Table 2 Performance metrics for Models A and B
From: Automated detection of pulmonary embolism from CT-angiograms using deep learning
Metric | Model A | Model B | ||
|---|---|---|---|---|
Stacks | Slices | Stacks | Slices | |
Accuracy | 90.2 (85.1–93.8) | 92.3 (91.9–92.7) | 87.3 (81.7–91.4) | 90.1 (89.6–90.5) |
Sensitivity | 86.6 (77.8–92.4) | 90.1 (89.0–91.2) | 83.5 (74.3–90.0) | 90.8 (89.7–91.9) |
Specificity | 93.5 (86.5–97.1) | 92.7 (92.2–93.1) | 90.7 (83.1–95.2) | 89.9 (89.4–90.4) |
PPV | 92.3 (84.3–96.6) | 69.9 (68.3–71.4) | 89.0 (80.3–94.3) | 62.3 (61.2–64.2) |
NPV | 88.5 (80.8–93.5) | 98.1 (97.8–98.3) | 85.8 (77.8–91.4) | 98.1 (97.9–98.3) |