Skip to main content

Table 2 The global performance of the CNNs on the NINA and NIA datasets are reported. In addition, image improvement techniques are not active

From: Refactoring and performance analysis of the main CNN architectures: using false negative rate minimization to solve the clinical images melanoma detection problem

IIQ not active

Net

Data augmentation

ACC (min)

ACC (max)

ACC (mean)

ACC (sd)

AlexNet

None

0.68

1

0.89

0.05

Yes

0.76

0.97

0.87

0.05

DenseNet

None

0.62

0.79

0.74

0.04

Yes

0.41

0.88

0.73

0.08

Google InceptionV3

None

0.56

0.94

0.74

0.07

Yes

0.32

0.71

0.55

0.07

GoogleNet

None

0.65

0.94

0.80

0.06

Yes

0.30

0.76

0.55

0.09

MobileNet

None

0.50

0.91

0.75

0.09

Yes

0.35

0.76

0.56

0.08

ShuffleNet

None

0.44

0.88

0.69

0.08

Yes

0.26

0.74

0.52

0.10

SqueezeNet

None

0.38

1

0.55

0.11

Yes

0.15

0.79

0.58

0.10

VGG

None

0.59

0.82

0.75

0.04

Yes

0.59

0.82

0.73

0.05

Back to article page

BMC Bioinformatics

ISSN: 1471-2105

Contact us