Portal Imaging (EPID)

Join the Competition Contribute to Benchmark

Standard reconstruction benchmark — forward model perfectly known, no calibration needed. Score = 0.5 × clip((PSNR−15)/30, 0, 1) + 0.5 × SSIM

#	Method	Score	PSNR (dB)	SSIM	Source
🥇	Score-CT Score-CT Song et al., NeurIPS 2024 39.92 dB SSIM 0.984 Checkpoint unavailable	0.907	39.92	0.984	✓ Certified	Song et al., NeurIPS 2024
🥈	DiffusionCT DiffusionCT Kazemi et al., ECCV 2024 39.68 dB SSIM 0.982 Checkpoint unavailable	0.902	39.68	0.982	✓ Certified	Kazemi et al., ECCV 2024
🥉	CTFormer CTFormer Li et al., ICCV 2024 39.45 dB SSIM 0.980 Checkpoint unavailable	0.897	39.45	0.980	✓ Certified	Li et al., ICCV 2024
4	CT-ViT CT-ViT Guo et al., NeurIPS 2024 39.15 dB SSIM 0.978 Checkpoint unavailable	0.891	39.15	0.978	✓ Certified	Guo et al., NeurIPS 2024
5	DOLCE DOLCE Liu et al., ICCV 2023 38.32 dB SSIM 0.971 Checkpoint unavailable	0.874	38.32	0.971	✓ Certified	Liu et al., ICCV 2023
6	DuDoTrans DuDoTrans Wang et al., MLMIR 2022 37.68 dB SSIM 0.962 Checkpoint unavailable	0.859	37.68	0.962	✓ Certified	Wang et al., MLMIR 2022
7	Learned Primal-Dual Learned Primal-Dual Adler & Oktem, IEEE TMI 2018 36.42 dB SSIM 0.947 Checkpoint unavailable	0.831	36.42	0.947	✓ Certified	Adler & Oktem, IEEE TMI 2018
8	FBPConvNet FBPConvNet Jin et al., IEEE TIP 2017 35.81 dB SSIM 0.939 Checkpoint unavailable	0.816	35.81	0.939	✓ Certified	Jin et al., IEEE TIP 2017
9	RED-CNN RED-CNN Chen et al., IEEE TMI 2017 33.56 dB SSIM 0.908 Checkpoint unavailable	0.763	33.56	0.908	✓ Certified	Chen et al., IEEE TMI 2017
10	PnP-DnCNN PnP-DnCNN Zhang et al., 2017 33.45 dB SSIM 0.905 Try in SpecLab →	0.760	33.45	0.905	✓ Certified	Zhang et al., 2017
11	PnP-ADMM PnP-ADMM Venkatakrishnan et al., 2013 32.64 dB SSIM 0.891 Try in SpecLab →	0.740	32.64	0.891	✓ Certified	Venkatakrishnan et al., 2013
12	TV-ADMM TV-ADMM Sidky et al., 2008 30.15 dB SSIM 0.862 Try in SpecLab →	0.683	30.15	0.862	✓ Certified	Sidky et al., 2008
13	FBP FBP Kak & Slaney, 1988 27.38 dB SSIM 0.790 Try in SpecLab →	0.601	27.38	0.790	✓ Certified	Kak & Slaney, 1988

Dataset: PWM Benchmark (13 algorithms)

Blind Reconstruction Challenge — forward model has unknown mismatch, must calibrate from data. Score = 0.4 × PSNR_norm + 0.4 × SSIM + 0.2 × (1 − ‖y − Ĥx̂‖/‖y‖)

#	Method	Overall Score	Public PSNR / SSIM	Dev PSNR / SSIM	Hidden PSNR / SSIM	Trust	Source
🥇	CTFormer + gradient CTFormer + gradient Li et al., ICCV 2024 Score 0.790 Correct & Reconstruct →	0.790	0.839 36.63 dB / 0.976	0.790 33.38 dB / 0.956	0.742 30.44 dB / 0.923	✓ Certified	Li et al., ICCV 2024
🥈	Score-CT + gradient Score-CT + gradient Song et al., NeurIPS 2024 Score 0.790 Correct & Reconstruct →	0.790	0.865 38.22 dB / 0.983	0.784 33.2 dB / 0.954	0.720 29.19 dB / 0.904	✓ Certified	Song et al., NeurIPS 2024
🥉	CT-ViT + gradient CT-ViT + gradient Guo et al., NeurIPS 2024 Score 0.779 Correct & Reconstruct →	0.779	0.857 37.86 dB / 0.982	0.758 32.08 dB / 0.944	0.721 29.41 dB / 0.907	✓ Certified	Guo et al., NeurIPS 2024
4	DuDoTrans + gradient DuDoTrans + gradient Wang et al., MLMIR 2022 Score 0.771 Correct & Reconstruct →	0.771	0.820 35.73 dB / 0.972	0.761 31.15 dB / 0.933	0.731 30.39 dB / 0.923	✓ Certified	Wang et al., MLMIR 2022
5	DiffusionCT + gradient DiffusionCT + gradient Kazemi et al., ECCV 2024 Score 0.751 Correct & Reconstruct →	0.751	0.844 37.8 dB / 0.981	0.719 29.73 dB / 0.913	0.690 27.89 dB / 0.878	✓ Certified	Kazemi et al., ECCV 2024
6	Learned Primal-Dual + gradient Learned Primal-Dual + gradient Adler & Oktem, IEEE TMI 2018 Score 0.744 Correct & Reconstruct →	0.744	0.824 34.93 dB / 0.967	0.714 29.08 dB / 0.902	0.693 27.9 dB / 0.879	✓ Certified	Adler & Oktem, IEEE TMI 2018
7	PnP-DnCNN + gradient PnP-DnCNN + gradient Zhang et al., IEEE TIP 2017 Score 0.733 Correct & Reconstruct →	0.733	0.787 32.4 dB / 0.947	0.731 29.18 dB / 0.903	0.682 28.11 dB / 0.883	✓ Certified	Zhang et al., IEEE TIP 2017
8	FBPConvNet + gradient FBPConvNet + gradient Jin et al., IEEE TIP 2017 Score 0.727 Correct & Reconstruct →	0.727	0.795 33.43 dB / 0.956	0.712 28.39 dB / 0.889	0.674 26.56 dB / 0.847	✓ Certified	Jin et al., IEEE TIP 2017
9	DOLCE + gradient DOLCE + gradient Liu et al., ICCV 2023 Score 0.723 Correct & Reconstruct →	0.723	0.828 36.42 dB / 0.975	0.698 27.43 dB / 0.868	0.643 25.89 dB / 0.829	✓ Certified	Liu et al., ICCV 2023
10	PnP-ADMM + gradient PnP-ADMM + gradient Venkatakrishnan et al., IEEE GlobalSIP 2013 Score 0.713 Correct & Reconstruct →	0.713	0.770 30.89 dB / 0.929	0.697 28.38 dB / 0.889	0.673 27.3 dB / 0.865	✓ Certified	Venkatakrishnan et al., IEEE GlobalSIP 2013
11	RED-CNN + gradient RED-CNN + gradient Chen et al., IEEE TMI 2017 Score 0.693 Correct & Reconstruct →	0.693	0.763 31.34 dB / 0.935	0.673 26.18 dB / 0.837	0.642 24.88 dB / 0.798	✓ Certified	Chen et al., IEEE TMI 2017
12	TV-ADMM + gradient TV-ADMM + gradient Sidky et al., Phys. Med. Biol. 2008 Score 0.691 Correct & Reconstruct →	0.691	0.730 28.53 dB / 0.891	0.695 27.61 dB / 0.872	0.648 25.05 dB / 0.804	✓ Certified	Sidky et al., Phys. Med. Biol. 2008
13	FBP + gradient FBP + gradient Kak & Slaney, IEEE Press 1988 Score 0.623 Correct & Reconstruct →	0.623	0.644 24.7 dB / 0.792	0.633 24.83 dB / 0.797	0.593 22.88 dB / 0.726	✓ Certified	Kak & Slaney, IEEE Press 1988

Complete score requires all 3 tiers (Public + Dev + Hidden).

Join the competition →

Scoring: 0.4 × PSNR_norm + 0.4 × SSIM + 0.2 × (1 − ‖y − Ĥx̂‖/‖y‖) PSNR 40% · SSIM 40% · Consistency 20%

Public 3 scenes

Full-access development tier with all data visible.

What you get & how to use

What you get: Measurements (y), ideal forward operator (H), spec ranges, ground truth (x_true), and true mismatch spec.

How to use: Load HDF5 → compare reconstruction vs x_true → check consistency → iterate.

What to submit: Reconstructed signals (x_hat) and corrected spec as HDF5.

Public Leaderboard

#	Method	Score	PSNR	SSIM
1	Score-CT + gradient	0.865	38.22	0.983
2	CT-ViT + gradient	0.857	37.86	0.982
3	DiffusionCT + gradient	0.844	37.8	0.981
4	CTFormer + gradient	0.839	36.63	0.976
5	DOLCE + gradient	0.828	36.42	0.975
6	Learned Primal-Dual + gradient	0.824	34.93	0.967
7	DuDoTrans + gradient	0.820	35.73	0.972
8	FBPConvNet + gradient	0.795	33.43	0.956
9	PnP-DnCNN + gradient	0.787	32.4	0.947
10	PnP-ADMM + gradient	0.770	30.89	0.929
11	RED-CNN + gradient	0.763	31.34	0.935
12	TV-ADMM + gradient	0.730	28.53	0.891
13	FBP + gradient	0.644	24.7	0.792

Spec Ranges (4 parameters)

Parameter	Min	Max	Unit
isocenter_shift	-0.4	0.8	mm
beam_energy_variation	5.96	6.08	MV
detector_sag	-0.2	0.4	mm
scatter_kernel_width	4.6	5.8	mm

View Details →

Dev 3 scenes

Blind evaluation tier — no ground truth available.

What you get & how to use

What you get: Measurements (y), ideal forward operator (H), and spec ranges only.

How to use: Apply your pipeline from the Public tier. Use consistency as self-check.

What to submit: Reconstructed signals and corrected spec. Scored server-side.

Dev Leaderboard

#	Method	Score	PSNR	SSIM
1	CTFormer + gradient	0.790	33.38	0.956
2	Score-CT + gradient	0.784	33.2	0.954
3	DuDoTrans + gradient	0.761	31.15	0.933
4	CT-ViT + gradient	0.758	32.08	0.944
5	PnP-DnCNN + gradient	0.731	29.18	0.903
6	DiffusionCT + gradient	0.719	29.73	0.913
7	Learned Primal-Dual + gradient	0.714	29.08	0.902
8	FBPConvNet + gradient	0.712	28.39	0.889
9	DOLCE + gradient	0.698	27.43	0.868
10	PnP-ADMM + gradient	0.697	28.38	0.889
11	TV-ADMM + gradient	0.695	27.61	0.872
12	RED-CNN + gradient	0.673	26.18	0.837
13	FBP + gradient	0.633	24.83	0.797

Spec Ranges (4 parameters)

Parameter	Min	Max	Unit
isocenter_shift	-0.48	0.72	mm
beam_energy_variation	5.952	6.072	MV
detector_sag	-0.24	0.36	mm
scatter_kernel_width	4.52	5.72	mm

Submit Reconstruction View Details →

Hidden 3 scenes

Fully blind server-side evaluation — no data download.

What you get & how to use

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.

Hidden Leaderboard

#	Method	Score	PSNR	SSIM
1	CTFormer + gradient	0.742	30.44	0.923
2	DuDoTrans + gradient	0.731	30.39	0.923
3	CT-ViT + gradient	0.721	29.41	0.907
4	Score-CT + gradient	0.720	29.19	0.904
5	Learned Primal-Dual + gradient	0.693	27.9	0.879
6	DiffusionCT + gradient	0.690	27.89	0.878
7	PnP-DnCNN + gradient	0.682	28.11	0.883
8	FBPConvNet + gradient	0.674	26.56	0.847
9	PnP-ADMM + gradient	0.673	27.3	0.865
10	TV-ADMM + gradient	0.648	25.05	0.804
11	DOLCE + gradient	0.643	25.89	0.829
12	RED-CNN + gradient	0.642	24.88	0.798
13	FBP + gradient	0.593	22.88	0.726

Spec Ranges (4 parameters)

Parameter	Min	Max	Unit
isocenter_shift	-0.28	0.92	mm
beam_energy_variation	5.972	6.092	MV
detector_sag	-0.14	0.46	mm
scatter_kernel_width	4.72	5.92	mm

Submit Algorithm View Details →

Blind Reconstruction Challenge

Challenge

Given measurements with unknown mismatch and spec ranges (not exact params), reconstruct the original signal. A method must be evaluated on all three tiers for a complete score. Scored on a composite metric: 0.4 × PSNR_norm + 0.4 × SSIM + 0.2 × (1 − ‖y − Ĥx̂‖/‖y‖).

Input

Measurements y, ideal forward model H, spec ranges

Output

Reconstructed signal x̂

Spec DAG — Forward Model Pipeline

Π → D

Π Projection

D Detector

Mismatch Parameters

Symbol	Parameter	Description	Nominal	Perturbed
i_s	isocenter_shift	Isocenter shift (mm)	0.0	0.4
b_e	beam_energy_variation	Beam energy variation (MV)	6.0	6.04
d_s	detector_sag	Detector sag (mm)	0.0	0.2
s_k	scatter_kernel_width	Scatter kernel width (mm)	5.0	5.4

Credits System

40%

Platform Profit Pool

Revenue allocated to benchmark rewards

30%

Winner Share

Top algorithm receives from pool

$100

Min Withdrawal

Minimum payout threshold

Spec Primitives Reference (11 primitives)

P Propagation

Free-space or medium propagation kernel (Fresnel, Rayleigh-Sommerfeld).

M Mask / Modulation

Spatial or spatio-temporal amplitude modulation (coded aperture, SLM pattern).

Π Projection

Geometric projection operator (Radon transform, fan-beam, cone-beam).

F Fourier Sampling

Sampling in the Fourier / k-space domain (MRI, ptychography).

C Convolution

Shift-invariant convolution with a point-spread function (PSF).

Σ Summation / Integration

Summation along a physical dimension (spectral, temporal, angular).

D Detector

Sensor readout with gain g and noise model η (Gaussian, Poisson, mixed).

S Structured Illumination

Patterned illumination (block, Hadamard, random) applied to the scene.

W Wavelength Dispersion

Spectral dispersion element (prism, grating) with shift α and aperture a.

R Rotation / Motion

Sample or gantry rotation (CT, electron tomography).

Λ Wavelength Selection

Spectral filter or monochromator selecting a wavelength band.