Table 1 The cross-validation metrics of the staging performance of all evaluated methods.

From: An autoencoder and vision transformer based interpretability analysis on the performance differences in automated staging of second and third molars

Method

Tooth 37

Tooth 38

Accuracy

MAE

Kappa

Accuracy

MAE

Kappa

ViT only

0.712 (0.025)

0.375 (0.026)

0.680 (0.028)

0.462 (0.020)

0.867 (0.054)

0.402 (0.022)

AE + ViT

0.815 (0.022)

0.252 (0.025)

0.794 (0.033)

0.543 (0.052)

0.711 (0.051)

0.492 (0.058)

DenseNet only

0.810 (0.030)

0.216 (0.065)

0.788 (0.034)

0.535 (0.066)

0.679 (0.201)

0.483 (0.073)

AE + DenseNet

0.748 (0.023)

0.314 (0.021)

0.720 (0.026)

0.485 (0.049)

0.792 (0.085)

0.427 (0.054)

  1. The best metrics for each tooth are reported as bold text. All metrics are presented as mean (std).
Back to article page