Table 1 Sensitivity (\(\%\)), the number of true positive nodules (\(\#\)TPs), the number of ground-truth nodules (\(\#\)GTs) measured using the in-house ABUS datasets A and B.

From: 3-D breast nodule detection on automated breast ultrasound using faster region-based convolutional neural networks and U-Net

Nodule category

ABUS dataset

Sensitivity (\(\%\))

\(\#\)TPs / \(\#\)GTs

Transverse

Coronal

Sagittal

Transverse

Coronal

Sagittal

Total

A

\(90.98\pm 7.29\)

\(84.41\pm 13.20\)

\(83.38\pm 10.38\)

13.9 / 15.2

12.8 / 15.2

12.7 / 15.2

B

93.65

87.30

87.30

59 / 63

55 / 63

55 / 63

BI-RADS 3

A

\(85.24\pm 21.44\)

\(80.79\pm 19.16\)

\(79.64\pm 16.91\)

4.1 / 4.7

3.8 / 4.7

3.8 / 4.7

B

91.30

78.26

69.57

21 / 23

18 / 23

16 / 23

BI-RADS 4

A

\(88.73\pm 20.46\)

\(85.09\pm 15.98\)

\(82.72\pm 14.85\)

7.6 / 8.3

6.8 / 8.3

6.8 / 8.3

B

94.12

91.18

97.06

32 / 34

31 / 34

33 / 34

BI-RADS 5

A

\(100.00\pm 0.00\)

\(100.00\pm 0.00\)

\(94.44\pm 16.67\)

2.2 / 2.2

2.2 / 2.2

2.1 / 2.2

B

100.00

100.00

100.00

6 / 6

6 / 6

6 / 6

Suspicious

A

\(90.91\pm 15.41\)

\(87.55\pm 14.19\)

\(85.13\pm 11.02\)

9.8 / 10.5

9.0 / 10.5

8.9 / 10.5

B

95.00

92.50

97.50

38 / 40

37 / 40

39 / 40

  1. Performance values are reported for a set of nodule groups, such as total nodules (BI-RADS categories 3, 4, and 5), each BI-RADS category, and suspicious nodules (BI-RADS categories 4 and 5). Mean (sensitivity, \(\#\)TPs, and \(\#\)GTs) and standard deviation (sensitivity) values are reported for the in-house ABUS dataset A, and a single value is reported for the in-house ABUS dataset B.
Back to article page