Skip to main content
. 2023 Aug 12;34(1):436–443. doi: 10.1007/s00330-023-10095-3

Table 1.

Breakdown of article characteristics

Characteristics n (%)
Publication year
2021 106
2022 151
Journal
European Radiology 195 (75.9%)
European Journal of Radiology 44 (17.1%)
Radiology 12 (4.7%)
Radiology: Imaging Cancer 4 (1.6%)
Radiology: Artificial Intelligence 2 (0.8%)
Topic
Abdominal 52 (20.2%)
Chest 52 (20.2%)
Neuroradiology 43 (16.7%)
Genitourinary 36 (14.0%)
Head and neck 26 (10.1%)
Breast 21 (8.2%)
Musculoskeletal 15 (5.8%)
Cardiovascular 4 (1.6%)
Oncology 3 (1.2%)
Pediatric 3 (1.2%)
Chest/neuroradiology 1 (0.4%)
Chest/abdominal 1 (0.4%)
Modality
MRI 110 (42.8%)
CT 109 (42.4%)
PET/CT 17 (6.6%)
US 15 (5.8%)
Mammography 3 (1.2%)
Angiography 1 (0.4%)
X-ray 1 (0.4%)
MRI, CT 1 (0.4%)
Study design
Retrospective 248 (96.5%)
Prospective 8 (3.11%)
Model utility
Classification 144 (56.03%)
Prognostication 79 (30.74%)
Detection 30 (11.67%)
Classification, prognostication 4 (1.55%)
Model type
LASSO/LR 139 (54.4%)
ML classifiers* 64 (24.9%)
RF** 20 (7.7%)
SVM*** 20 (7.7%)
DL 11 (4.2%)
Not mentioned 3 (1.1%)

Abbreviations: LASSO least absolute shrinkage and selection operator, LR logistic regression, RF random forest, SVM support vector machine, DL deep learning, ML machine learning

*In most cases, multiple ML modeling methods were combined. Some of the methods are listed here: linear classifier, K-nearest neighbor, passive-aggressive classifier, perceptron, ridge classifier, AdaBoost, Naïve-Bayes, ElasticNet, Gradient Boosting Decision Tree, AutoML Ensemble

**The number of papers only used the random forest method for modelling, but not as a part of combined ML classifiers

***The number of papers only used the support vector machine method for modelling, but not as a part of combined ML classifiers