Skip to main content

Table 3 The diagnostic performance of CCTA and DL-LEM in identifying ≥ 75% stenosis

From: Use of a deep-learning-based lumen extraction method to detect significant stenosis on coronary computed tomography angiography in patients with severe coronary calcification

  Evaluation 1 Evaluation 2
CCTA group DL-LEM group CCTA group DL-LEM group
Sensitivity 72.1% (64.8–78.7%) 72.7% (65.4–79.2%) 100% (93.4–100%) 100% (93.4–100%)
Specificity 75.1% (71.8–78.2%) 77.3% (74.1–80.3%) 0% (0–3.7%) 10.9%** (6.4–17.1%)
Positive predictive value 40.9% (35.3–46.7%) 43.4% (37.6–49.3%) 35.5% (29.3–42.1%) 38.2% (31.6–45.1%)
Negative predictive value 91.8% (89.3–93.9%) 92.2% (89.8–94.2%) N/A (0–100%) 100% (71.3–100%)
Accuracy 74.5% (71.5–77.4%) 76.4%** (73.5–79.2%) 35.5% (29.3–42.1%) 42.5%** (36.0–49.2%)
  1. N/A not available
  2. **p < 0.01. Data are presented with 95% confidence intervals in parentheses