Skip to main content

Table 1 Cross-validated results for automated CNN four-class Perugini grade classification

From: Convolutional neural networks for detection of transthyretin amyloidosis in 2D scintigraphy images

 

AUC

ACC

Grade 0

Grade 1

Grade 2

Grade 3

Precis

Recall

Precis

Recall

Precis

Recall

Precis

Recall

Linear

0.86

0.66

0.86

0.68

0.37

0.59

0.29

0.61

0.91

0.83

Residual

0,86

0,67

0,88

0,68

0,40

0,65

0,22

0,65

0,83

0,83

VGG16

0.87

0.74

0.83

0.83

0.44

0.45

0.46

0.57

1.00

0.79

ResNet50

0.78

0.66

0.81

0.78

0.33

0.31

0.11

0.04

0.20

0.76

InceptionV3

0.75

0.62

0.80

0.70

0.30

0.40

0.06

0.13

0.59

0.64

MobileNet

0.76

0.74

0.76

0.97

0.42

0.07

0.00

0.00

1.00

0.17

  1. ACC = accuracy; AUC = area under the curve; Precis = precision