Talbot-Lau X-ray Grating Interferometry

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
🥇	GratingFormer GratingFormer Grating interferometry transformer, 2024 32.0 dB SSIM 0.943 Checkpoint unavailable	0.755	32.0	0.943	✓ Certified	Grating interferometry transformer, 2024
🥈	DPC-Net DPC-Net Differential phase contrast CNN, 2021 29.09 dB SSIM 0.902 Checkpoint unavailable	0.686	29.09	0.902	✓ Certified	Differential phase contrast CNN, 2021
🥉	PCA Retrieval PCA Retrieval Zanette et al., PMB 2012 24.26 dB SSIM 0.778 Try in SpecLab →	0.543	24.26	0.778	✓ Certified	Zanette et al., PMB 2012
4	Phase Stepping Phase Stepping Weitkamp et al., Opt. Express 2005 23.57 dB SSIM 0.753 Try in SpecLab →	0.519	23.57	0.753	✓ Certified	Weitkamp et al., Opt. Express 2005

Dataset: PWM Benchmark (4 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
🥇	GratingFormer + gradient GratingFormer + gradient Grating interferometry transformer, 2024 Score 0.700 Correct & Reconstruct →	0.700	0.741 30.04 dB / 0.917	0.706 28.63 dB / 0.893	0.653 26.28 dB / 0.840	✓ Certified	Grating interferometry transformer, 2024
🥈	DPC-Net + gradient DPC-Net + gradient Differential phase contrast CNN, 2021 Score 0.549 Correct & Reconstruct →	0.549	0.716 27.95 dB / 0.880	0.508 19.6 dB / 0.579	0.424 17.33 dB / 0.467	✓ Certified	Differential phase contrast CNN, 2021
🥉	PCA Retrieval + gradient PCA Retrieval + gradient Zanette et al., PMB 2012 Score 0.535 Correct & Reconstruct →	0.535	0.560 21.34 dB / 0.661	0.544 20.98 dB / 0.645	0.501 20.33 dB / 0.614	✓ Certified	Zanette et al., PMB 2012
4	Phase Stepping + gradient Phase Stepping + gradient Weitkamp et al., Opt. Express 2005 Score 0.524 Correct & Reconstruct →	0.524	0.557 21.54 dB / 0.670	0.513 19.98 dB / 0.598	0.503 20.05 dB / 0.601	✓ Certified	Weitkamp et al., Opt. Express 2005

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 5 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	GratingFormer + gradient	0.741	30.04	0.917
2	DPC-Net + gradient	0.716	27.95	0.88
3	PCA Retrieval + gradient	0.560	21.34	0.661
4	Phase Stepping + gradient	0.557	21.54	0.67

Spec Ranges (4 parameters)

Parameter	Min	Max	Unit
grating_alignment_(rotation)	-0.1	0.2	deg
inter_grating_distance_error	-0.2	0.4	-
phase_stepping_error	-1.0	2.0	perstep
grating_defect_fraction	-0.6	1.2	-

View Details →

Dev 5 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	GratingFormer + gradient	0.706	28.63	0.893
2	PCA Retrieval + gradient	0.544	20.98	0.645
3	Phase Stepping + gradient	0.513	19.98	0.598
4	DPC-Net + gradient	0.508	19.6	0.579

Spec Ranges (4 parameters)

Parameter	Min	Max	Unit
grating_alignment_(rotation)	-0.12	0.18	deg
inter_grating_distance_error	-0.24	0.36	-
phase_stepping_error	-1.2	1.8	perstep
grating_defect_fraction	-0.72	1.08	-

Submit Reconstruction View Details →

Hidden 5 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	GratingFormer + gradient	0.653	26.28	0.84
2	Phase Stepping + gradient	0.503	20.05	0.601
3	PCA Retrieval + gradient	0.501	20.33	0.614
4	DPC-Net + gradient	0.424	17.33	0.467

Spec Ranges (4 parameters)

Parameter	Min	Max	Unit
grating_alignment_(rotation)	-0.07	0.23	deg
inter_grating_distance_error	-0.14	0.46	-
phase_stepping_error	-0.7	2.3	perstep
grating_defect_fraction	-0.42	1.38	-

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

M → P → D

M Modulation

P Propagation

D Detector

Mismatch Parameters

Symbol	Parameter	Description	Perturbed
g_a	grating_alignment_(rotation)	Grating alignment (rotation) (deg)	0.1
i_d	inter_grating_distance_error	Inter-grating distance error (-)	0.2
p_s	phase_stepping_error	Phase stepping error (per step)	1.0
g_d	grating_defect_fraction	Grating defect fraction (-)	0.6

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.