Skip to main content

Table 5 Sample classification, thresholding approach results (expressed in percentage), no feature selection

From: Improved automated early detection of breast cancer based on high resolution 3D micro-CT microcalcification images

 

No Feature Selection

Threshold

Classifier

Accuracy

Sensitivity

Specificity

F score

50%

MLP

78.72 ±41.15

70.59

88.37

78.67

45%

SVM

78.72 ±41.15

74.51

83.72

78.75

40%

MLP

80.85 ±39.56

80.39

81.39

80.87

35%

AdaBoost

79.79 ±40.37

72.55

88.37

79.77

35%

MLP

78.72 ±41.15

80.39

76.74

78.72

30%

SVM

78.72 ±41.15

76.47

81.36

78.76

30%

AdaBoost

77.66 ±41.88

74.51

81.40

77.70

25%

RF

80.85 ±39.56

76.47

86.04

80.88

25%

AdaBoost

78.72 ±41.15

80.39

76.74

78.72

20%

RF

78.72 ±41.15

76.47

81.39

78.76

15%

RF

78.72 ±41.15

82.35

74.41

78.67

10%

RF

75.53 ±43.22

86.27

62.79

75.09

10%

AdaBoost

71.28 ±45.49

92.16

46.51

69.46

5%

RF

74.47 ±43.84

96.07

48.83

72.69

  1. Multi Layer Perceptron (MLP), Random Forest (RF), Support Vector Machine (SVM)