18 June 2020 : Clinical Research
Ensemble Deep Learning Model for Multicenter Classification of Thyroid Nodules on Ultrasound Images
Xi Wei1ACDEG*, Ming Gao2BEF, Ruiguo Yu3BCD, Zhiqiang Liu3BCE, Qing Gu4BC, Xun Liu5BC, Zhiming Zheng6BC, Xiangqian Zheng2BC, Jialin Zhu1ABCDEF, Sheng Zhang1BDDOI: 10.12659/MSM.926096
Med Sci Monit 2020; 26:e926096
Table 4 Comparison of the diagnostic performance of EDLC-TN with other four state-of-the-art algorithms.
| AUC | Sensitivity (%) | Specificity (%) | Accuracy (%) | |
|---|---|---|---|---|
| EDLC-TN | 0.941 (0.936–0.946) | 93.77 | 94.44 | 98.51 |
| ResNeXt | 0.882 (0.875–0.889)* | 85.53 | 90.86 | 82.83 |
| SE_Inception_v4 | 0.874 (0.866–0.881)* | 90.33 | 84.38 | 97.12 |
| SE_Net | 0.840 (0.832–0.848)* | 88.64 | 79.35 | 96.52 |
| Xception | 0.880 (0.872–0.887)* | 84.68 | 91.26 | 93.84 |
| EDLC-TN – ensemble deep learning classification model of thyroid nodules; AUC – area under the ROC curve; AUCs of and other three models were calculated by the method of DeLong et al. – The difference of AUCs between the and other four models was compared by Z-test, * | ||||