Skip to main content

Table 1 Clinical characteristics of the patients

From: Deep learning-based fully automated detection and segmentation of pelvic lymph nodes on diffusion-weighted images for prostate cancer: a multicenter study

Parameter

Model development dataset

External validation dataset

P

Training

Validation

Test

Overall

P

Hospital 2

Hospital 3

Hospital 4

Overall

P

No. of patients

919 (79.8)

116 (10.1)

116 (10.1)

1151

-

237 (59.1)

66 (16.5)

98 (24.4)

401

-

-

No. of PCa patients

820 (79.5)

104 (10.1)

107 (10.4)

1031

-

164 (58.2)

52 (18.4)

66 (23.4)

282

-

-

No. of non-PCa patients

99 (82.5)

12 (10.0)

9 (7.5)

120

-

73 (61.3)

14 (11.8)

32 (26.9)

119

-

-

No. of DWI images

1033 (78.9)

135 (10.3)

141 (10.8)

1309

-

237 (59.1)

66 (16.5)

98 (24.4)

401

-

-

Age (years)

70.0 ± 8.3

68.5 ± 9.6

69.1 ± 8.5

69.7 ± 8.5

0.971

71.33 ± 7.4

70.7 ± 7.2

72.1 ± 8.5

71.4 ± 7.6

0.394

0.000

tPSA (ng/ml)

16.7 [9.0, 56.4]

16.5 [8.4, 44.7]

11.3 [7.4, 43.2]

16.1 [8.8, 51.2]

0.262

19.1 [9.8, 63.9]

21.6 [8.4, 100.0]

16.9 [8.5, 61.4]

18.9 [9.3, 67.5]

0.734

0.161

No. of annotated LNs

25,335

3612

3560

32,507

-

4683

1118

1905

7707

-

-

No. of suspicious LNs

4821

725

751

6297

-

818

171

438

1427

-

-

Average LNs per patient

24.8 ± 9.3

25.8 ± 8.6

26.6 ± 9.3

25.1 ± 9.3

0.075

19.9 ± 8.2

16.7 ± 8.2

20.5 ± 8.0

19.5 ± 8.2

0.019

0.000

Short diameter of largest LNs (cm)

7.8 [6.0, 9.5]

8.7 [6.6, 10.3]

8.2 [6.4, 10.2]

7.9 [6.1, 9.7]

0.002

9.9 [8.4, 11.4]

9.9 [7.2, 13.2]

10.6 [9.2, 13.3]

10.0 [8.4, 12.1]

0.003

0.000

Volume of largest LNs (cm3)

5.8 [3.7, 11.3]

6.5 [4.0, 13.1]

6.0 [4.2, 15.2]

5.9 [3.8, 11.5]

0.064

0.8 [0.5, 1.3]

0.8 [0.5, 1.0]

0.7 [0.4, 1.2]

0.8 [0.5, 1.3]

0.120

0.000

  1. tPSA = total prostate-specific antigen, LN = lymph node
  2. The categorical variables are given as numbers (percentages). Quantitative variables were given as the median [Q1, Q3] for nonnormalized data