Table 3.
Summary of papers on AI for stratification of PDAC patients. The performance for the validation and test sets is reported with respective 95% Confidence Interval or standard deviation when it was provided.
| Authors (Year) | Ground Truth | Data | Approach | Model | Metric | Validation Performance | Test Performance | Dev. Cohort |
Test Cohort |
|---|---|---|---|---|---|---|---|---|---|
| An et al. (2021) [63] | LNM | CT + clinical | DL | Resnet-18 | AUC | 0.90 (0.88–0.92) |
* 0.92 (0.91–0.92) |
113 | * 35 |
| Chaddad et al. (2020) [64] | Short term vs. long-term survival | CT | DL + ML | CNN + RF | AUC | 0.72 | .. | 159 | .. |
| Song et al. (2013) [67] |
Grading 1 vs. 2 |
WSI | ML | SVM | AUC | 0.79 | .. | 240 | .. |
| Bianet al. (2022) [68] | LNM | MR | Radiomics | LR | AUC | 0.75 (0.68–0.82) |
* 0.81 (0.69–0.94) |
180 | * 45 |
| Shi et al. (2022) [65] | LNM | MR + clinical | Radiomics | LR | AUC | 0.909 (0.854–0.964) |
* 0.835 (0.751–0.919) ** 0.805 (0.720–0.890) |
199 | ** 52 |
| Bian et al. (2021) [69] |
TIL | MR | Radiomics | XGBoost | AUC | 0.86 (0.79–0.93) |
* 0.79 (0.64–0.93) |
116 | * 40 |
| Cen et al. (2021) [70] | Stage I–II vs. Stage III–IV |
CT | Radiomics | LR | AUC | 0.940 (0.871–0.979) |
* 0.912 (0.781–0.978) |
94 | * 41 |
| Zhang et al. (2021) [71] | Liver metastasis vs. other metastasis | CT | Radiomics | RF | AUC | 0.81 | .. | 77 | .. |
| Xing et al. (2021) [72] | Grading 1 vs. 2/3 |
18FDG PET-CT | Radiomics | XGBoost | AUC | .. | * 0.921 (0.846–0.996) |
99 | * 50 |
| Kaissis et al. (2020) [73] | QMS | CT | Radiomics | RF | AUC | 0.93 ± 0.01 | .. | 181 | .. |
| Chen et al. (2020) [74] | PV-SMV invasion | CT | Radiomics | ElasticNet | AUC | 0.871 (0.795–0.946) |
* 0.848 (0.724–0.971) |
88 | 58 |
| Liu et al. (2020) [75] |
LNM | CT | Radiomics | LR | AUC | 0.841 (0.768–0.925) |
.. | 85 | .. |
| Li et al. (2020) [76] |
LNM | CT + clinical | Radiomics | LR | AUC | .. | * 0.912 (0.778–1) |
118 | *41 |
| Chang et al. (2020) [66] | Grading 1/2 vs. 3 |
CT | Radiomics | LASSO | AUC | 0.961 (0.935–0.987) |
* 0.91 (0.864–0.956) ** 0.77 (0.661–0.878) |
151 | * 150 ** 100 |
| Longlong et al. (2020) [77] |
Grading 1 vs. 2 vs. 3 |
CT | Radiomics | RF | AUC | 0.77 (0.64–0.87) |
* 0.70 (0.47–0.86) |
58 | * 25 |
| Qiu et al. (2019) [78] |
Grading 1/2 vs. 3 |
CT | Radiomics | SVM | SEN, SPE, ACC |
78 95 86 |
.. | 56 | .. |
** external test set, * internal test set. Abbreviations are: LNM—lymph node metastasis, TIL—tumour infiltrating lymphocytes, Grading—grade comparison (low vs. high), QMS—quasi mesenchymal subtype, PV-SMV—portal vein superior mesenteric vein, DL—deep learning, ML—machine learning, SVM—supported vector machine, RF—random forest, LR—logistic regression, CNN—convolutional neural network, XGBoost—extreme gradient boost, AUC—area under the receiver operating characteristic curve, SEN—sensitivity, SPE—specificity, ACC—accuracy, Dev. Cohort—development cohort (training + validation).