Polarization

Polarization Microscopy

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
🥇	ScoreMicro ScoreMicro Wei et al., ECCV 2025 38.48 dB SSIM 0.981 Checkpoint unavailable	0.882	38.48	0.981	✓ Certified	Wei et al., ECCV 2025
🥈	DiffDeconv DiffDeconv Huang et al., NeurIPS 2024 38.12 dB SSIM 0.979 Checkpoint unavailable	0.875	38.12	0.979	✓ Certified	Huang et al., NeurIPS 2024
🥉	Restormer+ Restormer+ Zamir et al., ICCV 2024 37.65 dB SSIM 0.975 Checkpoint unavailable	0.865	37.65	0.975	✓ Certified	Zamir et al., ICCV 2024
4	DeconvFormer DeconvFormer Chen et al., CVPR 2024 37.25 dB SSIM 0.972 Checkpoint unavailable	0.857	37.25	0.972	✓ Certified	Chen et al., CVPR 2024
5	ResUNet ResUNet DeCelle et al., Nat. Methods 2021 35.85 dB SSIM 0.964 Checkpoint unavailable	0.830	35.85	0.964	✓ Certified	DeCelle et al., Nat. Methods 2021
6	Restormer Restormer Zamir et al., CVPR 2022 35.8 dB SSIM 0.962 Checkpoint unavailable	0.828	35.8	0.962	✓ Certified	Zamir et al., CVPR 2022
7	U-Net U-Net Ronneberger et al., MICCAI 2015 35.15 dB SSIM 0.956 Checkpoint unavailable	0.814	35.15	0.956	✓ Certified	Ronneberger et al., MICCAI 2015
8	CARE CARE Weigert et al., Nat. Methods 2018 34.5 dB SSIM 0.948 Checkpoint unavailable	0.799	34.5	0.948	✓ Certified	Weigert et al., Nat. Methods 2018
9	PnP-DnCNN PnP-DnCNN Zhang et al., IEEE TIP 2017 31.2 dB SSIM 0.890 Try in SpecLab →	0.715	31.2	0.890	✓ Certified	Zhang et al., IEEE TIP 2017
10	PnP-FISTA PnP-FISTA Bai et al., 2020 30.42 dB SSIM 0.872 Try in SpecLab →	0.693	30.42	0.872	✓ Certified	Bai et al., 2020
11	TV-Deconvolution TV-Deconvolution TV-regularized deconvolution 29.5 dB SSIM 0.845 Try in SpecLab →	0.664	29.5	0.845	✓ Certified	TV-regularized deconvolution
12	Wiener Filter Wiener Filter Analytical baseline 28.35 dB SSIM 0.805 Try in SpecLab →	0.625	28.35	0.805	✓ Certified	Analytical baseline
13	Richardson-Lucy Richardson-Lucy Richardson 1972 / Lucy 1974 27.1 dB SSIM 0.770 Try in SpecLab →	0.587	27.1	0.770	✓ Certified	Richardson 1972 / Lucy 1974

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
🥇	Restormer+ + gradient Restormer+ + gradient Zamir et al., ICCV 2024 Score 0.770 Correct & Reconstruct →	0.770	0.819 35.49 dB / 0.971	0.776 33.1 dB / 0.953	0.714 28.78 dB / 0.896	✓ Certified	Zamir et al., ICCV 2024
🥈	ScoreMicro + gradient ScoreMicro + gradient Wei et al., ECCV 2025 Score 0.763 Correct & Reconstruct →	0.763	0.827 35.66 dB / 0.972	0.746 31.35 dB / 0.935	0.717 29.03 dB / 0.901	✓ Certified	Wei et al., ECCV 2025
🥉	DeconvFormer + gradient DeconvFormer + gradient Chen et al., CVPR 2024 Score 0.754 Correct & Reconstruct →	0.754	0.811 34.53 dB / 0.965	0.749 30.64 dB / 0.926	0.701 29.03 dB / 0.901	✓ Certified	Chen et al., CVPR 2024
4	ResUNet + gradient ResUNet + gradient DeCelle et al., Nat. Methods 2021 Score 0.737 Correct & Reconstruct →	0.737	0.796 33.66 dB / 0.958	0.718 29.36 dB / 0.907	0.698 27.56 dB / 0.871	✓ Certified	DeCelle et al., Nat. Methods 2021
5	DiffDeconv + gradient DiffDeconv + gradient Huang et al., NeurIPS 2024 Score 0.725 Correct & Reconstruct →	0.725	0.822 35.36 dB / 0.970	0.697 27.51 dB / 0.870	0.657 26.3 dB / 0.840	✓ Certified	Huang et al., NeurIPS 2024
6	Restormer + gradient Restormer + gradient Zamir et al., CVPR 2022 Score 0.719 Correct & Reconstruct →	0.719	0.793 33.32 dB / 0.955	0.729 29.99 dB / 0.917	0.635 24.74 dB / 0.794	✓ Certified	Zamir et al., CVPR 2022
7	U-Net + gradient U-Net + gradient Ronneberger et al., MICCAI 2015 Score 0.691 Correct & Reconstruct →	0.691	0.808 33.48 dB / 0.957	0.670 25.88 dB / 0.829	0.594 23.6 dB / 0.754	✓ Certified	Ronneberger et al., MICCAI 2015
8	CARE + gradient CARE + gradient Weigert et al., Nat. Methods 2018 Score 0.682 Correct & Reconstruct →	0.682	0.778 32.63 dB / 0.949	0.665 25.81 dB / 0.827	0.603 23.56 dB / 0.753	✓ Certified	Weigert et al., Nat. Methods 2018
9	PnP-FISTA + gradient PnP-FISTA + gradient Bai et al., 2020 Score 0.676 Correct & Reconstruct →	0.676	0.704 27.46 dB / 0.869	0.659 25.71 dB / 0.824	0.666 26.46 dB / 0.844	✓ Certified	Bai et al., 2020
10	PnP-DnCNN + gradient PnP-DnCNN + gradient Zhang et al., IEEE TIP 2017 Score 0.668 Correct & Reconstruct →	0.668	0.750 29.7 dB / 0.912	0.652 25.85 dB / 0.828	0.602 23.29 dB / 0.742	✓ Certified	Zhang et al., IEEE TIP 2017
11	TV-Deconvolution + gradient TV-Deconvolution + gradient Rudin et al., Phys. A 1992 Score 0.663 Correct & Reconstruct →	0.663	0.694 27.11 dB / 0.861	0.654 25.51 dB / 0.818	0.642 25.59 dB / 0.820	✓ Certified	Rudin et al., Phys. A 1992
12	Wiener Filter + gradient Wiener Filter + gradient Analytical baseline Score 0.657 Correct & Reconstruct →	0.657	0.664 25.49 dB / 0.817	0.670 26.54 dB / 0.847	0.637 25.38 dB / 0.814	✓ Certified	Analytical baseline
13	Richardson-Lucy + gradient Richardson-Lucy + gradient Richardson, JOSA 1972 / Lucy, AJ 1974 Score 0.614 Correct & Reconstruct →	0.614	0.637 24.42 dB / 0.783	0.634 24.79 dB / 0.795	0.570 22.31 dB / 0.703	✓ Certified	Richardson, JOSA 1972 / Lucy, AJ 1974

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	ScoreMicro + gradient	0.827	35.66	0.972
2	DiffDeconv + gradient	0.822	35.36	0.97
3	Restormer+ + gradient	0.819	35.49	0.971
4	DeconvFormer + gradient	0.811	34.53	0.965
5	U-Net + gradient	0.808	33.48	0.957
6	ResUNet + gradient	0.796	33.66	0.958
7	Restormer + gradient	0.793	33.32	0.955
8	CARE + gradient	0.778	32.63	0.949
9	PnP-DnCNN + gradient	0.750	29.7	0.912
10	PnP-FISTA + gradient	0.704	27.46	0.869
11	TV-Deconvolution + gradient	0.694	27.11	0.861
12	Wiener Filter + gradient	0.664	25.49	0.817
13	Richardson-Lucy + gradient	0.637	24.42	0.783

Spec Ranges (3 parameters)

Parameter	Min	Max	Unit
extinction_ratio	-0.5	1.0	dB
retardance	-2.0	4.0	nm
alignment	-0.5	1.0	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	Restormer+ + gradient	0.776	33.1	0.953
2	DeconvFormer + gradient	0.749	30.64	0.926
3	ScoreMicro + gradient	0.746	31.35	0.935
4	Restormer + gradient	0.729	29.99	0.917
5	ResUNet + gradient	0.718	29.36	0.907
6	DiffDeconv + gradient	0.697	27.51	0.87
7	U-Net + gradient	0.670	25.88	0.829
8	Wiener Filter + gradient	0.670	26.54	0.847
9	CARE + gradient	0.665	25.81	0.827
10	PnP-FISTA + gradient	0.659	25.71	0.824
11	TV-Deconvolution + gradient	0.654	25.51	0.818
12	PnP-DnCNN + gradient	0.652	25.85	0.828
13	Richardson-Lucy + gradient	0.634	24.79	0.795

Spec Ranges (3 parameters)

Parameter	Min	Max	Unit
extinction_ratio	-0.6	0.9	dB
retardance	-2.4	3.6	nm
alignment	-0.6	0.9	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	ScoreMicro + gradient	0.717	29.03	0.901
2	Restormer+ + gradient	0.714	28.78	0.896
3	DeconvFormer + gradient	0.701	29.03	0.901
4	ResUNet + gradient	0.698	27.56	0.871
5	PnP-FISTA + gradient	0.666	26.46	0.844
6	DiffDeconv + gradient	0.657	26.3	0.84
7	TV-Deconvolution + gradient	0.642	25.59	0.82
8	Wiener Filter + gradient	0.637	25.38	0.814
9	Restormer + gradient	0.635	24.74	0.794
10	CARE + gradient	0.603	23.56	0.753
11	PnP-DnCNN + gradient	0.602	23.29	0.742
12	U-Net + gradient	0.594	23.6	0.754
13	Richardson-Lucy + gradient	0.570	22.31	0.703

Spec Ranges (3 parameters)

Parameter	Min	Max	Unit
extinction_ratio	-0.35	1.15	dB
retardance	-1.4	4.6	nm
alignment	-0.35	1.15	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

About the Imaging Modality

Polarization microscopy measures anisotropic optical properties by analysing the polarisation state of light through the sample. In Mueller matrix imaging, the sample is illuminated with known polarisation states and the output is analysed, yielding a 4x4 Mueller matrix at each pixel encoding birefringence, optical activity, and depolarisation. The LC-PolScope uses liquid crystal retarders for rapid modulation. Reconstruction involves solving for Mueller elements and Lu-Chipman decomposition into physically meaningful parameters.

Principle

Polarization microscopy exploits the birefringence (orientation-dependent refractive index) of ordered biological structures such as collagen fibers, spindle microtubules, and crystalline inclusions. By analyzing the polarization state of transmitted or reflected light, structural anisotropy can be measured without fluorescent labeling. Quantitative techniques (LC-PolScope) measure both retardance magnitude and slow-axis orientation.

How to Build the System

Mount a liquid-crystal universal compensator (LC-PolScope by OpenPolScope, or Abrio system) on a standard brightfield microscope. Use strain-free optics and rotate the analyzer while keeping the polarizer fixed (or use a rotating stage). For quantitative imaging, acquire 4-5 images at different compensator settings. A monochromatic light source (546 nm green filter) minimizes chromatic effects.

Common Reconstruction Algorithms

Mueller matrix decomposition (full polarimetric imaging)
Jones calculus for coherent polarization analysis
Background retardance subtraction
Stokes parameter reconstruction from intensity measurements
Deep-learning retardance estimation from fewer raw frames

Common Mistakes

Strain birefringence in optical components contaminating the measurement
Incorrect compensator calibration producing quantitative retardance errors
Not accounting for sample tilt introducing apparent birefringence artifacts
Using polychromatic light causing wavelength-dependent retardance errors
Ignoring depolarization effects in thick or scattering samples

How to Avoid Mistakes

Use strain-free objectives and verify zero retardance on a blank field
Calibrate the liquid-crystal compensator at each session using a known retarder
Ensure sample is flat and perpendicular to the optical axis
Use narrow-band illumination or measure dispersion for wavelength correction
For thick samples, consider Mueller matrix imaging to capture depolarization

Forward-Model Mismatch Cases

The widefield fallback treats light as a scalar intensity, but polarization microscopy measures the full Mueller matrix or Stokes parameters — the vector nature of light (birefringence, dichroism, depolarization) is completely lost
The fallback produces a single-channel image, but the correct operator generates 4+ channels (Stokes S0-S3 or multiple polarizer/analyzer orientations), each encoding different polarization properties of the sample

How to Correct the Mismatch

Use the polarization operator that generates images at multiple polarizer/analyzer angles (0, 45, 90, 135 degrees), encoding the sample's Jones or Mueller matrix at each pixel
Reconstruct birefringence retardance and orientation from the polarization-resolved measurements using Mueller calculus or Jones matrix decomposition

Experimental Setup — Signal Chain

Experimental setup diagram for Polarization Microscopy

Reconstruction Gallery — 4 Scenes × 3 Scenarios

Method: CPU_baseline | Mismatch: nominal (nominal=True, perturbed=False)

I Ideal II Mismatch (uncorrected) III Oracle correction

I — Ideal

Ground Truth

Measurement

Reconstruction

17.022206064497794 dB SSIM 0.38380508849556166

II — Mismatch

Ground Truth

Measurement

Reconstruction

14.93801985037615 dB SSIM 0.24399594480247574

III — Oracle

Ground Truth

Measurement (perturbed)

Reconstruction

20.039527108294998 dB SSIM 0.5012899560940253

I — Ideal

Ground Truth

Measurement

Reconstruction

14.231375949985203 dB SSIM 0.29689664293183293

II — Mismatch

Ground Truth

Measurement

Reconstruction

12.92413737908083 dB SSIM 0.20980780569697918

III — Oracle

Ground Truth

Measurement (perturbed)

Reconstruction

19.225830712320835 dB SSIM 0.547962131323724

I — Ideal

Ground Truth

Measurement

Reconstruction

8.520061941863302 dB SSIM 0.3871062840596513

II — Mismatch

Ground Truth

Measurement

Reconstruction

8.013928482144143 dB SSIM 0.23077294300010404

III — Oracle

Ground Truth

Measurement (perturbed)

Reconstruction

20.113884181775745 dB SSIM 0.3416888134023018

I — Ideal

Ground Truth

Measurement

Reconstruction

13.297073801541249 dB SSIM 0.5284335563352818

II — Mismatch

Ground Truth

Measurement

Reconstruction

11.66205075678512 dB SSIM 0.26941399804629346

III — Oracle

Ground Truth

Measurement (perturbed)

Reconstruction

19.61468818224941 dB SSIM 0.44080752632935544

Mean PSNR Across All Scenes

13.267679439471888 dB

I — Ideal

11.884534117096562 dB

II — Mismatch

19.748482546160247 dB

III — Oracle

Degradation: -1.4 dB Recovery: +7.9 dB

Per-scene PSNR breakdown (4 scenes)

Scene	I (PSNR)	I (SSIM)	II (PSNR)	II (SSIM)	III (PSNR)	III (SSIM)
scene_00	17.022206064497794	0.38380508849556166	14.93801985037615	0.24399594480247574	20.039527108294998	0.5012899560940253
scene_01	14.231375949985203	0.29689664293183293	12.92413737908083	0.20980780569697918	19.225830712320835	0.547962131323724
scene_02	8.520061941863302	0.3871062840596513	8.013928482144143	0.23077294300010404	20.113884181775745	0.3416888134023018
scene_03	13.297073801541249	0.5284335563352818	11.66205075678512	0.26941399804629346	19.61468818224941	0.44080752632935544
Mean	13.267679439471888	0.39906039295558193	11.884534117096562	0.23849767288646312	19.748482546160247	0.45793710678735156

Experimental Setup

Instrument: CRi Abrio / OpenPolScope

Objective: Plan Fluor 60x / 1.30 NA oil

Pixel Size Nm: 110

Wavelength Nm: 546

Polarisation States: 4

Retarder: liquid crystal variable retarder (Meadowlark)

Detector: sCMOS 2048x2048

Application: birefringence / collagen fibre mapping

Key References

Mehta et al., 'Quantitative polarized light microscopy using the LC-PolScope', Live Cell Imaging: A Laboratory Manual, CSHL Press (2010)
Lu & Chipman, 'Interpretation of Mueller matrices based on polar decomposition', J. Opt. Soc. Am. A 13, 1106-1113 (1996)

Canonical Datasets

OpenPolScope calibration data
Collagen SHG/polarisation histopathology datasets

Spec DAG — Forward Model Pipeline

M(polarizer) → C(PSF) → D(g, η₁)

M Polarizer / Analyzer (polarizer)

C PSF Convolution (PSF)

D Camera (g, η₁)

Mismatch Parameters

Symbol	Parameter	Description	Perturbed
ΔER	extinction_ratio	Extinction ratio error (dB)	0.5
Δδ	retardance	Retardance error (nm)	2.0
Δθ	alignment	Polarizer alignment error (deg)	0.5

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.