CT纹理分析技术鉴别良恶性孤立性肺结节

目的 探讨CT纹理分析技术鉴别诊断良恶性孤立性肺结节（SPN）的应用价值。方法 对97例SPN患者（恶性组54例，良性组43例）行胸部CT平扫，测量结节CT值及最大径。采用MaZda软件对肺结节进行CT纹理分析，获得纹理特征参数（熵、熵和、熵差、对比度、相关及均和）。比较2组CT值、最大径及各纹理特征参数。对差异有统计学意义的纹理参数及其不同组合建立ROC曲线，计算AUC，确定阈值，评价各参数及其不同组合鉴别诊断良恶性SPN的特异度及敏感度；比较各参数组合之间AUC的差异。结果 良恶性SPN最大径、CT值的差异均无统计学意义（t=-0.98、1.16，P=0.079、0.087）。良性组SPN纹理特征参数熵、熵和、熵差值均低于恶性组（P均<0.05），2组对比度、相关、均和的差异均无统计学意义（P均>0.05）。纹理特征参数熵、熵和、熵差鉴别诊断良恶性SPN的效能均较高（AUC均>0.700）；熵、熵和、熵差的阈值分别为1.564、1.212、0.987时，敏感度分别为72.70%、88.40%、63.60%。上述纹理特征参数的不同组合鉴别诊断良恶性SPN的效能均较高（AUC均>0.800），各组合间AUC的差异均无统计学意义（P均>0.05）。结论 基于CT平扫的纹理分析技术有助于鉴别良恶性SPN。

CT texture analysis in differential diagnosis of benign and malignant solitary pulmonary nodule

Objective To investigate the value of CT texture analysis in differential diagnosis of benign and malignant solitary pulmonary nodules (SPN). Methods CT scan was performed in 97 patients with SPN (54 in the malignant group and 43 in the benign group). The CT value and maximum diameter of the nodules were measured. The CT texture analysis of pulmonary nodules was performed using MaZda software to obtain texture parameters (Entropy, SumEntrp, DifEntrp, contrast, correlation and SumAver). The CT values, maximum diameters, and texture parameters of the two groups were compared. ROC curves were established for statistically significant texture parameters and their different combinations. AUC was calculated, thresholds were determined, and the specificity and sensitivity of different parameters and their different combinations of texture parameters in differential diagnosis of benign and malignant SPN were evaluated. The difference in AUC between the different parameter combination was compared. Results There was no significant difference in maximal diameter and CT value between the two groups (t=-0.98, 1.16, P=0.079, 0.087). The texture parameters of entropy, SumEntrp, DifEntrp of the benign group were lower than those of the malignant group (all P<0.05). There was no significant difference in the contrast, correlation and SumAver between the two groups (all P>0.05). The entropy, SumEntrp and DifEntrp were more effective in differential diagnosis of benign and malignant SPN (all AUC>0.700). When the threshold of entropy, SumEntrp and DifEntrp was 1.564, 1.212 and 0.987, the sensitivity was 72.70%, 88.40% and 63.60%, respectively. The different combinations of the above texture features were more effective in differential diagnosis of benign and malignant SPN (all AUC>0.800). There was no significant difference in AUC between the combinations (all P>0.05). Conclusion The texture analysis technique based on non-contrast CT scan can help to distinguish the benign and malignant SPN.