【摘要】 目的 探讨双层探测器光谱CT（dl-SDCT）多参数成像技术预测宫颈鳞癌淋巴血管间隙浸润（LVSI）的临床应用价值。方法 回顾分析82例经手术病理证实的宫颈鳞癌患者，LVSI+组42例和LVSI-组40例，治疗前均行全腹+盆腔光谱CT增强扫描，分析光谱CT动脉期（AP）、静脉期（VP）常规能量图像（CI）、虚拟单能量图像（VMIs）及碘密度图，测量宫颈癌病灶及不同背景组织的双期CT值，碘值（IC）,肿瘤最大径（MD）。计算能谱曲线斜率（K）,标准化碘浓度（NIC）。LVSI+组和LVSI-组间存在统计学差异的参数构建单因素模型预测宫颈鳞癌LVSI,采用受试者工作特征（ROC）曲线评估不同模型对LVSI状态的预测效能，并计算相应阈值，采用DeLong检验比较模型间曲线下面积（AUC）值差异。结果 肿瘤MD在LVSI+与LVSI-组之间差异具有统计学意义（P<0.05）;LVSI+组AP-CT_CI、AP-CT_{40 keV}、VP-CT_{40 keV}、AP-CT_{80 keV}、AP-CT_{100 keV}均高于LVSI-组（P<0.05）;LVSI+组AP-K、VP-K均高于LVSI-组（P<0.05）;LVSI+组AP-IC、VP-IC、AP-NIC_肌肉、AP-NIC_血管、VP-NIC_肌肉、VP-NIC_血管均高于LVSI-组（P<0.05）。肿瘤MD、AP-NIC_肌肉、VP-NIC_肌肉对于宫颈癌LVSI状态预测效能较好，各单因素ROC曲线诊断宫颈鳞癌患者LVSI状态的AUC分别为0.707,0.832,0.765。AP-NIC_肌肉单因素模型预测宫颈鳞癌LVSI状态效能最佳，敏感度及特异度分别为85.7%和85.0%,诊断阈值为2.90 mg/ml。肿瘤MD分别联合AP-NIC_肌肉、VP-NIC_肌肉构建M4、M5预测LVSI状态的AUC值分别为0.851,0.832,Delong检验结果显示M4、M5优于M1,差异均具有统计学意义（P均<0.05）。结论 光谱CT多参数可用于术前预测宫颈鳞癌LVSI状态，其中AP NIC_肌肉预测效能最佳，为临床治疗提供辅助诊断方法。更多还原

【Abstract】 Objective To investigate the clinical value of dual-layer spectral detector Computed Tomography（dl-SDCT） multi-parameter imaging in predicting lymph-vascular space invasion（LVSI） in cervical squamous cell carcinoma（CSCC）. Methods 82 patients with CSCC underwent abdominal and pelvic dl-SDCT enhanced scan before treatment and proved by pathology were divided into LVSI positive group（n=42） and LVSI negative group（n=40）. One fellow analyzed conventional images（CIs）,40-200 keV virtual monoenergetic images（VMIs） and iodine density maps of the arterial phase（AP） and venous phase（VP）, recorded the CT attenuation and iodine concentration（IC）for CSCC and background tissues, and tumor maximum diameter（MD） were assessed on spectral images. Monoenergetic curve slope（K） and normalized iodine concentration（NIC） were calculated. To establish models to assess LVSI status by the parameters with significantly differences, receiver operating characteristic（ROC） curves analysis was used to assess the predictive efficacy of variable models for LVSI status in CSCC, and the DeLong test was used for AUC comparation between models. Results There were significant difference in cervical lesion MD between LVSI positive group and LVSI negative group（P<0.05）. The AP-CT_CI、AP-CT_{40 keV}、VP-CT_{40 keV}、AP-CT_{80 keV}、AP-CT_{100 keV} in LVSI positive group were higher than LVSI negative group（P<0.05）. The AP-K、VP-K of LVSI positive group were higher than LVSI negative group（P<0.05）. The AP-IC、VP-IC、AP-NIC_muscle、AP-NIC_vascular、VP-NIC_muscle、VP-NIC_vascular of LVSI positive group were higher than LVSI negative group（P<0.05）.Three parameters had a better diagnostic efficacy in diagnosis of LVSI in CSCC,with area under the curve（AUC） of 0.707（MD）, 0.832(AP-NIC_muscle),0.765(VP-NIC_muscle), respectively. AP-NIC_muscle had the highest AUC, with sensitivity and specificity were 85.7% and 85.0%, with diagnostic threshold value of 2.90 mg/ml. The area under the ROC of combined model of tumor MD with AP-NIC muscle（M4） and VP-NIC muscle（M5） were 0.851,0.832,the DeLong test showed that M4 and M5 are better than M1（P<0.05）. Conclusion The dl-SDCT multi-parameter had high diagnostic efficacy for preoperative predicting LVSI status in CSCC, which AP-NIC muscle model was hopeful applied to clinical treatment as an auxiliary diagnostic methods.更多还原