| 摘要: |
| 目的：评价甲状腺超声图像处理软件（AmCAD-UT Detection）对甲状腺良恶性结节的诊断效能
方法：166例甲状腺结节由超声医师根据其二维超声及二维联合超声造影特征分别进行TI-RADS分级，CAD系统自动进行分级。建立ROC曲线，比较三种方法的曲线下面积，找出区分甲状腺良恶性结节的最佳截点，比较CAD系统与二维超声及二维联合超声造影的灵敏度、特异度、准确率，分析CAD系统描述的超声特征在良恶性结节中的差异，评估其诊断效能。
结果：CAD系统、二维超声、二维联合超声造影的AUC分别为0.859、0.833、0.930，诊断能力比较：二维联合超声造影＞CAD＞二维超声。CAD系统的灵敏度及准确率高于二维超声，特异度低于二维超声，二维联合超声造影诊断的特异度及准确率均高于CAD，差异有统计学意义（P＜0.05）。低回声、强回声点、不规则边缘等特征对诊断恶性结节意义较大。
结论：甲状腺超声图像处理软件（AmCAD-UT Detection）能够敏感地发现甲状腺恶性结节，有较高的诊断准确性，可全面而准确地量化结节特征，临床应用价值较高。 |
| 关键词: 甲状腺超声图像处理软件 甲状腺结节 二维超声 超声造影 TI-RADS |
| DOI： |
| 投稿时间：2019-03-04修订日期：2019-03-13 |
| 基金项目: |
|
| Application value of AmCAD-UT Detection for diagnosing benign and malignant thyroid nodules |
| Gao gui-fang,Chen wu |
| () |
| Abstract: |
| Objective To evaluate the diagnostic efficacy of AmCAD-UT Detection for benign and malignant thyroid nodules.Method The 166 thyroid nodules were graded TI-RADS by physicians according to their conventional two-dimensional ultrasound and two-dimensional ultrasound combined contrast-enhanced ultrasound (CEUS) features, and were graded automatically by AmCAD-UT Detection. The receiveroperator characteristic (ROC) curve was established based on the gold standard of postoperative pathological results, and the area under the curve(AUC)of the three methods was compared. And the applicable cutoff point for differentiating benign and malignant thyroid nodules was found, and the sensitivity, specificity and accuracy of CAD were compared with conventional two-dimensional ultrasound and two-dimensional ultrasound combined CEUS,and analyzed the difference of ultrasonic features described by CAD system in benign and malignant nodules to evaluate the diagnostic efficacy.Result The diagnostic ability of the three methods was compared: two-dimensional combined CEUS > CAD > conventional two-dimensional ultrasound(The AUC of the three methods was 0.930,0.859and 0.833, respectively). The sensitivity and accuracy of CAD were higher than conventional two-dimensional ultrasound,the specificity was lower than conventional two-dimensional ultrasound,and the specificity and accuracy of two-dimensional combined CEUS were higher than those of CAD(P＜0.05). Hypoechogenicity、microcalcifications、irregular margins were available for identifying benign and malignant nodules.Conclusion the AmCAD-UT Detection can detect sensitively thyroid malignant nodules, improve the accuracy of diagnosis, quantify features of thyroid nodules comprehensively and accurately, and have a significant clinical application value. |
| Key words: AmCAD-UT Detection thyroid nodules two-dimensional ultrasound contrast-enhanced ultrasound TI-RADS |
