Table 2 Averaged test classification performance of the QMR model across all cross-validation folds for all hyperparameter combinations.

From: Convolutional neural networks for accurate real-time diagnosis of oral epithelial dysplasia and oral squamous cell carcinoma using high-resolution in vivo confocal microscopy

Epochs

LR

Accuracy

Sens.

Spec.

Precision

F1 score

Overall rank

5

0.001

64.0%

0.07

1.00

0.95

0.13

12

5

0.01

86.0%

0.75

0.93

0.87

0.80

5

5

0.1

82.3%

0.75

0.87

0.79

0.76

18

10

0.001

73.1%

0.31

0.99

0.97

0.47

11

10

0.01

85.5%

0.69

0.96

0.91

0.78

2

10

0.1

83.1%

0.74

0.89

0.81

0.77

14

15

0.001

78.5%

0.51

0.96

0.88

0.64

17

15

0.01

88.1%

0.78

0.94

0.90

0.83

1

15

0.1

83.7%

0.72

0.91

0.84

0.77

12

20

0.001

83.0%

0.62

0.96

0.91

0.74

9

20

0.01

85.1%

0.70

0.95

0.89

0.78

7

20

0.1

83.5%

0.66

0.94

0.88

0.75

15

25

0.001

84.4%

0.68

0.95

0.89

0.77

8

25

0.01

85.2%

0.70

0.94

0.89

0.79

6

25

0.1

83.7%

0.68

0.93

0.87

0.76

16

30

0.001

85.6%

0.72

0.94

0.89

0.79

3

30

0.01

85.3%

0.71

0.95

0.89

0.79

3

30

0.1

83.8%

0.76

0.88

0.81

0.78

9

  1. LR = Learning rate, Sens. = Sensitivity, Spec. = Specificity