Table 1. Summary of artificial intelligence studies on ultrasound features identifying benign and malignant thyroid nodules
Study | Patients | Method | Accuracy (%) | Sensitivity (%) | Specificity (%) |
Zhao et al.20) | 106 | ML/SVM | 82 | 91 | 78 |
Zhang et al.22) | 826 | ML/SVM | 83 | 86.1 | 82.7 |
Park et al.21) | 286 | ML/SVM | 75.9 | 90.4 | 58.8 |
Ko et al.25) | 519 | DL/CNN | 87.3 | 90 | 82 |
Wu et al.24) | 1,396 | DL/CNN | 82 | 85 | 78 |
Kim et al.23) | 757 | DL/CNN | 85.1 | 81.8 | 86.1 |
ML: machine learning, SVM: support vector machine, DL: deep learning, CNN: convolutional neural network.