TABLE 5.
Performance comparison of KneeXNet with transformer-based models on the test set and an independent dataset.
| Model | Test set | Independent dataset | ||||
|---|---|---|---|---|---|---|
| Abnormality | ACL tear | Meniscal tear | Abnormality | ACL tear | Meniscal tear | |
| ViT | 0.962 0.005 | 0.943 0.006 | 0.937 0.006 | 0.955 0.006 | 0.935 0.007 | 0.929 0.007 |
| Swin Transformer | 0.970 0.004 | 0.952 0.005 | 0.946 0.005 | 0.963 0.005 | 0.944 0.006 | 0.938 0.006 |
| KneeXNet | 0.985 0.003 | 0.972 0.004 | 0.968 0.004 | 0.978 0.004 | 0.964 0.005 | 0.960 0.005 |
The bold values represent the best-performing results for each respective metric or category.