Vertebra Fracture Recognition on RSNA dataset (test)
94.51AccuracyCNN-Transformer
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| CNN-TransformerInput=-, View=-, Mode=Score Fusion2026.01 | 94.51 | 84.16 | 57.49 | 98.74 | 68.15 | 91.62 | |
| CNN-TransformerInput=2.5D (Max Projection Stacks), View=Axial2026.01 | 93.69 | 76.83 | 56.55 | 97.95 | 64.6 | 89.85 | |
| CNN-TransformerInput=2.5D (Slice Stacks), View=Axial2026.01 | 93.55 | 72.16 | 62.31 | 97.09 | 66.54 | 90.19 | |
| EfficientNetB0Input=3D, View=-2026.01 | 89.72 | 49.2 | 17.11 | 97.99 | 25.39 | 73.42 | |
| DenseNet121Input=2D, View=Axial2026.01 | 87.52 | 29.8 | 16.34 | 95.62 | 21.11 | 31.3 | |
| DenseNet121Input=2D, View=Fused2026.01 | 87.2 | 27.64 | 15.58 | 95.36 | 19.93 | 87.2 | |
| ResNet18Input=2D, View=Fused2026.01 | 60.63 | 10.33 | 37.12 | 63.31 | 16.16 | 60.63 | |
| DenseNet121Input=2D, View=Sagittal2026.01 | 50.05 | 12.93 | 67.8 | 48.03 | 21.72 | 50.05 | |
| DenseNet121Input=2D, View=Coronal2026.01 | 13.82 | 10.33 | 96.81 | 4.37 | 18.67 | 43.89 |