Hidden
MRI — Hidden Tier
(3 scenes)Fully blind server-side evaluation — no data download.
What you get
No data downloadable. Algorithm runs server-side on hidden measurements.
How to use
Package algorithm as Docker container / Python script. Submit via link.
What to submit
Containerized algorithm accepting y + H, outputting x_hat + corrected spec.
Parameter Specifications
🔒
True spec hidden — blind evaluation, only ranges available.
| Parameter | Spec Range | Unit |
|---|---|---|
| B0_inhomog | -1.05 – 3.45 | ppm |
| gradient_nonlin | -1.4 – 4.6 | % |
| coil_sensitivity | -3.5 – 11.5 | % |
| k_trajectory | -0.7 – 2.3 | % |
Hidden Tier Leaderboard
| # | Method | Score | PSNR | SSIM | Consistency | Trust | Source |
|---|---|---|---|---|---|---|---|
| 1 | SwinMR++ + gradient | 0.801 | 34.49 | 0.964 | 0.85 | ✓ Certified | Huang et al., IEEE TMI 2025 — multi-scale axial attention + INR head + k-space DC per module + LPIPS+SSIM+k-space joint loss + dynamic feature fusion |
| 2 | PnP-DnCNN-Pro + gradient | 0.797 | 34.55 | 0.965 | 0.83 | ✓ Certified | PnP-DnCNN-Pro IEEE TMI 2025 (DOI:10.1109/TMI.2025.3441240) — multi-scale DnCNN denoiser + adaptive mu/sigma schedule + SIREN INR output head + joint LPIPS+SSIM denoiser training + dynamic PnP regularization scheduling |
| 3 | HUMUS-Net++ + gradient | 0.789 | 33.6 | 0.958 | 0.85 | ✓ Certified | Fabian et al., dHUMUS-Net 2023 — k-space DC per module + dynamic multi-scale weighting + INR head + perceptual-structural loss + axial attention |
| 4 | HUMUS-Net + gradient | 0.789 | 34.18 | 0.962 | 0.81 | ✓ Certified | Fabian et al., NeurIPS 2022 |
| 5 | MRI-FM + gradient | 0.773 | 32.13 | 0.944 | 0.87 | ✓ Certified | Wang et al., Nature MI 2026 |
| 6 | U-Net++ + gradient | 0.771 | 31.84 | 0.941 | 0.88 | ✓ Certified | Chen & Boning, IEEE TMI 2024 (DOI: 10.1109/TMI.2024.3367890) — Residual U-Net + data consistency layers + plug-and-play prior + residual connections + dense skip paths |
| 7 | MR-IPT + gradient | 0.769 | 33.61 | 0.958 | 0.75 | ✓ Certified | Sci. Reports 2025 |
| 8 | HybridCascade++ + gradient | 0.767 | 33.16 | 0.954 | 0.77 | ✓ Certified | HybridCascade++ MICCAI 2021 + IEEE TMI 2025 — multi-scale cascade DC + SIREN INR warm-start + SSIM structural anchor + DRUNet polish + freq-blend LF/HF fusion |
| 9 | ReconFormer++ + gradient | 0.763 | 32.3 | 0.946 | 0.81 | ✓ Certified | Pan et al., IEEE TMI 2025 |
| 10 | SwinMR + gradient | 0.761 | 32.72 | 0.95 | 0.77 | ✓ Certified | Huang et al., MICCAI 2022 |
| 11 | E2E-VarNet + gradient | 0.750 | 31.37 | 0.935 | 0.81 | ✓ Certified | Sriram et al., MICCAI 2020 |
| 12 | MMR-Mamba + gradient | 0.744 | 31.09 | 0.932 | 0.8 | ✓ Certified | Zhao et al., Med. Image Anal. 2025 |
| 13 | PnP-DnCNN + gradient | 0.743 | 31.06 | 0.932 | 0.8 | ✓ Certified | Ahmad et al., IEEE TCI 2019 (DOI:10.1109/TCI.2019.2944521) |
| 14 | MoDL-Net++ + gradient | 0.733 | 30.39 | 0.923 | 0.8 | ✓ Certified | MoDL-Net++ IEEE TMI 2025 — multi-scale pyramid fusion + RDN/Swin deep prior + differentiable DC layers + LPIPS+SSIM+L1 joint loss + two-stage training strategy |
| 15 | PromptMR-SFM + gradient | 0.725 | 30.29 | 0.921 | 0.77 | ✓ Certified | PWM 2026 |
| 16 | MRDynamo + gradient | 0.711 | 28.23 | 0.886 | 0.87 | ✓ Certified | Chen et al., NeurIPS 2024 |
| 17 | ReconFormer + gradient | 0.704 | 28.18 | 0.885 | 0.84 | ✓ Certified | Guo et al., IEEE TMI 2024 |
| 18 | U-Net + gradient | 0.687 | 28.27 | 0.886 | 0.75 | ✓ Certified | Zbontar et al., arXiv 2018 |
| 19 | MRI-DiffusionNet + gradient | 0.675 | 26.66 | 0.85 | 0.84 | ✓ Certified | Song et al., ICCV 2024 |
| 20 | BrainID-MRI + gradient | 0.671 | 27.39 | 0.867 | 0.75 | ✓ Certified | Liu et al., CVPR 2025 |
| 21 | MoDL + gradient | 0.671 | 26.29 | 0.84 | 0.86 | ✓ Certified | Aggarwal et al., IEEE TMI 2019 |
| 22 | GRAPPA + gradient | 0.671 | 27.08 | 0.86 | 0.78 | ✓ Certified | Griswold et al., MRM 2002 |
| 23 | PromptMR + gradient | 0.668 | 26.27 | 0.839 | 0.85 | ✓ Certified | Bai et al., ECCV 2024 |
| 24 | BM3D-MRI + gradient | 0.652 | 25.22 | 0.809 | 0.88 | ✓ Certified | Eksioglu, Comput. Math. Meth. Med. 2016 |
| 25 | DCCNN + gradient | 0.650 | 25.22 | 0.809 | 0.87 | ✓ Certified | Schlemper et al., IEEE TMI 2018 |
| 26 | ESPIRiT + gradient | 0.646 | 25.51 | 0.818 | 0.82 | ✓ Certified | Uecker et al., MRM 2014 |
| 27 | SENSE + gradient | 0.644 | 26.15 | 0.836 | 0.74 | ✓ Certified | Pruessmann et al., MRM 1999 |
| 28 | Deep-ADMM-Net + gradient | 0.640 | 24.71 | 0.793 | 0.88 | ✓ Certified | Yang et al., NeurIPS 2016 |
| 29 | HybridCascade + gradient | 0.626 | 25.3 | 0.812 | 0.74 | ✓ Certified | Fastmri, arXiv 2020 |
| 30 | Zero-Filled IFFT + gradient | 0.594 | 23.3 | 0.743 | 0.82 | ✓ Certified | Pruessmann et al., MRM 1999 |
| 31 | Score-MRI + gradient | 0.592 | 23.51 | 0.751 | 0.78 | ✓ Certified | Chung & Ye, Med. Image Anal. 2022 |
| 32 | LORAKS + gradient | 0.578 | 22.92 | 0.728 | 0.79 | ✓ Certified | Haldar, IEEE TMI 2014 |
| 33 | ALOHA + gradient | 0.564 | 22.45 | 0.709 | 0.78 | ✓ Certified | Jin et al., IEEE TMI 2016 |
| 34 | L1-Wavelet + gradient | 0.511 | 19.91 | 0.594 | 0.87 | ✓ Certified | Lustig et al., MRM 2007 |
| 35 | k-t SPARSE-SENSE + gradient | 0.489 | 19.9 | 0.594 | 0.76 | ✓ Certified | Lustig et al., MRM 2006 |
Dataset
Scenes: 3
Scoring Formula
0.4 × PSNR_norm + 0.4 × SSIM + 0.2 × (1 − ‖y − Ĥx̂‖/‖y‖)
PSNR: 40%
SSIM: 40%
Consistency: 20%