Table 2 Performance of the deep learning models for frame-level diagnosis in the internal and external validation.

From: Diagnosis of coronary layered plaque by deep learning

 

AUC (95% CI)

Sensitivity (%) (95% CI)

Specificity (%) (95% CI)

Accuracy (%) (95% CI)

Internal validation

 ViT-based model

0.860 (0.855—0.866)

77.7 (76.4–79.0)

77.6 (77.4–77.8)

77.6 (77.4–77.8)

 Standard CNN-based model

0.799 (0.792–0.805)

71.7 (70.2–74.0)

73.8 (73.6–74.0)

73.8 (73.6–73.9)

External validation

 ViT-based model

0.845 (0.837–0.853)

76.5 (74.6–78.4)

76.0 (75.7–76.3)

76.0 (75.7–76.3)

 Standard CNN-based model

0.791 (0.782–0.800)

71.4 (69.3–73.4)

71.9 (71.6–72.3)

71.9 (71.6–72.2)

  1. AUC area under the curve, CI confidence interval, PPV positive predictive value, NPV negative predictive value, DL deep learning, ViT vision transformer, CNN convolutional neural network.
Back to article page