Table 6.
Applications of radiomics-based metastasis prediction
| Image modality | Number of patients | Cancer | Target | Number of radiomics features | Commercial or open-source software | Method | References |
|---|---|---|---|---|---|---|---|
| FDG-PET | 174 | OPC | The risk of DM | 2–3 | Matlab, Stata/MP |
ML: LOOCV, Cox proportional-hazards regression, Fine and Gray’s proportional sub-hazards model, LR, fivefold CV SM: Kaplan–Meier analysis, log-rank test, Spearman correlation analysis |
[29] |
| MRI | 176 | NPC | DM | 7 | PyRadiomics, Python, ITK-SNAP, R, SPSS |
ML: mRMR, LASSO, LR, Mutual information, Bootstrap-resampling SM: ICC, t-test, Kaplan–Meier analysis, log-rank test, Fisher's exact test, Chi-square test, or Mann–Whitney U test |
[39] |
| MRI | 236 | Tongue cancer | LNM | 15/17/18/25/10 | ITK-SNAP, AIMT, Python, R, SPSS |
ML: PCA, SVM, Cox regression analysis, fivefold CV SM: DeLong test, Spearman correlation analysis, Kaplan–Meier analysis, log-rank test |
[34] |
| MRI | 346 | Rectal cancer | LNM | 4/5/10 | GE Healthcare, 3D Slicer, R, SPSS |
ML: LASSO, LR, Cox analysis SM: ICC, Wilcoxon test, Hosmer–Lemeshow test, t-test, Nonparametric test, Chi-square test, and Fisher’s exact test, DeLong test |
[35] |
| US | 126 | Thyroid cancer | LNM | 91 | ITK-SNAP, Ultrosomics, SPSS |
ML: LASSO, PCA, DT, Naive Bayes, KNN, LR, SVM, Bagging, RF, Extremely RF, AdaBoost, Gradient boosting DT SM: t-test, Chi-square test or Fisher’s exact test |
[41] |
| US | 205 | NPC | LNM | 7 | GE Healthcare, R, Python |
ML: mRMR, LR, LASSO SM: ICC, DeLong test |
[42] |
| PET | 76 | Primary prostate cancer | LNM, DM | 22 | RaCaT, Python |
ML: RF, CV, PCA SM: Chi-square test, DeLong test, ICC, Z-score |
[43] |
CT computed tomography, MRI magnetic resonance imaging, FDG fluorodeoxyglucose, PET positron emission tomography, US ultrasonography, ML machine learning, SM statistical method, OPC oropharyngeal cancer, NPC nasopharyngeal carcinoma, DM distant metastasis, LNM Lymph node metastasis, LOOCV leave one out cross validation, LR logistic regression, CV cross validation, mRMR maximum relevance minimum redundancy, LASSO least absolute shrinkage and selection operator, ICC intraclass correlation coefficients, PCA principal component analysis, SVM support vector machine, DT decision tree, KNN K-nearest neighbors, RF random forest, AdaBoost adaptive boosting