Table 1 Performance metrics for various EfficientNet models.

From: Artificial intelligence based sonographic differentiation between skull fractures and normal sutures in young children

EfficientNet

Precision

Recall

Accuracy

F1

PR AUC

(95% CI)

EfficientNet-B0

0.640

0.706

0.645

0.668

0.696

(0.615–0.771)

EfficientNet-B1

0.892

0.722

0.799

0.798

0.823

(0.750–0.885)

EfficientNet-B2

0.900

0.787

0.799

0.836

0.900

(0.849–0.943)

EfficientNet-B3

0.872

0.711

0.770

0.783

0.866

(0.805–0.919)

EfficientNet-B4

0.849

0.759

0.773

0.799

0.860

(0.801–0.911)

EfficientNet-B5

0.878

0.770

0.786

0.820

0.871

(0.816–0.919)

EfficientNet-B6

0.921

0.778

0.806

0.841

0.913

(0.870–0.948)

EfficientNet-B7

0.841

0.748

0.770

0.790

0.852

(0.794–0.902)

Mean

0.849

0.756

0.770

0.796

0.848

(0.788–0.900)

SD

± 0.083

± 0.027

± 0.048

± 0.051

± 0.049

(± 0.073to ± 0.052)

  1. Precision, Recall, Accuracy, F1 score and PR AUC (Precision-Recall, including 95% CI) values are listed.
Back to article page