Table 1.
Whole cohort (n = 1288) | Development set (n = 915) | Validation set (n = 373) |
p value** | |
---|---|---|---|---|
Patient age —yr. (SD) | 46.5 (16.0) | 47.0 (16.0) | 45.3 (16.0) | 0.089# |
- < 50 years —no. (%) | 785 (60.9) | 549 (60.0) | 236 (63.3) | 0.304* |
- ≥ 50 years —no. (%) | 503 (39.1) | 366 (40.0) | 137 (36.7) | 0.304* |
Clinically suspicious palpability (n = 1288) |
< 0.001* | |||
No —no. (%) | 631 (49.0) | 495 (54.1) | 136 (36.5) | |
Yes —no. (%) | 657 (51.0) | 420 (45.9) | 237 (63.5) | |
Breast mass dimensions on B-mode breast ultrasound (n = 1288) | ||||
Mass size in longest axis —mm. (SD) | 31.2 (14.6) | 30.9 (14.7) | 31.8 (14.5) | 0.317# |
Mass size in perpendicular plane —mm. (SD) | 19.9 (10.4) | 18.1 (9.2) | 24.2 (11.9) | < 0.001# |
Mass size in orthogonal plane —mm. (SD) | 23.8 (12.5) | 24.6 (12.9) | 22.0 (11.2) | < 0.001# |
Tissue composition (n = 1288) | ||||
Homogeneous background texture; fat —no. (%) | 259 (20.1) | 199 (21.7) | 60 (16.1) | 0.026* |
Heterogeneous background texture —no. (%) | 563 (43.7) | 395 (43.2) | 168 (45.0) | 0.581* |
Homogeneous background texture; fibroglandular —no. (%) | 466 (36.2) | 321 (35.1) | 145 (38.9) | 0.222* |
Shape of mass (n = 1288) | ||||
Oval —no. (%) | 906 (70.3) | 654 (71.5) | 252 (67.6) | 0.184* |
Round —no. (%) | 68 (5.3) | 43 (4.7) | 25 (6.7) | 0.187* |
Irregular —no. (%) | 314 (24.4) | 218 (23.8) | 96 (25.7) | 0.513* |
Orientation of mass (n = 1288) |
0.948 | |||
Parallel —No. (%) | 1046 (81.2) | 744 (81.3) | 302 (81.0) | |
Not parallel —No. (%) | 242 (18.8) | 171 (18.7) | 71 (19.0) | |
Margin of mass (n = 1288) |
0.011 | |||
Circumscribed —No. (%) | 619 (48.1) | 461 (49.6) | 158 (42.4) | |
Non-Circumscribed —No. (%) | 669 (51.9) | 454 (49.6) | 215 (57.6) | |
- Microlobulated margin —no. (%) | 69 (5.4) | 48 (5.2) | 21 (5.6) | 0.888* |
- Indistinct margin —no. (%) | 573 (44.5) | 391 (42.7) | 182 (48.8) | 0.054* |
- Angular margin —no. (%) | 99 (7.7) | 71 (7.8) | 28 (7.5) | 0.969* |
- Spiculated margin —no. (%) | 19 (1.5) | 13 (1.4) | 6 (1.6) | 1.00* |
Echo pattern (n = 1288) | ||||
Anechoic —no. (%) | 2 (0.2) | 0 (0.0) | 2 (0.5) | 0.151* |
Complex cystic and solid —no. (%) | 21 (1.6) | 16 (1.7) | 5 (1.3) | 0.778* |
Hypoechoic —no. (%) | 1065 (82.7) | 775 (84.7) | 290 (77.7) | 0.004* |
Isoechoic —no. (%) | 101 (7.8) | 62 (6.8) | 39 (10.5) | 0.035* |
Heterogeneous —no. (%) | 90 (7.0) | 57 (6.2) | 33 (8.8) | 0.121* |
Hyperechoic —no. (%) | 9 (0.7) | 5 (0.5) | 4 (1.1) | 0.510* |
Posterior features (n = 1288) | ||||
None —no. (%) | 759 (58.9) | 522 (57.0) | 237 (63.5) | 0.037* |
Enhancement —no. (%) | 370 (28.7) | 275 (30.1) | 95 (25.5) | 0.114* |
Combined pattern —no. (%) | 15 (1.2) | 9 (1.0) | 6 (1.6) | 0.508* |
Shadowing —no. (%) | 144 (11.2) | 109 (11.9) | 35 (9.4) | 0.227* |
Calcification (n = 1288) |
0.552 | |||
No calcification —no. (%) | 1231 (95.6) | 877 (95.8) | 354 (94.9) | |
Calcification —no. (%) | 57 (4.4) | 38 (4.2) | 19 (5.1) | |
Histopathological results (n = 1288) |
0.010 | |||
Benign —no. (%) | 920 (71.4) | 673 (73.6) | 247 (66.2) | |
- Fibroadenoma —no. (%) | 528 (57.5) | 381 (72.2) | 147 (27.8) | 0.504* |
- Lipoma —no. (%) | 4 (0.4) | 3 (75.0) | 1 (25.0) | 1* |
- Atypia —no. (%) | 1 (0.1) | 1 (100) | 0 (0.0) | 1* |
- Condense cyst —no. (%) | 65 (7.1) | 55 (84.6) | 10 (15.4) | 0.043* |
- Other —no. (%) | 320 (34.9) | 231 (72.2) | 89 (27.8) | 0.708* |
Malignant —no. (%) | 368 (28.6) | 242 (26.4) | 126 (33.8) | |
- No special type —no. (%) | 250 (67.9) | 154 (61.6) | 96 (38.4) | 0.020* |
- Invasive lobular carcinoma —no. (%) | 25 (6.8) | 13 (52.0) | 12 (48.0) | 0.199* |
- Invasive tubular carcinoma —no. (%) | 7 (1.9) | 7 (100) | 0 (0.0) | 0.127* |
- Medullary carcinoma —no. (%) | 5 (1.4) | 2 (40.0) | 3 (60.0) | 0.455* |
- Papillary carcinoma —no. (%) | 12 (3.3) | 11 (91.7) | 1 (8.3) | 0.107* |
- Ductal carcinoma in situ —no. (%) | 26 (7.1) | 22 (84.6) | 4 (15.4) | 0.059* |
- Other —no. (%) | 43 (11.7) | 33 (76.7) | 10 (23.3) | 0.149* |
* p values refer to chi-square tests for binary feature evaluation (feature true vs. feature not true)
# p values refer to t-test to evaluate mean differences of continuous data
** p values refer to differences in the development and validation sets