Table 1 Data labeling method for each Kellgren-Lawrence grade in the study
Dataset (total number of images) | KL grade | Number of images (proportion, %) |
|---|---|---|
Training set (37,462) | 0 | 14,906 (39.8) |
1 | 7022 (18.7) | |
2 | 9149 (24.4) | |
3 | 4920 (13.1) | |
4 | 1465 (3.9) | |
Validation set (9186) | 0 | 3483 (37.9) |
1 | 1785 (19.4) | |
2 | 2257 (24.6) | |
3 | 1292 (14.1) | |
4 | 396 (4.3) | |
Test set (17,040) | 0 | 7148 (41.9) |
1 | 2689 (15.8) | |
2 | 3024 (17.7) | |
3 | 2922 (17.1) | |
4 | 1257 (7.4) |