Table 3.
Summary of studies evaluating the role of radiomics on differentiating IPMNs with and without advanced neoplasia.
| Author, Location, Year | Inclusion Criteria | Number of Patients | Image Type | Number of Radiomic Features | Best Model | Performance Training Set | Performance Internal Validation Set |
|---|---|---|---|---|---|---|---|
| Hanania, USA, 2016 | Surgically resected IPMN (2003–2011) | 53 (34 HGD, 19 LGD) | CECT | 360 | 10 radiomic features | AUC: 0.82 Sn: 85% Sp: 68% (*) |
AUC: 0.96 Sn: 97% Sp: 81% |
| Permuth, USA, 2016 | Surgically resected IPMN (2006–2011) | 38 (20 HGD, 18 LGD) | CECT | 112 | 14 radiomic features +blood 5 mi-RNAs | AUC: 0.92 Sn: 83% Sp: 89% PPV: 88% NPV: 85% |
AUC: 0.87 |
| Attiyeh, USA, 2019 | Surgically resected BD-IPMN (2005–2015) | 103 (27 HGD, 76 LGD) | CECT | 255 | Radiomic + clinical features | AUC: 0.79 Sn: 71% Sp: 82% PPV: 95% NPV: 79% |
-- |
| Harrington, USA, 2020 | Surgically resected IPMN | 33 (7 HGD, 26 LGD) |
CECT | 13 | Radiomic features + cyst fluid protein markers | AUC: 0.88 Sn: 71% Sp: 92% PPV: 71% NPV: 92% |
-- |
| Hoffman, USA, 2017 | Pathology proven BD-IPMN (2006–2015) | 18 (8 HGD, 10 LGD) |
MRI with DWI | -- | Entropy | AUC: 0.86 Sn: 100% Sp: 70% |
-- |
* Estimates obtained with the highest performing individual radiomic feature. IPMN: intraductal papillary mucinous neoplasm; BD: branch-duct; HGD: high-grade dysplasia; LGD: low-grade dysplasia; CECT: contrast-enhanced computed tomography; DWI: diffusion weighted imaging; AUC: area under the curse; Sn: sensitivity; Sp: specificity; PPV: positive predictive value; NPV: negative predictive value.