目的 了解肝多房棘球蚴病（HAE）病灶不同的超声表现，为提高HAE在常规超声下的检出率和诊断准确率提供可靠的依据。方法 将2018年1月至2021年4月于新疆医科大学第一附属医院行常规超声检查并经手术病理确诊为HAE的患者作为研究对象，收集研究对象性别、年龄、民族、超声影像学分型结果等相关资料。选取研究对象超声原始图像中HAE病灶最清晰完整的图像进行分析，记录病灶的超声表现（病灶位置、大小、边界和形态、实性部分回声、钙化情况、液化坏死情况、血流信号、胆管和血管的侵犯等）并赋值，多发病灶分开描述；结合手术后病理结果判定病灶超声诊断符合情况。对HAE病灶的各种超声表现进行单因素分析，将单因素分析有统计学意义的因素设为自变量，病灶超声诊断符合情况设为因变量，进行logistic多因素分析，建立回归模型并用受试者工作特征（ROC）曲线验证；采用logistic多因素分析筛选出的结果构建列线图，采用ROC曲线、校准曲线、决策分析曲线评估其性能。结果 纳入研究的HAE患者数为141例，年龄分布在9~65岁，平均年龄（37.4 ± 13.6）岁；男性71例，女性70例；藏族患者87例，占61.7%。141例HAE患者中28例为单发病灶，113例为多发病灶，最终纳入标准的HAE病灶数为170个。单因素分析结果显示，不同病灶位置对超声诊断符合情况的影响无统计学意义（χ² = 1.952，P > 0.05），病灶大小、边界和形态、内部回声、钙化、液化坏死、血流信号、胆管和血管的侵犯等因素对超声诊断符合情况的影响均有统计学意义（χ² = 39.026、18.601、15.743、47.205、34.151、6.597、21.766，均P < 0.05）。病灶大小（OR = 0.180，95% CI：0.020~1.645）、钙化情况（OR = 0.037，95% CI：0.002~0.590）、液化坏死情况（OR = 0.282，95% CI：0.042~1.867）、血流信号（OR = 20.746，95% CI：3.720~115.686）等4个因素与HAE病灶超声诊断符合情况存在相关性。logistic回归模型预测准确性分析结果显示，该回归模型灵敏度为96.7%（118/122），特异度为83.3%（40/48），阳性预测值为93.7%（118/126），阴性预测值为90.9%（40/44）；ROC曲线下面积为0.918（95% CI：0.859~0.977），灵敏度为95.1%，特异度85.4%。基于多因素logistic回归分析结果构建的列线图ROC曲线下面积为0.833，灵敏度为86.9%，特异度为70.8%；校准曲线与参考曲线接近，决策分析曲线均明显偏离两条参考线。结论 HAE患者的病灶大小、钙化、液化坏死和血流信号是影响超声诊断的重要因素。

Objective To understand the different ultrasonic manifestations of hepatic alveolar echinococcosis (HAE), to provide a reliable basis for improving the detection and diagnostic accuracy of HAE under conventional ultrasonography. Methods From January 2018 to April 2021, the patients who underwent routine ultrasound examination in the First Affiliated Hospital of Xinjiang Medical University and were diagnosed as HAE through surgery and pathology were recruited in the study. The relevant data, including gender, age, nationality, and ultrasonic image classification were collected. The clearest and complete images of HAE lesions in the original ultrasound images of the study subjects were selected for analysis, and the ultrasound manifestations of the lesions (location, size, boundary and shape, solid portion echo, calcification, liquefaction necrosis, blood flow signals, bile ducts and vascular invasion, etc.) were recorded, and values were assigned. Multiple lesions were described separately. The coincidence between post-operation pathological findings and the ultrasound diagnosis of the lesion was determined. Univariate analysis was conducted on the various ultrasound manifestations of HAE lesions, with statistically significant factors set as independent variables, while the coincidence with ultrasound diagnosis of lesions as dependent variables. A multivariate logistic regression analysis was conducted to establish a regression model, which was validated with the receiver operating characteristic (ROC) curve. The nomogram was constructed based on the results screened using multivariate logistic regression analysis, and its performance was evaluated using the ROC curve, calibration curve and decision analysis curve. Results The total number of HAE patients included in this study was 141. The age ranged from 9 to 65 years old, with an average age of (37.4 ± 13.6) years. There were 71 males and 70 females in this study, including 87 Tibetan patients accounting for 61.7%. Among the 141 HAE patients, 28 had single lesions, and 113 had multiple lesions. The final number of HAE lesions with inclusive criteria was 170. The results of univariate analysis showed that the difference in lesion location was not statistically significant (χ² = 1.952, P > 0.05). The differences in lesion size, boundary and shape, internal echo, calcification, liquefaction necrosis, blood flow signal, and bile ducts and vascular invasion were statistically significant (χ² = 39.026, 18.601, 15.743, 47.205, 34.151, 6.597, 21.766, all P < 0.05). The 4 factors including lesion size (OR = 0.180, 95% CI: 0.020-1.645), calcification (OR = 0.037, 95% CI: 0.002-0.590), liquefaction necrosis (OR = 0.282, 95% CI: 0.042-1.867), and blood flow signal (OR = 20.746, 95% CI: 3.720-115.686) were correlated with the ultrasound diagnosis compliance of HAE lesions. The accuracy analysis results of the logistic regression model predicted that the sensitivity of the regression model was 96.7% (118/122), the specificity was 83.3% (40/48), the positive predictive value was 93.7% (118/126) and the negative predictive value was 90.9% (40/44). The area under the ROC curve was 0.918 (95% CI: 0.859-0.977), the sensitivity was 95.1% and the specificity was 85.4%. The area under the ROC curve of the nomogram constructed based on the results of multivariate logistic regression analysis was 0.833. The sensitivity was 86.9% and the specificity was 70.8%. The calibration curve was close to the reference curve, and the decision analysis curve deviated significantly from both reference curves. Conclusion The size, calcification, liquefaction necrosis, and blood flow signal of the HAE lesions were important factors affecting ultrasound diagnosis.