TIRF

TIRF 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
🥇	DeconvFormer + gradient DeconvFormer + gradient Chen et al., CVPR 2024 Score 0.762 Correct & Reconstruct →	0.762	0.812 34.57 dB / 0.965	0.755 31.98 dB / 0.942	0.719 29.54 dB / 0.910	✓ Certified	Chen et al., CVPR 2024
🥈	Restormer+ + gradient Restormer+ + gradient Zamir et al., ICCV 2024 Score 0.761 Correct & Reconstruct →	0.761	0.839 36.18 dB / 0.974	0.754 30.57 dB / 0.925	0.689 27.94 dB / 0.880	✓ Certified	Zamir et al., ICCV 2024
🥉	DiffDeconv + gradient DiffDeconv + gradient Huang et al., NeurIPS 2024 Score 0.743 Correct & Reconstruct →	0.743	0.846 37.04 dB / 0.978	0.726 28.86 dB / 0.898	0.658 26.56 dB / 0.847	✓ Certified	Huang et al., NeurIPS 2024
4	ScoreMicro + gradient ScoreMicro + gradient Wei et al., ECCV 2025 Score 0.732 Correct & Reconstruct →	0.732	0.826 35.51 dB / 0.971	0.716 28.85 dB / 0.898	0.654 25.98 dB / 0.831	✓ Certified	Wei et al., ECCV 2025
5	Restormer + gradient Restormer + gradient Zamir et al., CVPR 2022 Score 0.712 Correct & Reconstruct →	0.712	0.796 33.96 dB / 0.961	0.724 29.43 dB / 0.908	0.617 24.08 dB / 0.771	✓ Certified	Zamir et al., CVPR 2022
6	PnP-FISTA + gradient PnP-FISTA + gradient Bai et al., 2020 Score 0.679 Correct & Reconstruct →	0.679	0.708 27.65 dB / 0.873	0.676 26.94 dB / 0.857	0.654 25.86 dB / 0.828	✓ Certified	Bai et al., 2020
7	CARE + gradient CARE + gradient Weigert et al., Nat. Methods 2018 Score 0.671 Correct & Reconstruct →	0.671	0.802 33.47 dB / 0.957	0.628 24.28 dB / 0.778	0.583 22.48 dB / 0.710	✓ Certified	Weigert et al., Nat. Methods 2018
8	ResUNet + gradient ResUNet + gradient DeCelle et al., Nat. Methods 2021 Score 0.671 Correct & Reconstruct →	0.671	0.796 33.5 dB / 0.957	0.655 26.01 dB / 0.832	0.563 22.05 dB / 0.692	✓ Certified	DeCelle et al., Nat. Methods 2021
9	TV-Deconvolution + gradient TV-Deconvolution + gradient Rudin et al., Phys. A 1992 Score 0.668 Correct & Reconstruct →	0.668	0.687 26.6 dB / 0.848	0.675 26.78 dB / 0.853	0.641 25.44 dB / 0.816	✓ Certified	Rudin et al., Phys. A 1992
10	U-Net + gradient U-Net + gradient Ronneberger et al., MICCAI 2015 Score 0.657 Correct & Reconstruct →	0.657	0.785 32.67 dB / 0.950	0.627 24.76 dB / 0.794	0.559 22.4 dB / 0.707	✓ Certified	Ronneberger et al., MICCAI 2015
11	PnP-DnCNN + gradient PnP-DnCNN + gradient Zhang et al., IEEE TIP 2017 Score 0.647 Correct & Reconstruct →	0.647	0.726 29.11 dB / 0.902	0.649 25.66 dB / 0.822	0.566 22.59 dB / 0.715	✓ Certified	Zhang et al., IEEE TIP 2017
12	Wiener Filter + gradient Wiener Filter + gradient Analytical baseline Score 0.627 Correct & Reconstruct →	0.627	0.672 26.27 dB / 0.839	0.622 23.95 dB / 0.767	0.587 22.62 dB / 0.716	✓ Certified	Analytical baseline
13	Richardson-Lucy + gradient Richardson-Lucy + gradient Richardson, JOSA 1972 / Lucy, AJ 1974 Score 0.600 Correct & Reconstruct →	0.600	0.673 25.74 dB / 0.825	0.576 22.28 dB / 0.702	0.550 21.99 dB / 0.690	✓ 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	DiffDeconv + gradient	0.846	37.04	0.978
2	Restormer+ + gradient	0.839	36.18	0.974
3	ScoreMicro + gradient	0.826	35.51	0.971
4	DeconvFormer + gradient	0.812	34.57	0.965
5	CARE + gradient	0.802	33.47	0.957
6	Restormer + gradient	0.796	33.96	0.961
7	ResUNet + gradient	0.796	33.5	0.957
8	U-Net + gradient	0.785	32.67	0.95
9	PnP-DnCNN + gradient	0.726	29.11	0.902
10	PnP-FISTA + gradient	0.708	27.65	0.873
11	TV-Deconvolution + gradient	0.687	26.6	0.848
12	Richardson-Lucy + gradient	0.673	25.74	0.825
13	Wiener Filter + gradient	0.672	26.27	0.839

Spec Ranges (3 parameters)

Parameter	Min	Max	Unit
incidence_angle	-0.3	0.6	deg
penetration_depth	-20.0	40.0	nm
refractive_index	1.51	1.525

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	DeconvFormer + gradient	0.755	31.98	0.942
2	Restormer+ + gradient	0.754	30.57	0.925
3	DiffDeconv + gradient	0.726	28.86	0.898
4	Restormer + gradient	0.724	29.43	0.908
5	ScoreMicro + gradient	0.716	28.85	0.898
6	PnP-FISTA + gradient	0.676	26.94	0.857
7	TV-Deconvolution + gradient	0.675	26.78	0.853
8	ResUNet + gradient	0.655	26.01	0.832
9	PnP-DnCNN + gradient	0.649	25.66	0.822
10	CARE + gradient	0.628	24.28	0.778
11	U-Net + gradient	0.627	24.76	0.794
12	Wiener Filter + gradient	0.622	23.95	0.767
13	Richardson-Lucy + gradient	0.576	22.28	0.702

Spec Ranges (3 parameters)

Parameter	Min	Max	Unit
incidence_angle	-0.36	0.54	deg
penetration_depth	-24.0	36.0	nm
refractive_index	1.509	1.524

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	DeconvFormer + gradient	0.719	29.54	0.91
2	Restormer+ + gradient	0.689	27.94	0.88
3	DiffDeconv + gradient	0.658	26.56	0.847
4	ScoreMicro + gradient	0.654	25.98	0.831
5	PnP-FISTA + gradient	0.654	25.86	0.828
6	TV-Deconvolution + gradient	0.641	25.44	0.816
7	Restormer + gradient	0.617	24.08	0.771
8	Wiener Filter + gradient	0.587	22.62	0.716
9	CARE + gradient	0.583	22.48	0.71
10	PnP-DnCNN + gradient	0.566	22.59	0.715
11	ResUNet + gradient	0.563	22.05	0.692
12	U-Net + gradient	0.559	22.4	0.707
13	Richardson-Lucy + gradient	0.550	21.99	0.69

Spec Ranges (3 parameters)

Parameter	Min	Max	Unit
incidence_angle	-0.21	0.69	deg
penetration_depth	-14.0	46.0	nm
refractive_index	1.5115	1.5265

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

Total internal reflection fluorescence (TIRF) microscopy selectively excites fluorophores within ~100-200 nm of the coverslip surface using the evanescent field generated when excitation light undergoes total internal reflection at the glass-sample interface. This provides exceptional axial selectivity for imaging membrane-associated events such as vesicle fusion and focal adhesions. The lateral image follows standard widefield PSF convolution but with near-zero out-of-focus background. Primary degradations include non-uniform evanescent field and interference fringes from coherent illumination.

Principle

Total Internal Reflection Fluorescence microscopy creates an evanescent wave that penetrates only ~100-200 nm into the sample when the excitation beam is totally internally reflected at the glass-sample interface. This provides excellent optical sectioning of membrane-proximal events (vesicle fusion, protein dynamics at the plasma membrane) with very low background.

How to Build the System

Use a TIRF-capable objective (60-100x, 1.49 NA oil) on an inverted microscope. Launch the laser at the critical angle through the objective periphery (objective-type TIRF) or through a prism (prism-type TIRF). Verify total internal reflection by observing the evanescent field depth with a calibration sample. Cells must be plated on clean, high-RI coverslips (#1.5H, 170 μm).

Common Reconstruction Algorithms

Single-particle tracking (SPT) algorithms
Multi-angle TIRF for axial sectioning (variable penetration depth)
Denoising (Gaussian filtering, wavelet, or deep-learning)
Photobleaching step analysis for molecular counting
Temporal median filtering for background subtraction

Common Mistakes

Laser angle not precisely at TIR, partially exciting bulk fluorescence
Dirty coverslips causing scattering and destroying evanescent field uniformity
Cells not well-adhered to the coverslip surface, out of evanescent field range
Using objectives with NA < 1.45, insufficient for TIR at aqueous interfaces
Evanescent field depth not calibrated, making quantitative axial analysis unreliable

How to Avoid Mistakes

Fine-tune the TIR angle while observing a known sample; verify exponential depth decay
Clean coverslips rigorously (plasma cleaning or acid wash) before plating cells
Use poly-L-lysine or fibronectin coating to ensure cells adhere to the coverslip
Use 1.49 NA objectives; 1.45 NA is the minimum for aqueous TIR
Calibrate evanescent field depth using fluorescent beads at known axial positions

Forward-Model Mismatch Cases

The widefield fallback illuminates the entire sample depth, but TIRF uses an evanescent wave that penetrates only ~100-200 nm from the coverslip — the fallback includes fluorescence from hundreds of nanometers deeper, adding massive background
The exponential axial intensity decay of the evanescent field (I(z) = I_0 * exp(-z/d), d~100 nm) is not modeled by the widefield fallback — quantitative axial information (membrane proximity) is lost

How to Correct the Mismatch

Use the TIRF operator that models evanescent-wave excitation: only fluorophores within ~200 nm of the glass-sample interface contribute signal, with exponentially decaying excitation intensity
Include the penetration depth d = lambda/(4*pi*sqrt(n1^2*sin^2(theta) - n2^2)) in the forward model; for multi-angle TIRF, model the depth-dependent excitation for each incidence angle

Experimental Setup — Signal Chain

Experimental setup diagram for TIRF 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: Nikon Eclipse Ti2-E TIRF / Olympus cellTIRF-4Line

Objective: Apo TIRF 100x / 1.49 NA oil

Pixel Size Nm: 65

Excitation Source: 488 nm laser (Coherent OBIS, 100 mW)

Evanescent Depth Nm: 100

Exposure Ms: 30

Frame Rate Fps: 33

Detector: Hamamatsu ORCA-Fusion BT sCMOS

Key References

Axelrod, 'Total internal reflection fluorescence microscopy in cell biology', Traffic 2, 764-774 (2001)

Canonical Datasets

Cell Tracking Challenge TIRF sequences
FPbase TIRF imaging examples

Spec DAG — Forward Model Pipeline

C(PSF_TIRF) → D(g, η₃)

C Evanescent-Field PSF (PSF_TIRF)

D EMCCD / sCMOS (g, η₃)

Mismatch Parameters

Symbol	Parameter	Description	Nominal	Perturbed
Δθ_i	incidence_angle	Incidence angle error (deg)	0	0.3
Δd	penetration_depth	Penetration depth error (nm)	0	20
Δn	refractive_index	Coverslip refractive index error	1.515	1.52

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.