Value of dual-layer spectral detector CT in differentiating the diagnosis of lung cancer and inflammatory nodules
孙奕波,余烨,陈杏彪,傅奕铖,程杰军,李晓倩,吴华伟,张莹
DOI: https://doi.org/10.3760/cma.j.cn112149-20210125-00061
2021-01-01
Abstract:Objective:To explore the value of dual-layer spectral detector CT in differentiating the diagnosis of lung cancer and inflammatory nodules.Methods:A total of 92 patients undergoing enhanced chest scan from March 2019 to September 2020 at Renji Hospital, School of Medicine, Shanghai Jiaotong University, were retrospectively enrolled in the study. The conventional CT parameters, spectral CT parameters were measured and the nodules′ morphological characteristics were analyzed. Later the factors with statistical significance were identified as independent variables in a logistic regression model to establish models for predicting malignant nodules. ROC curve was used to assess the diagnostic performance for the conventional CT model, spectral CT parameters and combined model, respectively. Differences in the area under the ROC curve (AUC) were analyzed by the DeLong test.Results:Lobulated sign (42 and 8, respectively, χ2=10.779, P=0.001), short burr sign (41 and 7, respectively, χ2=11.911, P=0.001), pleural indentation sign (45 and 9 respectively, χ2=11.705, P=0.001), vascular convergence sign (35 and 8, respectively, χ2=5.337, P=0.021) and the venous phase iodine concentrations (IC) value [(2.1±0.5) mg/ml, (2.3±0.5) mg/ml, t=-2.464, P=0.016], normalized iodine concentrations (NIC) value (0.40±0.06, 0.45±0.08, t=-6.943, P<0.001), and Z-effective (Z eff) values (8.38±0.21, 8.49±0.19, t=-2.122, P=0.037) were significantly different between the lung cancer group and the inflammatory group, while other CT signs and CT indicators were not significantly different between the lung cancer group and the inflammatory group ( P>0.05). The conventional CT model was established with lobulated sign, short burr sign, pleural indentation sign, vascular convergence sign, and the AUC for differential diagnosis of lung cancer and inflammatory nodules was 0.827. The spectral CT parameter model was established with venous phase IC, venous phase NIC, and venous phase Z eff value, and the AUC for differential diagnosis of lung cancer and inflammatory nodules was 0.899. The conventional CT model combined spectral CT parameter model was established with the significant factors in the univariate analysis, and the AUC for differential diagnosis of lung cancer and inflammatory nodules was 0.925. The AUC of the combined model showed no significant difference from that of the spectral CT parameter model ( Z=1.794, P=0.073). However, AUC of the combined model was significantly higher than that of evaluation based on conventional CT alone ( Z=2.156, P=0.031). Conclusion:Spectral CT parameters combined with conventional CT signs can improve the differential diagnosis efficiency between lung cancer and inflammatory nodules.