Table 2 Comparison of detection performances.
From: Deep learning application of vertebral compression fracture detection using mask R-CNN
Model | TP | FP | TN | FN | Sensitivity (%) | Specificity (%) | Accuracy (%) | F1-score |
|---|---|---|---|---|---|---|---|---|
Mask R-CNN | 71 | 15 | 126 | 18 | 79.8 | 89.4 | 85.7 | 82.6 |
Cascade R-CNN | 59 | 14 | 127 | 30 | 66.3 | 90.0 | 80.8 | 70.7 |
YOLOACT | 60 | 32 | 109 | 29 | 67.4 | 77.3 | 73.4 | 70.6 |
YOLOv5-l | 64 | 31 | 110 | 25 | 71.9 | 78.0 | 75.7 | 79.5 |
YOLOv5-m | 70 | 26 | 115 | 19 | 78.7 | 81.6 | 80.4 | 83.5 |
