Table 3.
AI-based US studies for HCC.
| N | Image | Target | AUC | Reference |
|---|---|---|---|---|
| 668 | B-mode | Histopathological types of primary liver cancer | HCC v. non-HCC 0.775 Intrahepatic cholangiocarcinoma vs. combined hepatocellular–cholangiocarcinoma 0.728 |
[79] |
| 2143 | B-mode | Liver cancer diagnosis | 0.924 for focal hepatic lesions Sensitivity (86.5% vs. 76.1%, p = 0.0084) Specificity (85.5% vs. 76.9%, p = 0.0051) Both superior to 15-year skilled radiologists |
[47] |
| 47 | B-mode, share wave elastography and viscosity imaging | Ki-67 | 0.94 | [48] |
| 47 | B-mode, share wave elastography and viscosity imaging | Microvascular invasion | 0.98 | [48] |
| 482 | CEUS | Microvascular invasion | 0.731 | [49] |
| 322 | B-mode | Microvascular invasion | 0.726 (0.806 for differentiation between M1 and M2) - |
[80] |
| 130 | B-mode, CEUS | Response to TACE | 0.93 by CEUS 0.80 by time–intensity curve (p = 0.034 vs. CEUS) 0.81 by B-mode (p = 0.039 vs. CEUS) |
[81] |
| 47 | B-mode, share wave elastography and viscosity imaging | Programmed cell death-1 | 0.97 | [48] |
| 318 | CEUS | Post-ablation early recurrence | 0.84 | [69] |
CEUS, contrast-enhanced ultrasound.