CT容积灌注诊断孤立性肺结节良恶性的临床应用评价

The Evaluation of CT Volume Based Fist Pass Perfusion in Diagnosis of SPNs

目的评价常规胸部CT结合CT容积灌注(CTVP)诊断孤立性肺结节(SPN)良、恶性的临床应用价值。资料与方法搜集行首过法CTVP的SPN65例,其中恶性43例,良性22例。先通过受试者工作特征(ROC)曲线明确CTVP参数诊断阈值,并通过四格表行2项参数的联合试验。由两名低年资和两名高年资医师回顾性盲法阅常规胸部CT图像,做出诊断(方法A);再阅常规胸部CT及CTVP资料,作出诊断(方法B)。采用ROC曲线比较每名医师采用诊断方法A与B的准确性有无统计学差异。通过组内相关系数(ICC)评价诊断方法A及B的一致性。结果当BF=50ml.100g-1.min-1、BV=3ml/100g、PS=10ml.100g-1.min-1时,CTVP诊断SPN良、恶性的敏感性分别为93%、86%及86%,特异性分别为59%、68%及77%。BF+PS及BV+PS的序列实验联合特异性均为82%。联合运用常规胸部CT和CTVP,四名医师诊断SPN良、恶性的敏感性分别为91%、95%、95%及95%;特异性分别为86%、91%、95%及82%。四名医师方法B的ROC曲线下面积均大于方法A(P均<0.05)。四名医师方法A的ICC=0.612;方法B的ICC=0.854。结论与仅采用常规胸部CT相比,结合首过法CTVP可提高放射科医师诊断SPN良、恶性的准确性和一致性。

Objective To investigate role of CT volume based perfusion（CTVP） combine with routine thorax CT in the diagnosis of SPNs.Materials and Methods 65 patients with 43 malignant and 22 benign nodules received first pass CTVP examination.BF,BV,MTT,and PS were determined by Perfusion 3 software.Two junior and two senior radiologists made a diagnosis by those routine thorax CT data（method A） and CTVP plus routine CT data（method B） blindly.ROC curves were taken to judge the diagnostic efficiency.ROC curves also were used to analyze diagnostic accuracy between method A and B.Intraclass correlation coefficients（ICC） were used to investigate the agreements for diagnosis before and after CTVP used.Results When BF 50 ml·100g-1 ·min-1,BV 3 ml/100g,and PS 10 ml·100g-1 ·min-1 was used,the sensitivity,specificity of diagnosing malignant and benign SPNs were 93%,86%,86% and 59%,86%,77%,respectively.The combined specificity of serial tests were all 82%.Significant statistical differences were found between method A and B in the area under the cuvre of every radiologist,respectively（P0.05 for all）,the sensitivity,specificity of diagnosing malignant and benign SPNs were 91%、95%、95%,95% and 86%,91%,95%,82% for four radiologist.After CTVP data were used,ICC was increased from 0.612 to 0.854.Conclusion Combined routine CT with first pass CTVP could improve the diagnostic accuracy and agreement in malignant and benign SPNs.