Skip to main content

Table 3 Decisions in readers before and after referencing CAD output

From: Artificial intelligence-supported lung cancer detection by multi-institutional readers with multi-vendor chest radiographs: a retrospective clinical validation study

CAD results

TP

TN

FN

FP

Reader results

    

Before CAD use

FN

FP

TP

TN

After CAD use

TP

FN

TN

FP

FN

TP

FP

TN

General physicians

 Reader 1

11

3

2

5

0

1

3

8

 Reader 2

8

6

5

9

0

2

1

10

 Reader 3

9

9

0

2

0

1

0

11

 Reader 4

6

11

1

9

0

3

0

11

 Reader 5

7

4

4

0

0

1

2

9

 Reader 6

16

8

13

3

0

0

6

4

 Reader 7

4

2

0

18

0

2

0

9

 Reader 8

3

0

3

1

0

2

1

9

 Reader 9

4

5

1

5

0

4

1

10

 All general physicians

68

48

29

52

0

16

14

81

Radiologists

 Reader 10

7

4

2

13

0

1

4

7

 Reader 11

5

4

0

6

0

1

0

11

 Reader 12

7

5

2

0

0

2

2

9

 Reader 13

11

10

11

2

1

1

2

9

 Reader 14

6

4

0

13

0

2

7

3

 Reader 15

2

6

5

7

0

4

2

8

 Reader 16

1

8

0

2

0

4

0

11

 Reader 17

3

8

0

10

0

3

0

11

 Reader 18

7

3

4

5

0

2

4

5

 All radiologists

49

52

24

58

1

20

21

74

  1. Data are n. CAD: computer-assisted detection, FN: false negative, TP: true positive, FP: false positive, TN: true negative