Table 3.

Summary of the performance of the analysis methods.

Metric	Frame-based	Segmental analysis	Lesion-based	Lesion-based with CP
Accuracy	0.78 ± 0.04	0.84 ± 0.04	0.71 ± 0.01	0.75 ± 0.02
Sensitivity	0.81 ± 0.06	0.87 ± 0.05	0.75 ± 0.01	0.75 ± 0.02
Specificity	0.75 ± 0.06	0.82 ± 0.05	0.68 ± 0.02	0.75 ± 0.01
AUC	0.79 ± 0.04	0.85 ± 0.02	0.73 ± 0.02	0.76 ± 0.02

The best performances were obtained by applying the segmental method (highlighted) with the Gaussian process regression (GPR) algorithm in combination with the CLE feature group. We also examined the effect of adding the calcification phenotype as an independent variable on the performance of the lesion-based model. We retrained the models of the lesion-based approach after adding the calcification type as an independent variable. The mean AUC was improved from 0.73 to 0.76, as indicated in the italics.

Significant values are in bold.