Table 5 Performance metrics of proposed models compared with earlier research using ANOVA FS
| Â | Methods | Database | Results (%) | ||||
|---|---|---|---|---|---|---|---|
AUROC | Accuracy | Sensitivity | Precision | Specificity | |||
Cai, J et al. [43] | Deep Learning | LIDC | 88.1 | 84.6 | 83.7 | – | 85.2 |
Wang et al. [44] | Deep Learning | LIDC | 92.75 | 85.23 | 92.79 | 84.56 | 72.89 |
Proposed model 1 | Radiomics + FGSVM | LIDC | 88.3 | 83.7 | 97.8 | 83 | 45.3 |
Proposed model 2 | Radiomics + MGSVM | LIDC | 90.4 | 86.4 | 92.3 | 89.5 | 70.6 |
Proposed model 3 | Radiomics + CGSVM | LIDC | 89.5 | 82.7 | 91.7 | 85.6 | 57.9 |
Proposed model 4 | Radiomics + Decision Tree | LIDC | 87.8 | 85.6 | 91.0 | 89.5 | 70.9 |
Proposed model 5 | Radiomics + Ensemble Boosted Trees | LIDC | 90.5 | 86.6 | 91.7 | 90.2 | 72.7 |
Proposed Model 6 | Radiomics + Ensemble Bagged Trees | LIDC | 91.0 | 86.6 | 90.5 | 91.2 | 76.1 |
Proposed model 7 | Radiomics + Ensemble RUSBoosted Trees | LIDC | 91.4 | 85.2 | 86.1 | 93.2 | 82.9 |
Proposed model 8 | Radiomics + Ensemble Subspace Discriminant | LIDC | 90.2 | 84.3 | 94.8 | 85.4 | 55.9 |
Proposed model 9 | Radiomics + Ensemble Subspace KNN | LIDC | 90.9 | 86.9 | 92.9 | 89.6 | 70.6 |