|
| Author | Year | Design | Dataset | Training cohort | Validation cohort | Test cohort | Model | Outcome | Performance reported |
|
| Baihua Zhang | 2021 | Retrospective multicenter on CT | 914 LUAD | 638 | NA | 71 internal; 205 external | SE-CNN + radiomics mapping | EGFR mutation | AUC 0.910 and 0.841 in internal and external test cohorts, respectively |
|
| Wei Mu | 2020 | Retrospective multicenter on PET/CT | 681 NSCLCs | 429 | 187 | 65 external | CNN | EGFR mutation treatment response | AUC 0.86, 0.83, and 0.81 in the training, internal validation, and external test cohorts, respectively |
|
| Shuo Wang | 2019 | Retrospective multicenter on CT | 844 LUAD | 603 | Five-fold cross validation; 241 independent | NA | CNN | EGFR mutation | AUC 0.85 in the primary cohort; AUC 0.81 in the independent validation cohort |
|
| Wei Zhao | 2019 | Retrospective multicenter on CT | 616 LUAD | 348 | 116 | 115 internal; 37 public | CNN 3D DenseNets | EGFR mutation | AUC 0.758 and 0.750 in the internal test set and public test set |
|
| Junfeng Xiong | 2018 | Retrospective single-center on CT | 503 LUAD | 345 | 158 | NA | CNN | EGFR mutation | An AUC (CNN) of 0.776 and an AUC (a fusion model of CNNs and clinical features) of 0.838 in the validation set |
|
| Panwen Tian | 2021 | Retrospective multicenter on CT | 939 NSCLCs | 750 | 93 | 96 | KNN | PD-L1 expression treatment response | AUC 0.78, 0.71, and 0.76 in the training, validation, and test cohorts |
|
| Ying Zhu | 2020 | Retrospective single-center on CT | 127 LUAD | NA | Five-fold cross validation | NA | CNN 3D DenseNets | PD-L1 expression | AUC more than 0.750 |
|
| Zhengbo Song | 2020 | Retrospective multicenter on CT | 1028 NSCLCs | 651 | 286 | 91 | CNN 3D ResNet10 | ALK fusion status
Treatment response | AUC(CNN) 0.8046 and 0.7754 in the primary and validation cohorts, AUC (trained by both CT images and clinicopathological information) 0.8540 and 0.8481 in the primary and validation cohorts |
|
