Table 2.
Radiomics studies in NSCLC with an aspect of biology as the endpoint. The column labeled ‘Data’ specifies the total number of patients involved in the study, in brackets split by training and validation cohorts if applicable and specifying other cancer types of cohorts if applicable. This table has been simplified to clarify presentation – more details for each study are available in Supplementary Table 3.
| Reference | Stage | Endpoint | Data (training + validation) | Radiomic features in final model | Result |
|---|---|---|---|---|---|
| Genomics | |||||
| Aerts et al. 2016 [56] | Early stage | EGFR | 47 diagnostic CT and follow-up | Shape and texture | AUC = 0.74−0.91 |
| Rios Velazquez et al. 2017 [57] | 1−4 | EGFR, KRAS | 705 diagnostic CT (353 + 352) |
Shape, first order and texture | AUC = 0.69−0.80 |
| Mei et al. 2018 [58] | Not specified | EGFR | 296 CT | Texture | AUC = 0.664 |
| Digumarthy et al. 2019 [59] | Not specified | EGFR | 93 CT | First order | AUC = 0.713 |
| Jia et al. 2019 [60] | 1−4 | EGFR | 504 CT (345 + 158) |
Shape, first order and texture | AUC = 0.802 |
| Li et al. 2019 [61] | 1−4 | EGFR subtypes (19Del and L858R) | 312 CT (236 + 76) |
Shape, first order and texture | AUC = 0.775−0.793 |
| Tu et al. 2019 [62] | 1−4 | EGFR | 404 CT (243 + 161) |
First order and texture | AUC = 0.775 |
| Yang et al. 2019 [63] | Not specified | EGFR | 467 CT (306 + 161) |
Shape, first order and texture | AUC = 0.789 |
| Wang et al. 2019 [64] | 1−2 | EGFR, TP53 | 61 CT (41 + 20) |
First order and texture | AUC = 0.604, 0.586 |
| Wang et al. 2019 [64] | 1−2 | Tumor mutation burden | 61 CT (41 + 20) |
Texture | AUC = 0.606 |
| Signaling pathways | |||||
| Grossman et al. 2017 [15] | 1−3 | Various | 351 CT (262 + 89) |
Shape, first order and texture | AUC = 0.62−0.72 |
| Bak et al. 2018 [65] | 1−4 | Various | 57 CT | First order and texture | OR = 0.08−23.94 |
| Histopathology | |||||
| Patil et al. 2016 [66] | Not specified | ADC, LCC, SCC, NOS | 317 pCT | Shape, first order and texture | 88 % accuracy |
| Wu et al. 2016 [67] | 1−4 | ADC, SCC | 350 pCT (198 + 152) |
First order and texture | AUC = 0.72 |
| Ferreira Junior et al. 2018 [38] | 1−4 | ADC, SCC | 68 CT (52 + 16) |
Not specified | AUC = 0.81 |
| Zhu et al. 2018 [68] | Not specified | ADC, SCC | 129 CT (81 + 48) |
First order and texture | AUC = 0.893 |
| Digumarthy et al. 2019 [59] | Not specified | ADC, SCC | 93 CT | First order | AUC = 0.744 |
| E et al. 2019 [69] | Not specified | ADC, SCC, SCLC | 229 CT | Shape, first order and texture | AUC = 0.657−0.875 |
| Ferreira-Junior et al. 2019 [45] | 1−4 | ADC, SCC | 85 CT | Shape, first order, texture | AUC = 0.88 |
| Liu et al. 2019 [70] | Not specified | ADC, LCC, SCC, NOS | 349 CT (278 + 71) |
Not specified | AUC = 0.86 |
| Zhou et al. 2018 [71] | 1−4 | Ki-67 | 110 CT | Shape and texture | AUC = 0.61−0.77 |
| Gu et al. 2019 [72] | Not specified | Ki-67 | 245 CT | First order and texture | AUC = 0.776 |
| Song et al. 2017 [73] | 1−3 | Micropapillary pattern | 339 CT | First order | AUC = 0.751 |
| Chen et al. 2018 [74] | Not specified | Degree of differentiation | 487 CT (303 + 184) |
First order and texture | AUC = 0.782 |
| She et al. 2018 [75] | Not specified | Invasive vs non-invasive adenocarcinoma | 402 CT (207 + 195) |
Shape, first order and texture | AUC = 0.89 |
| Yang et al. 2019 [76] | Not specified | Invasive vs non-invasive adenocarcinoma | 192 CT (116 + 76) |
First order and texture | AUC = 0.77 |
Abbreviations: ADC, adenocarcinoma; AUC, area under the curve; CI, concordance index; EGFR, epidermal growth factor receptor; KRAS, Kirsten rat sarcoma viral oncogene homolog; LCC, large cell carcinoma; NOS, not otherwise specified; OR, odds ratio; SCC, squamous cell carcinoma.