Table 2.
Mains studies regarding histology and radio-genomic characterization.
| Reference | Application | Number of cases | Imaging modality | Algorithm | Segmentation | Feature types | No of features | Validation | Results |
|---|---|---|---|---|---|---|---|---|---|
| Histology subtypes | |||||||||
| Wu et al. (40) | Prediction of histology subtype | 350 (198 for Training) | CT | Naïve Baye’s classifier | Manually segmented | Shape, 1st order, 2nd order |
5 | Independent | AUC 0,72 |
| Raniery Ferreira et al. (41) | Prediction of histology subtype | 68 (52 for Training) | CT | RBF-based ANN | Semi-Automatically segmented | Shape, 1st order, 2nd order |
100 | Sample split | AUC 0,71 |
| Genomic alterations | |||||||||
| Zhang et al. (45) | Prediction of EGFR mutation | 180 (140 for Training) | CT | multivariate analysis | Manually segmented | Clinical, Shape, 1st order, 2nd order |
7 | Sample split | AUC 0,87 |
| Velazquez et al. (46) | Prediction of EGFR and KRAS mutation | 381 (190 for Training) | CT | RFC | Manually segmented | Clinical, Shape, 1st order, 2nd order |
25 | Independent | AUC 0,86 |
| Zhao et al. (47) | Prediction of EGFR subtype | 637 (322 for Training) | CT | multivariate analysis | Manually segmented | Clinical, Shape, 1st order, 2nd order |
11 | Sample split | AUC 0,76 |
| Wang et al. (48) | Prediction of EGFR mutation | 843 (603 for Training) | CT | CNN | Manual segmentation | Deep features | NE | Independent | AUC 0,81 |
| Zhang et al. (49) | Prediction of EGFR mutation | 248 (175 for Training) | PET, CT | Logistic regression | Semi-Automatically segmented | Clinical, Shape, 1st order, 2nd order |
13 | Sample split | AUC 0,87 |
| Yoon et al. (51) | Prediction of ALK status | 539 | PET, CT | Logistic regression | Semi-Automatically segmented | Clinical, Shape, 1st order, 2nd order |
7 | Cross validation | sensitivity and specificity, 0.73 and 0.70, respectively |
ALK, anaplastic lymphoma kinase; ANN, artificial neural network; AUC, area under the curve; CNN, convolutional neural network; CT, computed tomography; EGFR, epidermal growth factor receptor, KRAS, Kirsten rat sarcoma viral oncogene homolog, NE, not evaluable; PET, positron emission tomography; RBF, radial basis function; SVM, support vector machine.