Table 2.
Radiomics analysis details of included studies
| Study | Segmentation and software | Observers and agreement | Feature extraction software | Non-radiomics features | Validation dataset | Model type | Phase classification |
|---|---|---|---|---|---|---|---|
| Nie [20] | Manual; ITK-SNAP | 2 URs; ICC | Radiomics cloud platform | Clinical parameters | Separate data from other institution | 3 | 1 |
| Shi [21] | Manual; Image J | Not documented | Image J | None | Exactly the same data | 1a | 0 |
| Wang [22] | Manual; not documented | 2 SRs; none | Pyradiomics | None | Tenfold cross-validation | 1b | 0 |
| Wang [23] | Manual; Image J | 2 SRs; none | Pyradiomics | Clinical parameters | Tenfold cross-validation | 1b | 0 |
| Wu [24] | Manual; ITK-SNAP | 2 URs; ICC | Artificial intelligence kit | Radiologists’ assessment | Exactly the same data | 1a | 0 |
| Yin [25] | Manual; ITK-SNAP | 2 SRs; ICC | Artificial intelligence kit | None | Randomly splitting data | 2a | 0 |
| Yin [26] | Manual; ITK-SNAP | 2 URs; ICC | Artificial intelligence kit | None | Randomly splitting data | 2a | 1 |
| Yin [27] | Manual; ITK-SNAP | 2 SRs; ICC | Artificial intelligence kit | Clinical parameters | Randomly splitting data | 2a | 1 |
| Yin [28] | Semi-automatic; MITK | 2 SRs; ICC | Artificial intelligence kit | Clinical parameters | Randomly splitting data | 2a | 1 |
ICC intraclass coefficient, SR radiologist with relevant subspecialist expertise, UR radiologist with unspecified expertise