目的·探讨研究双层探测器光谱电子计算机断层扫描（computed tomography，CT）虚拟单能量图像（virtual monoenergetic image，VMI）在胰腺导管腺癌（pancreatic ductal adenocarcinoma，PDAC）术前评估中的最佳成像能级及应用价值。方法·回顾性分析2019年1月—2019年6月于上海交通大学医学院附属瑞金医院双层探测器光谱CT上行胰腺动态增强扫描并经手术病理证实为PDAC的60例患者影像资料。扫描完成后均行胰腺实质期、门静脉期常规120 千伏峰值（kilovolt peak，kVp）混合能量图像（polyenergetic image，PI）以及40、50、60和70 keV VMI重建。测量PI以及40、50、60和70 keV VMI中病灶、正常胰腺实质、胰周动静脉及腹壁脂肪的CT值，以腹壁脂肪CT值标准差（standard deviation，SD）为图像噪声值，计算病灶-胰腺实质（caner-pancreas，CA-P）、胰周动脉-病灶（artery-caner，A-CA）、胰周静脉-病灶（vein-caner，V-CA）的对比噪声比（contrast-to-noise ratio，CNR），采用配对t检验与单因素方差分析比较不同期相及不同重建图像中CT值、CNR和图像噪声的差异，并采用Bonferroni 法进行组内两两比较。基于病灶-胰腺实质对比度、病灶边界清晰度、图像总体质量对各组图像进行主观评分，采用Wilcoxon检验与Friedman检验进行不同期相及不同重建图像组间对比。结果·胰腺实质期各组图像中CNR_CA-P，及病灶-胰腺实质对比度、病灶边界清晰度主观评分优于门静脉期（均P<0.05）。不同能级VMI中，CNR_CA-P、CNR_A-CA、CNR_V-CA随着能级的降低而明显升高，而图像噪声随着能级降低轻度升高（均P=0.000）。VMI_{40 keV}图像中CNR_CA-P、CNR_A-CA、CNR_V-CA值最高，而图像噪声仍低于PI（均P=0.000）。VMI_{40 keV}各图像质量主观评分不低于PI。结论·双层探测器光谱CT VMI_{40 keV}能够明显提高PDAC病灶-胰腺实质、胰周血管-病灶对比噪声比，且拥有较高图像质量，能够优化PDAC成像与术前评估。

Objective ·To explore the optimal energy level of virtual monoenergetic images (VMIs) derived from a dual-layer spectral detector computed tomography (CT) and its application value in the preoperative evaluation of pancreatic ductal adenocarcinoma (PDAC). Methods ·From January 2019 to June 2019, the images of sixty patients who underwent pancreas dynamic enhanced scanning on a dual-layer spectral detector CT and were pathologically confirmed of PDAC in Ruijin Hospital, Shanghai Jiao Tong University School of Medicine were retrospectively analyzed. The conventional 120 kVp polyenergetic image (PI), and 40, 50, 60, and 70 keV VMIs were generated. The CT values of lesions, normal pancreatic parenchyma, peripheral pancreatic arteries and veins, and abdominal subcutaneous fat were measured, and the contrast-to-noise ratios (CNRs) of cancer-pancreatic parenchyma (CA-P), artery-cancer (A-CA), and vein-cancer (V-CA) were calculated with the standard deviation (SD) of abdominal subcutaneous fat CT value as image noise. The differences were analyzed by using paired t test and one‐way ANOVA test. Pairwise comparisons within groups were performed by using the Bonferroni method. Subjective scores were obtained based on the lesion-pancreatic parenchyma contrast, lesion boundary definition, and the overall image quality, and compared by using Wilcoxon test and Friedman test. Results ·Among PI and all VMIs, the CNR_CA-P and the subjective scores of lesion-pancreas parenchyma contrast and the lesion boundary definition in pancreatic parenchymal phase were higher than those in portal vein phase (all P=0.000). The CNR_CA-P , CNR_A-CA, and CNR_V-CA of VMIs increased significantly as the energy level decreased from 70 keV to 40 keV, while the image noise only subtly increased (all P=0.000). The VMI_{40 keV} had the highest CNR_CA-P , CNR_A-CA, and CNR_V-CA with lower image noise than PI (all P=0.000). The subjective scores of VMI_{40 keV} were not lower than PI. Conclusion ·The VMI_{40 keV} derived from the dual-layer spectral detector CT performs a better image quality, and can significantly improve the CNR_CA-P , CNR_A-CA, and CNR_V-CA, which can optimize the preoperative imaging and the evaluation of pancreatic ductal adenocarcinoma.