Skip to main content

Table 2 The characteristics of patients and lymph nodes

From: Development and validation of the 3D U-Net algorithm for segmentation of pelvic lymph nodes on diffusion-weighted images

Characteristics

Model development dataset

Hold-out test dataset

P value

Training set

Validation set

Testing set

P value

Patients with suspicious LNs

Patients without suspicious LNs

P value

No. of patients

309

43

41

–

37

40

–

–

Age, mean ± SD (years)

70.3 ± 9.6

71.2 ± 7.3

70.8 ± 8.8

0.785

69.4 ± 8.8

71.4 ± 8.4

0.305

0.990

PSA, median (ng/ml)

13.00 (7.13,23.43)

14.21

(8.20, 26.73)

12.20

(7.42, 21.30)

0.552

15.49

(9.25, 26.65)

10.69

(7.10, 18.82)

0.031

0.088

No. of annotated LNs

8139

1258

1110

–

–

–

–

–

Average LNs per patient

27 (5, 30)

29 (6, 33)

27 (3, 35)

0.537

–

–

–

–

No. of suspicious LNs

1374

186

230

–

201

–

  

Short diameter of largest LNs (cm)

0.83 ± 0.30

0.93 ± 0.33

0.83 ± 0.29

0.145

–

–

–

–

Volume of largest LNs (cm3)

9.03

(4.60, 17.32)

10.93

(5.30, 17.40)

7.09

(5.37, 14.08)

0.628

–

–

–

–

Scanners

        

3.0 T Achieva

96

16

12

–

12

15

–

–

3.0 T Discovery

113

13

15

–

14

13

–

–

3.0 T Interia

100

14

14

–

11

12

–

–

  1. PSA prostate-specific antigen, LN lymph node