Table 2.

Classification metrics of HTNet with ultrasound images as input.

Classification metrics	Internal-testing set 1 (n = 4303)	Internal-testing set 2 (n = 185)	External-testing set (n = 563)
Accuracy (95% CI)	0.832 (0.821–0.843)	0.832 (0.771–0.883)	0.821 (0.786–0.851)
Sensitivity (95% CI)	0.826 (0.803–0.847)	0.846 (0.719–0.931)	0.842 (0.773–0.897)
Specificity (95% CI)	0.835 (0.821–0.848)	0.827 (0.752–0.887)	0.813 (0.772–0.849)