Table 2.

Characteristics of included studies

Characteristics	No. of studies (%)
Journal speciality
Imaging	28/57 (49.1%)
Clinical	20/57 (35.1%)
Generalist	7/57 (12.3%)
Medical physics	2/57 (3.5%)
Specialty of the 1st author*
Radiologist	36/57 (63.2%)
Radiotherapist	1/57 (1.8%)
Clinician	11/57 (19.3%)
Physicist, Computer scientist	9/57 (15.8%)
Geographical origin of the study
Asia	43/57 (75.4%)
Europe	6/57 (10.5%)
North America	8/57 (14.0%)
Imaging modality
CT	37/57 (64.9%)
MRI	20/57 (35.10%)
Study objectives**
Differential diagnosis	25/57 (43.9%)
Response evaluation	4/57 (7.0%)
Prognosis prediction	29/57 (50.9%)
Segmentation method
Manual	53/57 (93.0%)
Semi-automatic	4/57 (7.0%)
Publication of negative results
No	55/57 (96.5%)
Yes	2/57 (3.5%)
Type of ROI
2D	11/57 (19.3%)
3D	43/57 (75.4%)
2D&3D	3/57 (5.3%)
Model type
Type 1a: developed model validated with exactly the same data	10/57 (17.5%)
Type 1b: developed model validated with resampling data	6/57 (10.5%)
Type 2a: developed model validated with randomly splitting data	27/57 (47.4%)
Type 2b: developed model validated with non-randomly splitting data	4/57 (7.0%)
Type 3: developed model validated with separate data	9/57 (15.8%)
Type 4: validation only	1/57 (1.8%)
Phase classification***
Discovery science: experimental	10/57 (17.5%)
Phase 0: < 100 patients; retrospective; internal validation Or > 100 patients; retrospective; internal validation	34/57 (59.6%)
Phase I: < 100 patients; retrospective; external validation	2/57 (3.5%)
Phase II: > 100 patients; retrospective; external validation	9/57 (15.8%)
Phase III: > 100 patients; prospective; external validation	2/57 (3.5%)
Phase IV: real world	0/57

*Thirteen studies claimed joint first author, the author whose name was written first was assessed

**One study discussed two topics

***Two studies were classified as phase III due to prospective design, although they were conducted with less than 100 patients and without external validation