Skip to main content

Table 1 Baseline clinical-radiological features of the datasets

From: Deep learning-assisted diagnosis of benign and malignant parotid tumors based on ultrasound: a retrospective study

Variable

 

Training cohort(n = 721, image = 1638)

Validation cohort (n = 82, image = 194)

Internal-test cohort (n = 89, image = 192)

External-test cohort(n = 88, image = 187)

Benign (n = 506, Image = 1155)

Malignant (n = 215, Image = 483)

p-value

Benign (n = 57, Image = 136)

Malignant (n = 25, Image = 58)

p-value

Benign (n = 64, Image = 139)

Malignant (n = 25, Image = 53)

p-value

Benign (n = 79, Image = 165)

Malignant (n = 9, Image = 22)

p-value

Age (years)

 

56.00(44.00,64.00)

59.00(49.00,69.00)

0.001

57.00(41.00,69.00)

52.00(37.00,65.50)

0.372

56.50(43.75,63.00)

59.00(52.00,71.50)

0.203

65.00(54.00,75.00)

72.00(59.00,77.50)

0.380

Max-diameter (mm)

 

23.00(17.00,32.00)

24.00(18.00,32.00)

0.401

24.00(16.00,34.00)

22.50(15.00,27.20)

0.126

22.95(16.75,28.25)

26.00(17.00,38.70)

0.172

24.00(18.00,31.00)

24.00(16.00,76.00)

0.634

Sex

Female

186 (40.5%)

78 (38.6%)

0.644

24 (43.6%)

9 (36.0%)

0.520

20 (37.0%)

5 (20.8%)

0.157

41 (51.9%)

4 (44.4%)

0.736

 

Male

273 (59.5%)

124 (61.4%)

 

31 (56.4%)

16 (64.0%)

 

34 (63.0%)

19 (79.2%)

 

38 (48.1%)

5 (55.6%)

 

Location

Superficial

289 (63.0%)

113 (55.9%)

0.183

30 (54.5%)

17 (68.0%)

0.257

36 (66.7%)

16 (66.7%)

1.000

67 (84.8%)

8 (88.9%)

1.000

 

Deep

166 (36.2%)

88 (43.6%)

 

25 (45.5%)

8 (32.0%)

 

18 (33.3%)

8 (33.3%)

 

12 (15.2%)

1 (11.1%)

 
 

Both

4 (0.9%)

1 (0.5%)

 

0 (0%)

0 (0%)

 

0 (0.0%)

0 (0.0%)

 

0 (0.0%)

0 (0.0%)

 

Shape

Regular

365 (72.1%)

87 (40.5%)

<0.001

41 (71.9%)

12 (48.0%)

0.037

48 (75.0%)

8 (32.0%)

0.004

77 (97.5%)

2 (33.3%)

<0.001

 

Irregular

141 (27.9%)

128 (59.5%)

 

16 (28.1%)

13 (52.0%)

 

16 (25.0%)

17 (68.0%)

 

2 (2.5%)

6 (66.7%)

 

Margin

Clear

491 (97.0%)

102 (47.4%)

<0.001

53 (93.0%)

9 (36.0%)

<0.001

64 (100.0%)

4 (16.0%)

<0.001

79 (100.0%)

6 (66.7%)

0.001

 

Unclear

15 (3.0%)

113 (52.6%)

 

4 (7.0%)

16 (64.0%)

 

0 (0.0%)

21 (84.0%)

 

0 (0.0%)

3 (33.3%)

 

Echo

Homogeneous

103 (20.4%)

42 (19.5%)

0.801

11 (19.3%)

5 (20.0%)

1.000

10 (15.6%)

1 (4.0%)

0.171

38 (48.1%)

2 (22.2%)

0.174

 

Heterogeneous

403 (79.6%)

173 (80.5%)

 

46 (80.7%)

20 (80.0%)

 

54 (84.4%)

24 (96.0%)

 

41 (51.9%)

7 (77.8%)

 

Cystic areas

Absent

424 (83.8%)

177 (82.3%)

0.628

51 (89.5%)

20 (80.0%)

0.297

61 (95.3%)

23 (92.0%)

0.617

60 (79.5%)

5 (55.6%)

0.058

 

Present

82 (16.2%)

38 (17.7%)

 

6 (10.5%)

5 (20.0%)

 

3 (4.7%)

2 (8.0%)

 

19 (24.1%)

4 (44.4%)

 

Posterior acoustic enhancement

Absent

55 (10.9%)

114 (53.0%)

<0.001

10 (17.5%)

11 (44.0%)

0.012

6 (9.4%)

20 (80.0%)

<0.001

3 (3.8%)

5 (55.6%)

<0.001

Present

451 (89.1%)

101 (47.0%)

 

47 (82.5%)

14 (56.0%)

 

58 (90.6%)

5 (20.0%)

 

76 (96.2%)

4 (44.4%)

 

Calcification

Absent

486 (96.0%)

189 (87.9%)

<0.001

54 (94.7%)

22 (88.0%)

0.363

61 (95.3%)

19 (76.0%)

0.013

77 (97.5%)

5 (55.6%)

0.001

 

Present

20 (4.0%)

26 (12.1%)

 

3 (5.3%)

3 (12.0%)

 

3 (4.7%)

6 (24.0%)

 

2 (2.5%)

4 (44.4%)

Â