.847)on spectral CT were both independent predictors(both P＜0.05),and spectral CT model was constructed.The AUC of spectral CT model was 0.894,higher than that of conventional CT model(Z=2.127,P=0.033).The AUC of combined model was 0.924,higher than that of conventional CT model(Z=2.302,P=0.021)but not significantly different with that of spectral CT model(Z=0.827,P=0.408).Conclusion Multi-parameter spectral CT could effectively differentiate grade G2-G3 grade pNET and pNEC.
NETL
NSTL
万方数据
