Derived aims and key findings of studies examining 1p19q status of IDH-mut LGG
| First Author and Year | Derived Aim | Key Findings |
|---|---|---|
| Han 202035 | To determine if clinical and standard imaging factors improve classification | The AUC (95% CI) = 0.753 (0.654–0.852) for clinical plus radiomic features versus AUC = 0.760 (0.663-0.857) for just radiomic features; radiomic features were superior to clinical features alone, AUC = 0.627 (0.551–0.703) |
| Kocak 202026 | To determine the best ML classifier | The neural network produced the highest AUC (95% CI) = 0.869 (0.751–0.981); sensitivity of 87.5%, specificity of 75.8% |
| Lu 201828 | To determine the best ML classifier | Classification occurred with an AUC = 0.92 (sensitivity of 88.5%, specificity of 86.2%) using quadratic SVM |
| Shofty 201831 | To determine the best ML classifier | Classification occurred with an AUC = 0.87 (sensitivity of 92%, specificity of 83%) using ensemble bagged trees classifier |
| Zhou 201734 | To determine if VASARI annotations were superior to standard radiomic analysis for classification | Texture features classified with an AUC (± 95% CI) = 0.96 ± 0.01; sensitivity of 90% ± 2%, specificity of 89% ± 2%VASARI features classified with an AUC = 0.78 ± 0.02; sensitivity of 72% ± 3%, specificity of 67% ±3% |
| Fukuma 201922 | To determine if integration of CNN deep learning with radiomic features improved classification | Conventional radiomic features (± 95% CI): accuracy = 59.0 ± 9.0%; AUC = 0.656 ± 0.113CNN features: accuracy = 84.0 ± 9.3%; AUC = 0.868 ± 0.099CNN and conventional radiomic features: accuracy = 79.8 ± 11.0%; AUC = 0.861 ± 0.116 |