Table 1 Patient characteristics of all 1392 women included in our cohort divided in the development dataset (n = 706) and test dataset (n = 686).
From: Deep learning-based breast cancer grading and survival analysis on whole-slide histopathology images
Patient characteristics | Development dataset | Test dataset | p value |
|---|---|---|---|
n | 706 | 686 | |
Age at biopsy | 1.00 | ||
Median years (Interquartile range) | 36 (33–38) | 36 (33–38) | |
Tumor grade (n (%)) | 0.25 | ||
Grade 1 | 113 (16) | 89 (13) | |
Grade 2 | 244 (35) | 238 (35) | |
Grade 3 | 349 (49) | 359 (52) | |
Nuclear score (n (%)) | 0.66 | ||
1 | 20 (3) | 16 (2) | |
2 | 335 (47) | 315 (46) | |
3 | 349 (49) | 354 (52) | |
Tubular score (n (%)) | 0.62 | ||
1 | 46 (7) | 43 (6) | |
2 | 135 (19) | 118 (17) | |
3 | 524 (74) | 524 (76) | |
Mitoses score (n (%)) | 0.64 | ||
1 | 249 (35) | 230 (34) | |
2 | 170 (24) | 161 (23) | |
3 | 286 (41) | 295 (43) | |
Subtype (n (%)) | 0.06 | ||
HR+/HER2− | 394 (56) | 345 (50) | |
HR−/HER2− | 182 (26) | 218 (32) | |
HR+/HER2+ | 85 (12) | 74 (11) | |
HR−/HER2+ | 36 (5) | 42 (6) | |
Tumor size (n (%)) | 0.22 | ||
1A–B | 128 (18) | 114 (17) | |
1C | 365 (52) | 366 (54) | |
2–3 | 195 (28) | 198 (29) | |
Missing | 18 (3) | 8 (1) | |
Lymphovascular invasion (n (%)) | 0.81 | ||
Absent | 587 (83) | 566 (83) | |
Present | 119 (17) | 120 (17) | |
Local treatment (n (%)) | 0.33 | ||
Conserving surgery with radiotherapy | 448 (63) | 451 (66) | |
Mastectomy without radiotherapy | 213 (30) | 184 (27) | |
Other | 45 (6) | 51 (7) |
