Differential Interference Contrast (DIC)

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
🥇	DiffusionDIC DiffusionDIC Luo 2023 39.2 dB SSIM 0.950 Checkpoint unavailable	0.878	39.2	0.950	✓ Certified	Luo 2023
🥈	PhysPhase-Net PhysPhase-Net Barbastathis 2019 37.4 dB SSIM 0.935 Checkpoint unavailable	0.841	37.4	0.935	✓ Certified	Barbastathis 2019
🥉	SwinDIC SwinDIC Liang 2021 36.1 dB SSIM 0.921 Checkpoint unavailable	0.812	36.1	0.921	✓ Certified	Liang 2021
4	PhaseNet-DIC PhaseNet-DIC Sinha 2020 33.7 dB SSIM 0.884 Checkpoint unavailable	0.754	33.7	0.884	✓ Certified	Sinha 2020
5	PnP-DIC PnP-DIC Kamilov 2017 32.2 dB SSIM 0.869 Try in SpecLab →	0.721	32.2	0.869	✓ Certified	Kamilov 2017
6	DIC-CNN DIC-CNN Rivenson 2018 31.4 dB SSIM 0.856 Checkpoint unavailable	0.701	31.4	0.856	✓ Certified	Rivenson 2018
7	TV-DIC TV-DIC Bostan 2014 27.8 dB SSIM 0.793 Try in SpecLab →	0.610	27.8	0.793	✓ Certified	Bostan 2014
8	Phase-DLSIM Phase-DLSIM Stephens 2003 25.9 dB SSIM 0.762 Try in SpecLab →	0.563	25.9	0.762	✓ Certified	Stephens 2003
9	DIC-Deconv DIC-Deconv Preza 1999 24.1 dB SSIM 0.731 Try in SpecLab →	0.517	24.1	0.731	✓ Certified	Preza 1999

Dataset: PWM Benchmark (9 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
🥇	SwinDIC + gradient SwinDIC + gradient Liang et al., ICCV 2021 (DIC) Score 0.762 Correct & Reconstruct →	0.762	0.798 33.5 dB / 0.957	0.763 32.12 dB / 0.944	0.724 29.24 dB / 0.904	✓ Certified	Liang et al., ICCV 2021 (DIC)
🥈	DiffusionDIC + gradient DiffusionDIC + gradient Luo et al., Nat. Photonics 2023 (DIC) Score 0.761 Correct & Reconstruct →	0.761	0.835 36.48 dB / 0.976	0.744 30.23 dB / 0.920	0.705 28.38 dB / 0.889	✓ Certified	Luo et al., Nat. Photonics 2023 (DIC)
🥉	PhysPhase-Net + gradient PhysPhase-Net + gradient Barbastathis et al., Optica 2019 Score 0.727 Correct & Reconstruct →	0.727	0.814 35.1 dB / 0.968	0.700 28.68 dB / 0.894	0.668 26.26 dB / 0.839	✓ Certified	Barbastathis et al., Optica 2019
4	PhaseNet-DIC + gradient PhaseNet-DIC + gradient Sinha et al., Optica 2020 Score 0.683 Correct & Reconstruct →	0.683	0.788 32.35 dB / 0.946	0.650 25.95 dB / 0.831	0.610 23.75 dB / 0.760	✓ Certified	Sinha et al., Optica 2020
5	PnP-DIC + gradient PnP-DIC + gradient Kamilov et al., Optica 2017 (DIC) Score 0.659 Correct & Reconstruct →	0.659	0.739 29.5 dB / 0.909	0.645 25.72 dB / 0.824	0.592 23.53 dB / 0.751	✓ Certified	Kamilov et al., Optica 2017 (DIC)
6	DIC-CNN + gradient DIC-CNN + gradient Rivenson et al., Optica 2018 Score 0.643 Correct & Reconstruct →	0.643	0.722 28.43 dB / 0.890	0.637 24.55 dB / 0.788	0.570 23.0 dB / 0.731	✓ Certified	Rivenson et al., Optica 2018
7	Phase-DLSIM + gradient Phase-DLSIM + gradient Stephens & Allen, J. Biomed. Opt. 2003 Score 0.574 Correct & Reconstruct →	0.574	0.620 23.93 dB / 0.766	0.582 22.81 dB / 0.724	0.519 20.33 dB / 0.614	✓ Certified	Stephens & Allen, J. Biomed. Opt. 2003
8	DIC-Deconv + gradient DIC-Deconv + gradient Preza et al., JOSA A 1999 Score 0.560 Correct & Reconstruct →	0.560	0.606 22.99 dB / 0.731	0.559 22.1 dB / 0.694	0.516 20.35 dB / 0.615	✓ Certified	Preza et al., JOSA A 1999
9	TV-DIC + gradient TV-DIC + gradient Bostan et al., IEEE TIP 2014 Score 0.542 Correct & Reconstruct →	0.542	0.685 26.21 dB / 0.838	0.512 20.42 dB / 0.619	0.429 17.35 dB / 0.468	✓ Certified	Bostan et al., IEEE TIP 2014

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	DiffusionDIC + gradient	0.835	36.48	0.976
2	PhysPhase-Net + gradient	0.814	35.1	0.968
3	SwinDIC + gradient	0.798	33.5	0.957
4	PhaseNet-DIC + gradient	0.788	32.35	0.946
5	PnP-DIC + gradient	0.739	29.5	0.909
6	DIC-CNN + gradient	0.722	28.43	0.89
7	TV-DIC + gradient	0.685	26.21	0.838
8	Phase-DLSIM + gradient	0.620	23.93	0.766
9	DIC-Deconv + gradient	0.606	22.99	0.731

Spec Ranges (3 parameters)

Parameter	Min	Max	Unit
shear_amount	80.0	140.0	nm
bias_retardation	-0.15	0.15	nm
prism_orientation	-0.6	1.2	deg

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	SwinDIC + gradient	0.763	32.12	0.944
2	DiffusionDIC + gradient	0.744	30.23	0.92
3	PhysPhase-Net + gradient	0.700	28.68	0.894
4	PhaseNet-DIC + gradient	0.650	25.95	0.831
5	PnP-DIC + gradient	0.645	25.72	0.824
6	DIC-CNN + gradient	0.637	24.55	0.788
7	Phase-DLSIM + gradient	0.582	22.81	0.724
8	DIC-Deconv + gradient	0.559	22.1	0.694
9	TV-DIC + gradient	0.512	20.42	0.619

Spec Ranges (3 parameters)

Parameter	Min	Max	Unit
shear_amount	76.0	136.0	nm
bias_retardation	-0.15	0.15	nm
prism_orientation	-0.72	1.08	deg

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	SwinDIC + gradient	0.724	29.24	0.904
2	DiffusionDIC + gradient	0.705	28.38	0.889
3	PhysPhase-Net + gradient	0.668	26.26	0.839
4	PhaseNet-DIC + gradient	0.610	23.75	0.76
5	PnP-DIC + gradient	0.592	23.53	0.751
6	DIC-CNN + gradient	0.570	23.0	0.731
7	Phase-DLSIM + gradient	0.519	20.33	0.614
8	DIC-Deconv + gradient	0.516	20.35	0.615
9	TV-DIC + gradient	0.429	17.35	0.468

Spec Ranges (3 parameters)

Parameter	Min	Max	Unit
shear_amount	86.0	146.0	nm
bias_retardation	-0.15	0.15	nm
prism_orientation	-0.42	1.38	deg

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 → C → D

M Modulation

C Convolution

D Detector

Mismatch Parameters

Symbol	Parameter	Description	Nominal	Perturbed
s_a	shear_amount	Shear amount (nm)	100.0	120.0
b_r	bias_retardation	Bias retardation (nm)	0.0	0.0
p_o	prism_orientation	Prism orientation (deg)	0.0	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.