Skip to main content

Table 7 Sample classification, multiple instance-learning algorithms results (expressed in percentage)

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

 

Accuracy

Sensitivity

Specificity

AUC

F score

Nr of features

NSK

75.53

80.39

69.76

80.94

75.44

150

STK

65.95

82.35

46.51

72.87

64.70

150

sMIL

73.40

78.43

67.44

79.66

73.30

50

MISVM

73.40

62.75

86.04

70.59

73.21

150

miSVM

72.34

64.70

81.39

70.63

72.28

150

MissSVM

72.34

58.82

88.37

74.33

71.93

30

  1. Area Under the Curve (AUC), Normalized set kernel (NSK), Statistics kernel (STK), Sparse multiple instance learning (sMIL), Maximum bag margin support vector machine (MISVM), Maximum pattern margin support vector machine (miSVM), Multi instance learning by semi-supervised support vector machine (MissSVM)