探讨径向支气管镜超声（EBUS-GS）的声像图特征对肺部良、恶性病变的评估价值。

回顾性分析解放军第九〇三医院从2018年1月至2019年6月行EBUS-GS检查的154例患者的临床资料和检查结果，以病理诊断为“金标准”，采用四格表分析EBUS-GS的声像图特征对肺部良、恶性病变诊断的敏感度、特异度、阳性预测值、阴性预测值及诊断准确性。采用χ²检验和二分类Logistic回归模型分析EBUS-GS的声像图特征对肺部恶性疾病预测的应用价值。

154例患者中病理诊断良性病变为96例，恶性病变54例，4例未明确病理。以病理诊断为“金标准”，得出超声声像图评价准确率为98.05%（151/154），与病理科诊断的一致率为90.26%（139/154），敏感度为88.42%（84/95），特异度为98.15%（53/54），阳性预测值为98.82%（84/85），阴性预测值为82.81%（53/64）。EBUS-GS检查的恶性病变的声像图多为不规则，边界不清晰、不连续，回声基本均匀，内部大多为团块状的低回声，少部分含气的会出现点状高回声。而良性病变的声像图多为规则（圆形或椭圆形为主），边界清晰、连续，回声大多不均匀，内部以高回声为主，大多会出现点状或斑状的高回声。Logistic回归分析显示，EBUS-GS的声像特征对预测肺部恶性疾病具有一定的价值，差异均具有统计学意义（P均＜0.05）。

EBUS-GS的超声声像图特征对肺部良、恶性疾病的诊断具有一定的参考价值，有助于对其进行初步评估，可以为未来的临床早期诊断提供新的依据。

To assess the value of endobronchial ultrasound with guide-sheath (EBUS-GS) in the diagnosis of benign and malignant lung diseases.

A retrospective analysis was performed on the clinical data and test results of 154 patients with lung lesions who underwent routine EBUS-GS at the PLA 903 Hospital from January 2018 to June 2019. Using the pathological diagnosis as the gold standard, a four-fold table was used to calculate the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of EBUS-GS in the diagnosis of benign and malignant lung lesions. The χ² test and two-class Logistic regression were used to assess the value of EBUS-GS characteristics in the prediction of malignancy of lung diseases.

Among the 154 patients who underwent EBUS-GS examination, 96 had benign lesions, 54 had malignant lesions, and 4 had lesions whose nature was unclear. According to the gold standard of pathological diagnosis, the accuracy of EBUS-GS was 98.05% (151/154), the consistency rate with the pathology diagnosis was 90.26% (139/154), the sensitivity was 88.42% (84/95), the specificity was 98.15% (53/54), the positive predictive value was 98.82% (84/85), and the negative predictive value was 82.81% (53/64). On EBUS-GS, malignant lesions mostly had an irregular shape, unclear and discontinuous boundary, and basically uniform echo; the interior of the lesions mostly exhibited clumpy-like low echo and some dot-like high echo in the tissue containing gas. In contrast, benign lesions mostly had a regular (round or oval-based) shape, clear and continuous boundary, and mostly uneven echo; the interior of the lesions manily exhibited high echo, with most showing dot or spot-shaped high echo. Logistic regression analysis demonstrated that the acoustic characteristics of EBUS-GS were of some value to predict malignant diseases in the lungs (P＜0.05).

EBUS-GS has appreciated value in the diagnosis of benign and malignant lung diseases, which is helpful for the preliminary evaluation of benign and malignant lung diseases and can provide a new basis for the early clinical diagnosis in the future.