Skip to main content

Table 4 Comparison of the averaged classification performances of different layers of CNN model across ten folds

From: DeepFHR: intelligent prediction of fetal Acidemia using fetal heart rate signals based on convolutional neural network

Layers

Type

Performance

Acc (%)

Se (%)

Sp (%)

QI (%)

AUC (%)

Training Time (second)

5

I – C – P – F – O

92.13

93.45

91.22

92.33

92.34

140.5

6

I – C – P – C – F - O

91.88

92.55

89.74

91.13

91.15

162.3

7

I – C – P – C – P – F - O

91.21

92.13

89.25

90.68

90.69

178.8

8

I – C – P – C – P – F – F – O

90.76

91.71

88.67

90.18

90.19

201.3

9

I – C – P – C – P – C – F – F - O

91.34

92.34

89.56

90.94

90.95

225.4

10

I – C – P – C – P – C – P – F – F - O

90.82

91.88

89.11

90.48

90.50

248.2

  1. Note: The best performance is indicated in bold. I = image input layer, C = convolution + ReLU + normalization layer, P = max pooling layer, F = fully-connected + dropout layer, O = classification output layer
Back to article page

BMC Medical Informatics and Decision Making

ISSN: 1472-6947

Contact us