| | Author | Year | Nation | Study design | Sample size | Extracted features of AI | Software |
| | Feng C et al. [26]. | 2018 | China | Retrospective | 90 | First-order statistics: Mean attenuation; 10th, 25th, 50th, 75th, and 90th percentile attenuation; skewness; kurtosis; entropy | CT kinetics | | Wang C et al. [27]. | 2019 | China | Retrospective | 333
Training cohort = 233
Validation cohort = 100 | First-order (histogram), haralick features, GLCM, GLRLM | AK | | Chen T et al. [25]. | 2019 | China | Retrospective | 222
Training cohort = 130
Validation cohort = 92 | GLV, GLRLM, GLSZM, NGTDM, GLSZM | MATLAB | | Yan J et al. [28]. | 2018 | China | Retrospective | 213 | First-order (histogram) gradient features, GLCM, GLRLM | MaZda | | Liu S et al. [29]. | 2018 | China | Retrospective | 78 | First-order (histogram) | Image analyzer | | Zhang L et al. [30]. | 2020 | China | Retrospective | 140
Training cohort = 100
Validation cohort = 40 | First-order features, shape and size features, second-order features (GLCM, GLRLM, GLSZM) features, and haralick features | AK | | Choi I et al. [24]. | 2019 | Korea | Retrospective | 145 | First-order statistics: Mean SD of mean, entropy, MPP, skewness, and kurtosis. Geometry with Gaussian filtration | MATLAB | | Ning Z et al. [31]. | 2018 | China | Retrospective | 231
Training cohort = 130
Validation cohort = 101 | First-order, second-order (GLCM, GLRLM, GLSZM, and NGTDM) features | MATLAB PYTHON | | Zhang Q et al. [32]. | 2020 | China | Retrospective | 339
Training cohort = 148
|Internal validation cohort = 41
External validation cohort = 150 | First-order statistics, features of shape, second-order features (GLCM, GLRLM, GLSZM) | PYTHON | | Li X et al. [33] | 2020 | China | Retrospective | 915
Training cohort = 680
Validation cohort = 54
Testing cohort = 181 | First-order (histogram), second-order (GLCM, GLRLM, GLSZM, NGTDM) and wavelet-filtered features | MATLAB |
|
|
Note. GLCM = gray-level cooccurrence matrix, GLRLM = gray-level run-length matrix, GLV = gray-level variance, GLSZM = gray-level size-zone matrix, NGTDM = Neighborhood gray-tone difference matrix.
|
