Hidden

CT — Hidden Tier

(20 scenes)

Fully blind: 20 real LoDoPaB-CT slices (validation split, patients 64–127) with adversarial modifications (metal inserts, lesions, calcifications).

What you get

No data download. Algorithm runs server-side on hidden measurements.

How to use

Package algorithm as Docker container / Python script accepting y + H, outputting x_hat + corrected spec.

What to submit

Containerized algorithm. Scored server-side against adversarial phantoms.

Parameter Specifications

🔒

True spec hidden — blind evaluation, only ranges available.

Parameter Spec Range Unit
center_offset_px -1.0 – 9.0 px
angle_error_deg -2.0 – 14.0 deg
beam_hardening_beta 0.07 – 0.37
detector_tilt_deg -0.5 – 5.5 deg

Hidden Tier Leaderboard

# Method Score PSNR SSIM Consistency Trust Source
1 CT-FM + gradient 0.795 33.58 0.958 0.88 ✓ Certified Wang et al., Nature MI 2026
2 Score-CT + gradient 0.791 33.62 0.958 0.86 ✓ Certified Gao et al., IEEE TMI 2024
3 TransCT + gradient 0.782 33.47 0.957 0.82 ✓ Certified Xia et al., MICCAI 2021
4 CTformer + gradient 0.771 33.11 0.954 0.79 ✓ Certified Wang et al., MICCAI 2023
5 CT-MAE + gradient 0.765 32.26 0.945 0.82 ✓ Certified Chen et al., MICCAI 2024
6 PINER-CT + gradient 0.757 32.17 0.944 0.79 ✓ Certified Sun et al., CVPR 2025
7 DiffusionMBIR + gradient 0.731 29.52 0.909 0.86 ✓ Certified Song et al., arXiv 2024
8 iCT-Net + gradient 0.723 29.5 0.909 0.82 ✓ Certified Li et al., IEEE TMI 2019
9 DuDoRNet + gradient 0.718 29.92 0.916 0.76 ✓ Certified Zhou et al., CVPR 2020
10 Eformer + gradient 0.711 29.42 0.908 0.77 ✓ Certified Wang et al., AAAI 2022
11 LEARN + gradient 0.711 28.35 0.888 0.86 ✓ Certified Chen et al., IEEE TPAMI 2018
12 PnP-ADMM + gradient 0.706 28.55 0.892 0.82 ✓ Certified Venkatakrishnan et al., GlobalSIP 2013
13 BM3D-CT + gradient 0.660 25.85 0.828 0.85 ✓ Certified Dabov et al., IEEE TIP 2007; Chen 2014
14 SART + gradient 0.652 25.88 0.829 0.81 ✓ Certified Andersen & Kak, Ultrason. Imaging 1984
15 CGLS + gradient 0.643 25.35 0.813 0.82 ✓ Certified Bjorck, SIAM 1996
16 DLCT + gradient 0.598 23.95 0.767 0.76 ✓ Certified Xu et al., IEEE TMI 2012
17 RED-CNN + gradient 0.597 23.18 0.738 0.85 ✓ Certified Chen et al., IEEE TMI 2017
18 WGAN-CT + gradient 0.593 23.42 0.747 0.8 ✓ Certified Wolterink et al., IEEE TMI 2017
19 CT-U-Net + gradient 0.575 23.02 0.732 0.76 ✓ Certified Han et al., Phys. Med. Biol. 2016
20 OSEM + gradient 0.573 23.02 0.732 0.75 ✓ Certified Hudson & Larkin, IEEE TMI 1994
21 FBPConvNet + gradient 0.563 22.1 0.694 0.82 ✓ Certified Jin et al., IEEE TMI 2017
22 FBP + gradient 0.489 19.92 0.595 0.76 ✓ Certified Kak & Slaney, IEEE Press 1988
23 TV-ADMM + gradient 0.473 19.37 0.568 0.76 ✓ Certified Sidky & Pan, Phys. Med. Biol. 2008
24 ART-TV + gradient 0.453 18.02 0.501 0.86 ✓ Certified Li et al., Med. Phys. 2004

Dataset

Scenes: 20

Scoring Formula

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

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