Table 2 Clinical and pathological data summaries in both training, validation and testing cohort.

From: Deep learning for the ovarian lesion localization and discrimination between borderline and malignant ovarian tumors based on routine MR imaging

 

Training group

Validation group

Testing group

P value

(N = 123)

(N = 29)

(N = 49)

Age (years)

45.74 ± 15.51

46.86 ± 11.22

47.04 ± 13.93

0.961

  < 30

25 (20.3%)

5 (17.2%)

8 (16.3%)

 

 30–50

45 (36.6%)

11 (37.9%)

20 (40.8%)

 

  > 50

53 (43.1%)

13 (44.9%)

21 (42.9%)

 

Ki-67 expression (%)

25.19 ± 24.30

32.25 ± 31.85

30.09 ± 27.67

0.946

  < 50

85 (78.0%)

17 (70.8%)

31 (68.9%)

 

 50–75

19 (17.4%)

4 (16.7%)

9 (20.0%)

 

  > 75

5 (4.6%)

3 (12.5%)

5 (11.1%)

 

CA-125 level(IU/L)

514.41 ± 887.01

363.24 ± 375.51

370.87 ± 565.14

0.956

  < 35

15 (18.3%)

5 (29.4%)

10 (30.3%)

 

 35–200

28 (34.1%)

4 (23.5%)

8 (24.2%)

 

 200–500

16 (19.5%)

2 (11.8%)

7 (21.3%)

 

  > 500

23 (28.1%)

6 (35.3%)

8 (24.2%)

 

Category

   

0.980

 Borderline tumor

62 (50.4%)

15 (51.7%)

25 (51.0%)

 

 Malignancies

61 (49.6%)

14 (48.3%)

24 (49.0%)

 

 Endometroid cancer

3 (4.9%)

1 (7.1%)

2 (8.3%)

 

 Low-grade adenocarcinoma

9 (14.7%)

4 (28.7%)

5 (20.8%)

 

 Clear cell type

4 (6.6%)

1 (7.1%)

3 (12.5%)

 

 High-grade Serous carcinoma

45 (73.8%)

7 (50.0%)

14 (58.4%)

 

 Mixed carcinoma

0 (0%)

1 (7.1%)

0 (0%)

Â