Table 1 General characteristics of the collected dataset
From: Artificial intelligence for phase recognition in complex laparoscopic cholecystectomy
Variable | Training | Testing | |
---|---|---|---|
No. of sites | 5 | 5 | |
No. of videos | 294 | 77 | |
No. frames (1 FPS) | 586,453 | 107,819 | |
Complexity level | 1 (%) | 58 (20%) | 21 (27%) |
2 (%) | 125 (43%) | 38 (49%) | |
3 (%) | 43 (15%) | 6 (8%) | |
4 (%) | 50 (17%) | 8 (10%) | |
5 (%) | 18 (6%) | 4 (5%) | |
Adverse events | Gallbladder perforation* (%) | 86 (29%) | 15 (19%) |
Major bile leakage* (%) | 27 (9%) | 1 (1%) | |
Incidental finding* (%) | 19 (6%) | 3 (4%) | |
Cholecystitis* (%) | 55 (19%) | 7 (9%) | |
Without complication (%) | 146 (50%) | 56 (73%) |