Table 3.
Radiomics studies in pancreatic cancer detection and diagnosis.
| Reference | Image | Software | Endpoints | Segmentation Process (Number of Readers) | Sample Size (Training + Validation) | Number of Features Extracted | Results |
|---|---|---|---|---|---|---|---|
| Benedetti et al. [48] |
CT | In house with Matlab | Discriminating histopathologic characteristics of PNET | Manual (1) | 39 | 69 | Best AUC = 0.86 |
| Bevilacqua et al. [49] |
PET/CT | In house with Matlab | Grade 1 vs. 2 primary PNET | Manual (1) | 25 + 26 (model A) 26 + 25 (model B) 51 (model C) |
60 | Best performance was achieved by model A test AUC = 0.90 |
| Bian et al. [50] |
MRI | Pyradiomics | G1 vs. G2/3 grades in patients with PNETs | Manual (2) | 157 | 1409 | AUC = 0.775 |
| Bian et al. [51] |
MRI | Pyradiomics | PNET grades | Manual (1) | 97 + 42 | 3328 | AUC = 0.851 (training) AUC = 0.736 (validation) |
| Canellas et al. [52] |
CT | TexRAD | Differentiating PNET grades |
Manual (2) | 101 | 36 | Accuracy of 79.3% for differentiating grade1 vs. grades 2/3. |
| Chang et al. [53] |
CT | IBEX | Histological grades of PDAC | Manual (2) | 151 + 150 (local) +100 (external validation) | 1452 | AUCs = 0.961 (training), AUC = 0.910 (local validation), and AUC = 0.770 (external validation) |
| Chen et al. [54] |
CT | Pyradiomics | Differentiating PDAC from normal pancreas | Manual (2) | 915 + 200 (local test) + 264 (external test) | 88 | AUC = 0.98 (local test) AUC = 0.91 (external test) |
| Chu et al. [55] |
CT | Pyradiomics | Differentiating PDAC from normal pancreas | Manual (3) | 255 + 125 | 478 | AUC = 0.999 |
| Deng et al. [56] |
MRI | IBEX | DifferentiatingPDAC and MFCP lesions | Manual (2) | 64 + 55 | 410 | AUCs for the T1WI, T2WI, A and, P and clinical models were 0.893, 0.911, 0.958, 0.997 and 0.516 in the primary cohort, and 0.882, 0.902, 0.920, 0.962 and 0.649 in the validation cohort. |
| Gu et al. [57] |
MRI | Artificial Intelligence Kit | SPN vs. differential diseases (PDAC, NET, and cystadenoma) | manual (2) | 48 + 113 | 2376 | In validation, AUC = 0.853 for T2 (best performing single sequence), AUC = 0.925 for multi-parametric MRI radiomics model, and AUC = 0.962 for radiomics + clinical model. |
| Li et al. [58] |
CT | Fire Voxel | Atypical PNET vs. PDAC | Manual (2) | 75 | <20 | Best AUC = 0.887 |
| Linning et al. [59] |
CT | In house with Matlab | PDAC vs. autoimmune pancreatitis | Manual (2) | 96 (5-fold cross validation) | 1160 | AUC = 0.977 |
| Liu et al. [60] |
PET/CT | Pyradiomics | PDAC vs. autoimmune pancreatitis | Manual (2) | 112 (10-fold cross validation) | 502 | AUC= 0.967 |
| Liu et al. [61] |
CT and MRI | Pyradiomics | PNET grades | Manual (2) | 82 + 41 | 1209 | AUC = 0.92 (training) AUC = 0.85 (validation) |
| Park et al. [62] |
CT | Pyradiomics | PDAC vs. autoimmune pancreatitis | Manual (4) | 120 + 62 | 431 | AUC = 0.975 |
| Reinert et al. [63] |
CT | Pyradiomics | Differentiating PDAC from PanNEN | Manual (1) | 95 | 92 | 8 features highly significant (p < 0.005) |
| Ren et al. [64] |
CT | Analysis Kit software | Pancreatic adenosquamous carcinoma vs. PDAC | Manual (1) | 112 7:3 ratio |
792 | Average AUC of 0.82 |
| Song et al. [65] |
MRI | Pyradiomics | Differentiating NF-PNET and SPN | Manual (2) | 79 (7:3 ratio) | 396 | AUC = 0.978 (radiomics) and AUC = 0.965 (radiomics + clinical) in the training set AUC = 0.907 (radiomics) and AUC = 0.920 (radiomics + clinical) in the validation set |
| Xing et al. [66] |
PET/CT | Pyradiomics | Pathological grades in PDAC | Manual (2) | 99 + 50 | about 3000 | AUC o = 0.994 (training) AUC = 0.921 (validation) |
| Zhang et al. [67] |
CT | LifeX | Pathological grades of PNETs | Manual (3) | 82 3:1 ratio |
40 | AUC = 0.82 (G1 vs. G2), 0.70 (G2 vs. G3), and 0.85 (G1 vs. G3), respectively |
| Zhao et al. [68] |
CT | In house with Matlab | Grade 1 vs. 2 in PNET | Manual (2) | 59 + 40 | 585 | AUC = 0.968 (training) AUC= 0.876 (validation) |
Abbreviations used in this table: Area Under Curve (AUC), Mass-forming Chronic Pancreatitis (MFCP), Pancreatic Neuroendocrine Neoplasm (PanNEN), Pancreatic Adenocarcinoma (PDAC), Neuroendocrine Tumor (NET) or Pancreatic Neuroendocrine Tumor (PNET), Solid Pseudopapillary Neoplasm (SPN).