Probe translation dimension estimation on Echocardiography ten standard planes
5.94MAE Plane 1 (mm)EchoCLIP+ours
Evaluation results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| EchoCLIP+oursTemporal context=Sequential, Pre-training data source=ultrasound, Trained Params=2.61M2025.10 | 5.94 | 5.55 | 6.03 | 5.17 | 4.99 | 6.9 | 4.77 | 4.87 | 5.11 | 4.63 | 5.4 | |
| USFM+oursTemporal context=Sequential, Pre-training data source=ultrasound, Trained Params=3.97M2025.10 | 6.04 | 5.4 | 5.4 | 5.2 | 5.01 | 6.72 | 4.94 | 5.13 | 5.07 | 4.6 | 5.35 | |
| BiomedCLIP+oursTemporal context=Sequential, Pre-training data source=ultrasound, Trained Params=3.94M2025.10 | 6.12 | 5.78 | 6.28 | 5.25 | 5.18 | 6.65 | 5.03 | 4.97 | 5.3 | 4.94 | 5.55 | |
| EchoCLIPTemporal context=Sequential, Pre-training data source=ultrasound, Trained Params=88.41M2025.10 | 6.31 | 6.03 | 6.48 | 5.76 | 5.76 | 7.31 | 7.03 | 7.05 | 6.96 | 6.95 | 6.56 | |
| USFMTemporal context=Sequential, Pre-training data source=ultrasound, Trained Params=87.04M2025.10 | 6.48 | 6.34 | 6.84 | 6.05 | 5.93 | 7.72 | 8.07 | 8.16 | 8.05 | 7.86 | 7.15 | |
| Decision-TTemporal context=Sequential, Pre-training data source=ImageNet, Trained Params=22.27M, Pre-trained encoder=DeiT2025.10 | 6.69 | 7.6 | 8.46 | 6.71 | 6.81 | 7.87 | 7.01 | 7.06 | 7.92 | 7.67 | 7.38 | |
| US-GuideNetTemporal context=Sequential, Pre-training data source=ImageNet, Trained Params=22.05M, Pre-trained encoder=DeiT2025.10 | 6.81 | 7.59 | 8.63 | 6.7 | 7.05 | 8.09 | 7.28 | 7.28 | 8.05 | 7.84 | 7.53 | |
| BiomedCLIPTemporal context=Sequential, Pre-training data source=ultrasound, Trained Params=86.15M2025.10 | 6.87 | 6.67 | 7.26 | 6.32 | 6.31 | 8.09 | 7.77 | 7.87 | 7.95 | 7.83 | 7.29 | |
| US-IJEPATemporal context=Single-frame, Pre-training data source=ultrasound, Trained Params=22.59M2025.10 | 8.28 | 7.82 | 8.36 | 7.74 | 7.74 | 8.84 | 8.46 | 8.55 | 9.01 | 8.67 | 8.35 | |
| EchoCLIPTemporal context=Single-frame, Pre-training data source=ultrasound, Trained Params=89.74M2025.10 | 8.29 | 7.78 | 8.45 | 7.53 | 7.6 | 9.17 | 8.16 | 8.22 | 8.58 | 8.35 | 8.21 | |
| DINOv2Temporal context=Single-frame, Pre-training data source=ImageNet, Trained Params=90.28M2025.10 | 8.31 | 7.82 | 8.36 | 7.7 | 7.66 | 9.09 | 8.4 | 8.52 | 8.7 | 8.62 | 8.22 | |
| US-MAETemporal context=Single-frame, Pre-training data source=ultrasound, Trained Params=22.60M2025.10 | 8.31 | 7.74 | 8.3 | 7.69 | 7.7 | 9.02 | 8.35 | 8.47 | 8.55 | 8.45 | 8.26 | |
| USFMTemporal context=Single-frame, Pre-training data source=ultrasound, Trained Params=92.94M2025.10 | 8.34 | 7.74 | 8.28 | 7.62 | 7.69 | 9.02 | 8.2 | 8.33 | 8.51 | 8.39 | 8.26 | |
| BiomedCLIPTemporal context=Single-frame, Pre-training data source=ultrasound, Trained Params=89.50M2025.10 | 8.4 | 7.89 | 8.44 | 7.79 | 7.78 | 9.16 | 8.62 | 8.73 | 8.88 | 8.75 | 8.44 | |
| DeiTTemporal context=Single-frame, Pre-training data source=ImageNet, Trained Params=22.60M2025.10 | 8.5 | 8.09 | 8.51 | 7.9 | 7.85 | 9.21 | 8.43 | 8.55 | 8.7 | 8.52 | 8.43 | |
| LVM-MedTemporal context=Single-frame, Pre-training data source=ultrasound, Trained Params=89.44M2025.10 | 8.55 | 8 | 8.42 | 7.95 | 7.89 | 9.31 | 8.7 | 8.88 | 9.07 | 8.8 | 8.56 | |
| US-MoCoTemporal context=Single-frame, Pre-training data source=ultrasound, Trained Params=22.60M2025.10 | 8.75 | 8.08 | 8.29 | 7.95 | 7.9 | 9.22 | 8.55 | 8.78 | 8.88 | 8.74 | 8.51 |