Skip to main content

Table 6 Performance comparison of the modified 3D RPN trained on three datasets stratified by the range of solid components of nodules

From: Development of a modified 3D region proposal network for lung nodule detection in computed tomography scans: a secondary analysis of lung nodule datasets

Number of false positive per scan

Sensitivity

CPM

0.125

0.25

0.5

1

2

4

8

Data source

Solid component range

        

LUNA16

0 to 10%

72.1%

0.779

81.7%

84.6%

85.6%

88.5%

88.5%

82.7%

10 to 50%

81.3%

0.859

93.0%

94.5%

96.1%

96.1%

96.1%

91.9%

50 to 100%

80.4%

0.893

92.9%

92.9%

95.5%

97.3%

97.3%

92.2%

LNOP

0 to 10%

35.0%

41.7%

53.3%

66.7%

83.3%

91.7%

95.0%

66.7%

10 to 50%

45.8%

56.2%

68.8%

78.4%

88.1%

95.9%

97.9%

75.9%

50 to 100%

45.3%

60.6%

67.2%

80.4%

91.2%

96.4%

98.9%

77.1%

LNHE

0 to 10%

42.9%

45.7%

60.0%

74.3%

77.1%

85.7%

91.4%

68.2%

10 to 50%

53.8%

55.8%

61.5%

75.0%

78.8%

88.5%

94.2%

72.5%

50 to 100%

51.6%

57.9%

63.6%

79.3%

88.3%

90.9%

94.5%

75.2%