Bioluminescence Tomography (BLT)

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
🥇	PhysDiff-BLT PhysDiff-BLT Physics-constrained diffusion for BLT, 2025 38.1 dB SSIM 0.967 Checkpoint unavailable	0.869	38.1	0.967	✓ Certified	Physics-constrained diffusion for BLT, 2025
🥈	ScoreBLT ScoreBLT Score-based BLT with uncertainty, 2024 36.5 dB SSIM 0.952 Checkpoint unavailable	0.834	36.5	0.952	✓ Certified	Score-based BLT with uncertainty, 2024
🥉	BLT-Former BLT-Former Transformer for optical tomography, MICCAI 2023 34.8 dB SSIM 0.929 Checkpoint unavailable	0.794	34.8	0.929	✓ Certified	Transformer for optical tomography, MICCAI 2023
4	DiffusionPINN-BLT DiffusionPINN-BLT Cai et al., Phys. Med. Biol. 68:035005, 2023 32.9 dB SSIM 0.902 Checkpoint unavailable	0.749	32.9	0.902	✓ Certified	Cai et al., Phys. Med. Biol. 68:035005, 2023
5	LISTA-BLT LISTA-BLT Gregor & LeCun ICML 2010; adapted BLT 2020 30.4 dB SSIM 0.864 Checkpoint unavailable	0.689	30.4	0.864	✓ Certified	Gregor & LeCun ICML 2010; adapted BLT 2020
6	BLT-CNN BLT-CNN Gao et al., Sci. Rep. 8:8363, 2018 29.1 dB SSIM 0.838 Checkpoint unavailable	0.654	29.1	0.838	✓ Certified	Gao et al., Sci. Rep. 8:8363, 2018
7	PnP-ADMM (BLT) PnP-ADMM (BLT) Venkatakrishnan et al., IEEE GlobalSIP 2013 25.6 dB SSIM 0.730 Try in SpecLab →	0.542	25.6	0.730	✓ Certified	Venkatakrishnan et al., IEEE GlobalSIP 2013
8	Tikhonov-PR Tikhonov-PR Han et al., Opt. Express 14(8):3673, 2006 22.8 dB SSIM 0.640 Try in SpecLab →	0.450	22.8	0.640	✓ Certified	Han et al., Opt. Express 14(8):3673, 2006
9	Tikhonov-BLT Tikhonov-BLT Lv et al., Phys. Med. Biol. 51:1479, 2006 19.5 dB SSIM 0.540 Try in SpecLab →	0.345	19.5	0.540	✓ Certified	Lv et al., Phys. Med. Biol. 51:1479, 2006

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
🥇	PhysDiff-BLT + gradient PhysDiff-BLT + gradient Physics-constrained diffusion for BLT, 2025 Score 0.750 Correct & Reconstruct →	0.750	0.825 35.78 dB / 0.972	0.735 29.41 dB / 0.907	0.689 27.93 dB / 0.879	✓ Certified	Physics-constrained diffusion for BLT, 2025
🥈	BLT-Former + gradient BLT-Former + gradient Transformer for optical tomography, MICCAI 2023 Score 0.717 Correct & Reconstruct →	0.717	0.782 32.76 dB / 0.950	0.708 28.99 dB / 0.900	0.660 26.72 dB / 0.851	✓ Certified	Transformer for optical tomography, MICCAI 2023
🥉	DiffusionPINN-BLT + gradient DiffusionPINN-BLT + gradient Cai et al., Phys. Med. Biol. 68:035005, 2023 Score 0.697 Correct & Reconstruct →	0.697	0.755 31.06 dB / 0.932	0.681 27.0 dB / 0.858	0.655 25.92 dB / 0.830	✓ Certified	Cai et al., Phys. Med. Biol. 68:035005, 2023
4	ScoreBLT + gradient ScoreBLT + gradient Score-based BLT with uncertainty, 2024 Score 0.695 Correct & Reconstruct →	0.695	0.826 35.41 dB / 0.970	0.660 26.46 dB / 0.844	0.599 23.81 dB / 0.762	✓ Certified	Score-based BLT with uncertainty, 2024
5	PnP-ADMM (BLT) + gradient PnP-ADMM (BLT) + gradient Venkatakrishnan et al., IEEE GlobalSIP 2013 Score 0.569 Correct & Reconstruct →	0.569	0.638 24.19 dB / 0.775	0.551 21.24 dB / 0.657	0.518 20.07 dB / 0.602	✓ Certified	Venkatakrishnan et al., IEEE GlobalSIP 2013
6	BLT-CNN + gradient BLT-CNN + gradient Gao et al., Sci. Rep. 8:8363, 2018 Score 0.559 Correct & Reconstruct →	0.559	0.713 27.74 dB / 0.875	0.526 20.98 dB / 0.645	0.439 17.64 dB / 0.482	✓ Certified	Gao et al., Sci. Rep. 8:8363, 2018
7	LISTA-BLT + gradient LISTA-BLT + gradient Gregor & LeCun, ICML 2010; adapted BLT 2020 Score 0.544 Correct & Reconstruct →	0.544	0.740 29.19 dB / 0.904	0.487 19.65 dB / 0.582	0.405 16.5 dB / 0.426	✓ Certified	Gregor & LeCun, ICML 2010; adapted BLT 2020
8	Tikhonov-PR + gradient Tikhonov-PR + gradient Han et al., Opt. Express 14(8):3673, 2006 Score 0.484 Correct & Reconstruct →	0.484	0.521 20.06 dB / 0.602	0.482 19.56 dB / 0.577	0.450 18.61 dB / 0.530	✓ Certified	Han et al., Opt. Express 14(8):3673, 2006
9	Tikhonov-BLT + gradient Tikhonov-BLT + gradient Lv et al., Phys. Med. Biol. 51:1479, 2006 Score 0.379 Correct & Reconstruct →	0.379	0.423 16.7 dB / 0.435	0.376 15.84 dB / 0.394	0.337 14.04 dB / 0.312	✓ Certified	Lv et al., Phys. Med. Biol. 51:1479, 2006

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	ScoreBLT + gradient	0.826	35.41	0.97
2	PhysDiff-BLT + gradient	0.825	35.78	0.972
3	BLT-Former + gradient	0.782	32.76	0.95
4	DiffusionPINN-BLT + gradient	0.755	31.06	0.932
5	LISTA-BLT + gradient	0.740	29.19	0.904
6	BLT-CNN + gradient	0.713	27.74	0.875
7	PnP-ADMM (BLT) + gradient	0.638	24.19	0.775
8	Tikhonov-PR + gradient	0.521	20.06	0.602
9	Tikhonov-BLT + gradient	0.423	16.7	0.435

Spec Ranges (3 parameters)

Parameter	Min	Max	Unit
optical_property_error_(mu_a,_mu_s')	-4.0	8.0	relative
source_depth_ambiguity	-1.0	2.0	mm
autofluorescence_background	-6.0	12.0	-

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	PhysDiff-BLT + gradient	0.735	29.41	0.907
2	BLT-Former + gradient	0.708	28.99	0.9
3	DiffusionPINN-BLT + gradient	0.681	27.0	0.858
4	ScoreBLT + gradient	0.660	26.46	0.844
5	PnP-ADMM (BLT) + gradient	0.551	21.24	0.657
6	BLT-CNN + gradient	0.526	20.98	0.645
7	LISTA-BLT + gradient	0.487	19.65	0.582
8	Tikhonov-PR + gradient	0.482	19.56	0.577
9	Tikhonov-BLT + gradient	0.376	15.84	0.394

Spec Ranges (3 parameters)

Parameter	Min	Max	Unit
optical_property_error_(mu_a,_mu_s')	-4.8	7.2	relative
source_depth_ambiguity	-1.2	1.8	mm
autofluorescence_background	-7.2	10.8	-

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	PhysDiff-BLT + gradient	0.689	27.93	0.879
2	BLT-Former + gradient	0.660	26.72	0.851
3	DiffusionPINN-BLT + gradient	0.655	25.92	0.83
4	ScoreBLT + gradient	0.599	23.81	0.762
5	PnP-ADMM (BLT) + gradient	0.518	20.07	0.602
6	Tikhonov-PR + gradient	0.450	18.61	0.53
7	BLT-CNN + gradient	0.439	17.64	0.482
8	LISTA-BLT + gradient	0.405	16.5	0.426
9	Tikhonov-BLT + gradient	0.337	14.04	0.312

Spec Ranges (3 parameters)

Parameter	Min	Max	Unit
optical_property_error_(mu_a,_mu_s')	-2.8	9.2	relative
source_depth_ambiguity	-0.7	2.3	mm
autofluorescence_background	-4.2	13.8	-

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

Src → R → P → D

Src Source

R Rotation

P Propagation

D Detector

Mismatch Parameters

Symbol	Parameter	Description	Perturbed
o_p	optical_property_error_(mu_a,_mu_s')	Optical property error (mu_a, mu_s') (relative)	4.0
s_d	source_depth_ambiguity	Source depth ambiguity (mm)	1.0
a_b	autofluorescence_background	Autofluorescence background (-)	6.0

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.