Table 2 MAE of the \(S_{\text {area}}\) score and IoU scores of U-Net variants across all folds.

From: Semantic modeling of cell damage prediction: a machine learning approach at human-level performance in dermatology

Fold

U-Net/16

U-Net/32

U-Net/64

MAE \([S_{area}]\) \(\downarrow\)

IoU \(\uparrow\)

MAE \([S_{area}]\) \(\downarrow\)

IoU \(\uparrow\)

MAE \([S_{area}]\) \(\downarrow\)

IoU \(\uparrow\)

1

0.031 ± 0.013

0.788 ± 0.020

0.021 ± 0.006

0.815 ± 0.016

0.026 ± 0.005

0.809 ± 0.028

2

0.051 ± 0.003

0.743 ± 0.015

0.048 ± 0.008

0.774 ± 0.007

0.054 ± 0.017

0.773 ± 0.008

3

0.065 ± 0.021

0.809 ± 0.011

0.073 ± 0.016

0.816 ± 0.008

0.063 ± 0.037

0.802 ± 0.015

4

0.055 ± 0.009

0.759 ± 0.015

0.052 ± 0.007

0.756 ± 0.018

0.046 ± 0.008

0.762 ± 0.014

5

0.038 ± 0.004

0.775 ± 0.008

0.057 ± 0.015

0.764 ± 0.025

0.041 ± 0.004

0.786 ± 0.015

6

0.037 ± 0.009

0.797 ± 0.021

0.040 ± 0.009

0.800 ± 0.028

0.037 ± 0.007

0.811 ± 0.021

7

0.045 ± 0.005

0.724 ± 0.014

0.048 ± 0.006

0.736 ± 0.018

0.046 ± 0.009

0.766 ± 0.016

8

0.018 ± 0.017

0.795 ± 0.033

0.019 ± 0.007

0.789 ± 0.017

0.020 ± 0.009

0.807 ± 0.018

9

0.058 ± 0.009

0.726 ± 0.002

0.046 ± 0.005

0.743 ± 0.013

0.049 ± 0.005

0.748 ± 0.020

10

0.017 ± 0.006

0.825 ± 0.016

0.019 ± 0.007

0.843 ± 0.024

0.014 ± 0.005

0.866 ± 0.043

All

0.045 ± 0.006

0.764 ± 0.004

0.047 ± 0.005

0.776 ± 0.004

0.044 ± 0.004

0.785 ± 0.003

  1. Significant values are in [bold].