MRI Reconstruction on Axial T2w brain images (test)
93.05SSIML-TGVN
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| L-TGVNExperiment=B3, Acceleration factor=10x, Trainable parameters=≈ 500M, Batch size=64, Gradient steps=20K, GPU VRAM=8x A100 (80 GB)2026.06 | 93.05 | 33.81 | 18.8 | |
| E2E-VNExperiment=B3, Acceleration factor=10x, Trainable parameters=≈ 500M, Batch size=64, Gradient steps=20K, GPU VRAM=8x A100 (80 GB)2026.06 | 92.12 | 32.84 | 21 | |
| L-TGVNExperiment=B2, Acceleration factor=15x, Trainable parameters=≈ 500M, Batch size=64, Gradient steps=20K, GPU VRAM=8x A100 (80 GB)2026.06 | 91.56 | 32.2 | 22.7 | |
| L-TGVNExperiment=B1, Acceleration factor=20x, Trainable parameters=≈ 500M, Batch size=64, Gradient steps=20K, GPU VRAM=8x A100 (80 GB)2026.06 | 90.89 | 31.27 | 25.3 | |
| E2E-VNExperiment=B2, Acceleration factor=15x, Trainable parameters=≈ 500M, Batch size=64, Gradient steps=20K, GPU VRAM=8x A100 (80 GB)2026.06 | 89.46 | 30.62 | 27 | |
| E2E-VNExperiment=B1, Acceleration factor=20x, Trainable parameters=≈ 500M, Batch size=64, Gradient steps=20K, GPU VRAM=8x A100 (80 GB)2026.06 | 87.65 | 29.13 | 32.2 | |
| MTransExperiment=B3, Acceleration factor=10x, Trainable parameters=≈ 500M, Batch size=64, Gradient steps=20K, GPU VRAM=8x A100 (80 GB)2026.06 | 83.94 | 28.28 | 35.1 | |
| MTransExperiment=B2, Acceleration factor=15x, Trainable parameters=≈ 500M, Batch size=64, Gradient steps=20K, GPU VRAM=8x A100 (80 GB)2026.06 | 79.44 | 26.03 | 45.6 | |
| MTransExperiment=B1, Acceleration factor=20x, Trainable parameters=≈ 500M, Batch size=64, Gradient steps=20K, GPU VRAM=8x A100 (80 GB)2026.06 | 77.05 | 25.02 | 51.2 | |
| DMSIExperiment=B3, Acceleration factor=10x, Trainable parameters=≈ 500M, Batch size=64, Gradient steps=70K, GPU VRAM=8x A100 (80 GB)2026.06 | 60.69 | 24.3 | 55.3 | |
| DMSIExperiment=B2, Acceleration factor=15x, Trainable parameters=≈ 500M, Batch size=64, Gradient steps=70K, GPU VRAM=8x A100 (80 GB)2026.06 | 54.98 | 22.72 | 66.7 | |
| DMSIExperiment=B1, Acceleration factor=20x, Trainable parameters=≈ 500M, Batch size=64, Gradient steps=70K, GPU VRAM=8x A100 (80 GB)2026.06 | 53.78 | 22.09 | 71.7 |