肺结节加强观察功能对双源CT低剂量扫描图像肺结节检出效能的影响

Influence of nodule enhanced viewing of dual-source CT on efficacy of detecting pulmonary nodule in low-dose CT

目的评价双源CT在低剂量扫描前提下肺结节加强观察（NEV）功能的诊断效能。方法在肺转移瘤或筛查肺癌随诊的患者中，采用数字表法随机抽取127例进行胸部CT低剂量扫描。CT图像由2名具有10年以上胸部影像诊断经验的医师，共同阅片诊断的肺结节为诊断标准。由住院医师分别采用和不采用辅助NEV分析软件阅读同一组CT图像，记录单独NEV软件、住院医师采用辅助NEV分析软件和不采用辅助情况下发现的肺结节数量和阅读时间。对结果采用秩和检验进行比较。同时以具有10年以上胸部影像诊断经验的医师的结果为标准，分别计算3种情况下诊断的敏感度和准确度。结果高年资医师、住院医师、NEV及NEV辅助医师发现最大径≤2．0cm的非钙化肺结节数量分别为570、404、768、593个，2／3以上的肺结节最大径〈0．5cm。住院医师发现的最大径≤2．0cm与〈0．5cm肺结节数量均明显少于NEV发现的，两者间差异有统计学意义（z值分别为-6．887和-7．235，P均〈0．01）。住院医师使用NEV后发现的最大径≤2．0cm与〈0．5cm肺结节数量比使用前均明显增多，两者问差异有统计学意义（z值分别为-6．606和-6．657，P均〈0．01）。住院医师、NEV、住院医师应用NEV发现最大径≤2．0cm肺结节的敏感度与准确度分别为61．4％、86．3％、95．3％与56．1％、58．1％、87．6％，发现最大径〈0．5em肺结节的敏感度与准确度分别为51．4％、88．1％、94．8％与47．0％、56．9％、87．5％。住院医师、NEV、住院医师借助NEV的阅读时间分别为120～444s／例、85～262s／例、131～1512s／例。住院医师使用与不使用NEV的阅读时间差异有统计学意义（Z=-9．781，P〈0．01）。住院医师平均花费（23±10）S判断NEV发现的每个肺结节可疑区域。结论NEV有助于住院医师发现肺结节，尤其是对〈0．5cm的肺结节。但是在判断肺结节的性质方面，NEV不能代替医师。

Objective To evaluate efficacy in detecting lung nodules at low-dose CT （LDCT）by nodule enhanced viewing（NEV）. Methods One hundred and twenty seven patients who were referred to undergo low-dose CT （LDCT） for the evaluation of pulmonary metastasis or screening lung cancer were selected randomly. Two radiologists with at least 10 years experience read the images with normal clinical reading speed to find actionable nodules ≤2. 0 cm in maximum diameter, and their consensus result was referred as ＂Standard＂. NEV was adopted to detect the pulmonary nodules. Two residents with experience of less than three years read first detected suspicious nodules and recorded reading time, first consensus and mean time were recorded. Then, they made second decisions on the images with the help of NEV and the resuhs and the reading time were recorded and analyzed by using wilcoxon test. The sensitivity and accuracy of NEV, residents and residents with NEV were analyzed. Results ＂Standard＂, resident, NEV and resident with NEV detected 570,404,768 and 593 lung nodules ≤2.0 cm in maximum diameter,respectively. More than 60% nodules were less than 0. 5 cm in maximum diameter. The performance of NEV in detecting nodules ≤2. 0 cm as well as nodules 〈 0. 5 cm in maximum diameter was significantly higher than that of the resident （Z = -6. 887,P 〈 0. 01 and Z = -7. 235,P 〈 0. 01 ） ,and the performance of resident with NEV indetecting nodules ≤2. 0 cm as well as nodules 〈 0. 5 cm in maximum diameter was significantly higher than that of resident without NEV（Z = -6. 606,P 〈 0. 01 and Z = -6. 657 ,P 〈 0. 01 ）. The resident, NEV and the resident with NEV detected nodules 〈 20 mm in maximum diameter with sensitivities of 61.4% ,86. 3% and 95.3%, and with accuracy of 56. 1%, 58. 1% and 87.6%, respectively. The resident achieved sensitivities of 51.4%, 88.1% and 94. 8%, and accuracy of 47. 0%, 56. 9% and 87.5% for nodules 〈 5mm in maximum diameter, respectively. The resident, NEV and resident with NEV spent 120 -444 s ,85 - 262 s and 131 - 1512 s per case to read the CT scans, respectively. The reading time of resident with NEV in was significantly higher than that of resident without NEV（ Z = -9. 781 ,P 〈 0. 01 ）. The resident spent 23 s per NEV mark. Conclusion NEV considerable improves the resident＇s performance in lung nodule detection, especially in maximum diameter 〈 0. 5 cm nodule detection.