Hidden

DEXA — 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
energy_offset -0.7 – 2.3 keV
soft_tissue -2.1 – 6.9 %
beam_overlap -0.014 – 0.046

Hidden Tier Leaderboard

# Method Score PSNR SSIM Consistency Trust Source
1 SwinDXA + gradient 0.742 30.46 0.924 0.84 ✓ Certified Liu et al., ICCV 2021 (DEXA adapt.)
2 PhysDXA + gradient 0.728 29.36 0.907 0.86 ✓ Certified Raissi et al., J. Comput. Phys. 2019 (DEXA)
3 PnP-DXA + gradient 0.699 28.35 0.888 0.8 ✓ Certified Venkatakrishnan et al., 2013 (DEXA adapt.)
4 DiffusionDXA + gradient 0.693 28.62 0.893 0.75 ✓ Certified Blattmann et al., arXiv 2023 (DEXA adapt.)
5 DXA-CNN + gradient 0.637 24.64 0.791 0.87 ✓ Certified Lee et al., Bone 2020
6 BML-Sep + gradient 0.618 24.04 0.77 0.85 ✓ Certified Lehmann et al., Med. Phys. 1981
7 DXA-U-Net + gradient 0.570 22.84 0.725 0.76 ✓ Certified Huo et al., IEEE TMED 2021
8 FBP-DEXA + gradient 0.568 22.06 0.693 0.85 ✓ Certified Mazess et al., Am. J. Clin. Nutr. 1990
9 TV-DEXA + gradient 0.565 21.81 0.682 0.87 ✓ Certified Sidky & Pan, Phys. Med. Biol. 2008 (DEXA)

Dataset

Scenes: 3

Scoring Formula

0.4 × PSNR_norm + 0.4 × SSIM + 0.2 × (1 − ‖y − Ĥx̂‖/‖y‖)

PSNR: 40% SSIM: 40% Consistency: 20%
Back to DEXA