Table 1.
Statistics of the overall NYU Breast Ultrasound Dataset, internal test set, and reader study set. This dataset was collected from NYU Langone Health over an eight-year period. Exam-level BI-RADS were issued by radiologists based on patients’ breast US exams. Breast densities were determined using existing screening and diagnostic mammography reports. Patients who were not matched with any mammograms were assigned "unknown” for breast density. Abbreviations: N, number; SD, standard deviation.
| Characteristic, unit | Overall | Internal test set | Reader study |
|---|---|---|---|
| Patients, N | 143,203 | 25,003 | 644 |
| Age, mean years (SD) | 53.7 (13.7) | 55.5 (12.7) | 52.8 (14.0) |
| < 40 yrs old, N (%) | 18,218 (12.7) | 1857 (7.4) | 90 (14.0) |
| 40 − 49 years old, N (%) | 33,955 (23.7) | 5811 (23.2) | 175 (27.2) |
| 50 − 59 years old, N (%) | 34,942 (24.4) | 6567 (26.3) | 146 (22.7) |
| 60 − 69 years old, N (%) | 26,671 (18.6) | 5198 (20.8) | 104 (16.1) |
| ≥70 years old, N (%) | 17,703 (12.4) | 3359 (13.4) | 81 (12.6) |
| Exams, N | 288,767 | 44,755 | 663 |
| Images, N | 5,442,907 | 858,636 | 13,582 |
| Average no. of images per exam, N | 18 | 19 | 20 |
| Exams associated with biopsy, N (%) | 28,914 (10.0) | 8337 (18.6) | 587 (88.5) |
| Breasts, N | 510,271 | 79,156 | 1024 |
| Breasts with benign findings, N | 26,843 | 7879 | 567 |
| Breasts with malignant findings, N | 5593 | 1324 | 73 |
| Exam-level BI-RADS | |||
| BI-RADS 0, N (%) | 14,078 (4.9) | 1092 (2.4) | 80 (12.1) |
| BI-RADS 1, N (%) | 86,347 (29.9) | 12,374 (27.6) | 56 (8.4) |
| BI-RADS 2, N (%) | 136,322 (47.2) | 21,675 (48.4) | 80 (12.1) |
| BI-RADS 3, N (%) | 27,711 (9.6) | 3586 (8.0) | 25 (3.8) |
| BI-RADS 4, N (%) | 22,133 (7.7) | 5578 (12.5) | 391 (59.0) |
| BI-RADS 5, N (%) | 1348 (0.5) | 338 (0.8) | 22 (3.3) |
| BI-RADS 6, N (%) | 518 (0.2) | 69 (0.2) | 3 (0.5) |
| Unknown BI-RADS, N (%) | 310 (0.1) | 43 (0.1) | 6 (0.9) |
| Exam-level mammographic density | |||
| A (breasts are almost entirely fatty), N (%) | 5384 (1.9) | 695 (1.6) | 13 (2.0) |
| B (scattered areas of fibroglandular density), N (%) | 69,948 (24.2) | 11,048 (24.7) | 143 (21.6) |
| C (breasts are heterogeneously dense), N (%) | 165,855 (57.4) | 26,509 (59.2) | 376 (56.7) |
| D (the breasts are extremely dense), N (%) | 31,829 (11.0) | 5189 (11.6) | 76 (11.5) |
| Unknown density, N (%) | 15,751 (5.5) | 1314 (2.9) | 55 (8.3) |