Intravascular Ultrasound (IVUS)

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
🥇	ScoreUS ScoreUS Johnson et al., ECCV 2025 36.28 dB SSIM 0.962 Checkpoint unavailable	0.836	36.28	0.962	✓ Certified	Johnson et al., ECCV 2025
🥈	DiffUS DiffUS Chen et al., NeurIPS 2024 35.95 dB SSIM 0.958 Checkpoint unavailable	0.828	35.95	0.958	✓ Certified	Chen et al., NeurIPS 2024
🥉	AttentionBeam AttentionBeam Xu et al., ECCV 2024 35.52 dB SSIM 0.952 Checkpoint unavailable	0.818	35.52	0.952	✓ Certified	Xu et al., ECCV 2024
4	BeamDATA BeamDATA Smith et al., ICCV 2024 35.32 dB SSIM 0.951 Checkpoint unavailable	0.814	35.32	0.951	✓ Certified	Smith et al., ICCV 2024
5	BeamFormer BeamFormer Li et al., ICCV 2024 35.15 dB SSIM 0.948 Checkpoint unavailable	0.810	35.15	0.948	✓ Certified	Li et al., ICCV 2024
6	UltrasoundFormer UltrasoundFormer Park et al., CVPR 2024 34.85 dB SSIM 0.945 Checkpoint unavailable	0.803	34.85	0.945	✓ Certified	Park et al., CVPR 2024
7	Phase-ADMM-Net Phase-ADMM-Net Hou et al., IEEE TMI 2022 33.95 dB SSIM 0.940 Checkpoint unavailable	0.786	33.95	0.940	✓ Certified	Hou et al., IEEE TMI 2022
8	PnP-TV PnP-TV TV regularization for ultrasound 33.1 dB SSIM 0.953 Try in SpecLab →	0.778	33.1	0.953	✓ Certified	TV regularization for ultrasound
9	MU-Net MU-Net Hyun et al., IEEE TUFFC 2022 33.2 dB SSIM 0.928 Checkpoint unavailable	0.767	33.2	0.928	✓ Certified	Hyun et al., IEEE TUFFC 2022
10	ABLE ABLE Luijten et al., IEEE TMI 2020 31.85 dB SSIM 0.905 Checkpoint unavailable	0.733	31.85	0.905	✓ Certified	Luijten et al., IEEE TMI 2020
11	PnP-ADMM PnP-ADMM Goudarzi et al., 2020 28.12 dB SSIM 0.810 Try in SpecLab →	0.624	28.12	0.810	✓ Certified	Goudarzi et al., 2020
12	PW-DAS PW-DAS Plane wave synthesis 26.15 dB SSIM 0.735 Try in SpecLab →	0.553	26.15	0.735	✓ Certified	Plane wave synthesis
13	DAS-CF DAS-CF Capon filter variant 25.8 dB SSIM 0.720 Try in SpecLab →	0.540	25.8	0.720	✓ Certified	Capon filter variant
14	DAS DAS Analytical baseline 24.5 dB SSIM 0.680 Try in SpecLab →	0.498	24.5	0.680	✓ Certified	Analytical baseline

Dataset: PWM Benchmark (14 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
🥇	ScoreUS + gradient ScoreUS + gradient Johnson et al., ECCV 2025 Score 0.738 Correct & Reconstruct →	0.738	0.798 33.34 dB / 0.956	0.727 29.63 dB / 0.911	0.690 27.65 dB / 0.873	✓ Certified	Johnson et al., ECCV 2025
🥈	UltrasoundFormer + gradient UltrasoundFormer + gradient Park et al., CVPR 2024 Score 0.736 Correct & Reconstruct →	0.736	0.806 33.64 dB / 0.958	0.730 30.18 dB / 0.920	0.673 26.28 dB / 0.840	✓ Certified	Park et al., CVPR 2024
🥉	DiffUS + gradient DiffUS + gradient Chen et al., NeurIPS 2024 Score 0.736 Correct & Reconstruct →	0.736	0.799 34.06 dB / 0.961	0.725 29.29 dB / 0.905	0.684 27.56 dB / 0.871	✓ Certified	Chen et al., NeurIPS 2024
4	BeamFormer + gradient BeamFormer + gradient Li et al., ICCV 2024 Score 0.723 Correct & Reconstruct →	0.723	0.783 32.45 dB / 0.947	0.734 29.36 dB / 0.907	0.652 26.31 dB / 0.841	✓ Certified	Li et al., ICCV 2024
5	AttentionBeam + gradient AttentionBeam + gradient Xu et al., ECCV 2024 Score 0.721 Correct & Reconstruct →	0.721	0.792 33.49 dB / 0.957	0.709 28.27 dB / 0.886	0.661 26.68 dB / 0.850	✓ Certified	Xu et al., ECCV 2024
6	BeamDATA + gradient BeamDATA + gradient Smith et al., ICCV 2024 Score 0.719 Correct & Reconstruct →	0.719	0.789 33.25 dB / 0.955	0.709 28.47 dB / 0.890	0.660 26.16 dB / 0.836	✓ Certified	Smith et al., ICCV 2024
7	Phase-ADMM-Net + gradient Phase-ADMM-Net + gradient Hou et al., IEEE TMI 2022 Score 0.707 Correct & Reconstruct →	0.707	0.790 32.33 dB / 0.946	0.680 26.45 dB / 0.844	0.651 25.94 dB / 0.830	✓ Certified	Hou et al., IEEE TMI 2022
8	PnP-TV + gradient PnP-TV + gradient TV regularization for ultrasound Score 0.662 Correct & Reconstruct →	0.662	0.753 30.3 dB / 0.921	0.651 25.19 dB / 0.808	0.583 22.77 dB / 0.722	✓ Certified	TV regularization for ultrasound
9	PnP-ADMM + gradient PnP-ADMM + gradient Goudarzi et al., 2020 Score 0.618 Correct & Reconstruct →	0.618	0.697 26.98 dB / 0.858	0.602 23.78 dB / 0.761	0.556 22.37 dB / 0.706	✓ Certified	Goudarzi et al., 2020
10	MU-Net + gradient MU-Net + gradient Hyun et al., IEEE TUFFC 2022 Score 0.613 Correct & Reconstruct →	0.613	0.756 30.69 dB / 0.927	0.574 22.82 dB / 0.724	0.508 19.89 dB / 0.593	✓ Certified	Hyun et al., IEEE TUFFC 2022
11	DAS-CF + gradient DAS-CF + gradient Capon filter, IEEE 1969 Score 0.610 Correct & Reconstruct →	0.610	0.614 23.72 dB / 0.758	0.610 23.94 dB / 0.766	0.607 23.83 dB / 0.762	✓ Certified	Capon filter, IEEE 1969
12	ABLE + gradient ABLE + gradient Luijten et al., IEEE TMI 2020 Score 0.591 Correct & Reconstruct →	0.591	0.762 30.55 dB / 0.925	0.547 21.02 dB / 0.647	0.463 18.31 dB / 0.515	✓ Certified	Luijten et al., IEEE TMI 2020
13	PW-DAS + gradient PW-DAS + gradient Plane wave synthesis baseline Score 0.578 Correct & Reconstruct →	0.578	0.622 23.95 dB / 0.767	0.568 21.77 dB / 0.680	0.545 21.59 dB / 0.672	✓ Certified	Plane wave synthesis baseline
14	DAS + gradient DAS + gradient Analytical baseline Score 0.530 Correct & Reconstruct →	0.530	0.567 21.61 dB / 0.673	0.534 21.2 dB / 0.655	0.488 19.76 dB / 0.587	✓ Certified	Analytical baseline

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 3 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	UltrasoundFormer + gradient	0.806	33.64	0.958
2	DiffUS + gradient	0.799	34.06	0.961
3	ScoreUS + gradient	0.798	33.34	0.956
4	AttentionBeam + gradient	0.792	33.49	0.957
5	Phase-ADMM-Net + gradient	0.790	32.33	0.946
6	BeamDATA + gradient	0.789	33.25	0.955
7	BeamFormer + gradient	0.783	32.45	0.947
8	ABLE + gradient	0.762	30.55	0.925
9	MU-Net + gradient	0.756	30.69	0.927
10	PnP-TV + gradient	0.753	30.3	0.921
11	PnP-ADMM + gradient	0.697	26.98	0.858
12	PW-DAS + gradient	0.622	23.95	0.767
13	DAS-CF + gradient	0.614	23.72	0.758
14	DAS + gradient	0.567	21.61	0.673

Spec Ranges (3 parameters)

Parameter	Min	Max	Unit
catheter_rotation_non_uniformity	-2.0	4.0	-
ring_down_artifact	-4.0	8.0	-
sound_speed_in_plaque	1508.0	1604.0	m/s

View Details →

Dev 3 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	BeamFormer + gradient	0.734	29.36	0.907
2	UltrasoundFormer + gradient	0.730	30.18	0.92
3	ScoreUS + gradient	0.727	29.63	0.911
4	DiffUS + gradient	0.725	29.29	0.905
5	AttentionBeam + gradient	0.709	28.27	0.886
6	BeamDATA + gradient	0.709	28.47	0.89
7	Phase-ADMM-Net + gradient	0.680	26.45	0.844
8	PnP-TV + gradient	0.651	25.19	0.808
9	DAS-CF + gradient	0.610	23.94	0.766
10	PnP-ADMM + gradient	0.602	23.78	0.761
11	MU-Net + gradient	0.574	22.82	0.724
12	PW-DAS + gradient	0.568	21.77	0.68
13	ABLE + gradient	0.547	21.02	0.647
14	DAS + gradient	0.534	21.2	0.655

Spec Ranges (3 parameters)

Parameter	Min	Max	Unit
catheter_rotation_non_uniformity	-2.4	3.6	-
ring_down_artifact	-4.8	7.2	-
sound_speed_in_plaque	1501.6	1597.6	m/s

Submit Reconstruction View Details →

Hidden 3 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	ScoreUS + gradient	0.690	27.65	0.873
2	DiffUS + gradient	0.684	27.56	0.871
3	UltrasoundFormer + gradient	0.673	26.28	0.84
4	AttentionBeam + gradient	0.661	26.68	0.85
5	BeamDATA + gradient	0.660	26.16	0.836
6	BeamFormer + gradient	0.652	26.31	0.841
7	Phase-ADMM-Net + gradient	0.651	25.94	0.83
8	DAS-CF + gradient	0.607	23.83	0.762
9	PnP-TV + gradient	0.583	22.77	0.722
10	PnP-ADMM + gradient	0.556	22.37	0.706
11	PW-DAS + gradient	0.545	21.59	0.672
12	MU-Net + gradient	0.508	19.89	0.593
13	DAS + gradient	0.488	19.76	0.587
14	ABLE + gradient	0.463	18.31	0.515

Spec Ranges (3 parameters)

Parameter	Min	Max	Unit
catheter_rotation_non_uniformity	-1.4	4.6	-
ring_down_artifact	-2.8	9.2	-
sound_speed_in_plaque	1517.6	1613.6	m/s

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

P → D

P Propagation

D Detector

Mismatch Parameters

Symbol	Parameter	Description	Nominal	Perturbed
c_r	catheter_rotation_non_uniformity	Catheter rotation non-uniformity (-)	0.0	2.0
r_a	ring_down_artifact	Ring-down artifact (-)	0.0	4.0
s_s	sound_speed_in_plaque	Sound speed in plaque (m/s)	1540.0	1572.0

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.