Scanning Acoustic Microscopy (SAM)

Scanning Acoustic Microscopy (SAM)

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
🥇 DiffusionSAM 0.807 35.0 0.948 ✓ Certified Score-based diffusion for SAM, 2024
🥈 AcousticFormer 0.784 34.0 0.935 ✓ Certified Zhu et al., Ultrasonics 2024
🥉 PINN-SAM 0.749 32.5 0.915 ✓ Certified Guo et al., IEEE UFFC 2024
4 Self-Sup Deconv 0.712 31.0 0.890 ✓ Certified He et al., IEEE Trans. Instrum. Meas. 2024
5 SAM-Net 0.672 29.5 0.860 ✓ Certified Guo et al., Ultrasonics 2022
6 PnP-ADMM 0.577 26.5 0.770 ✓ Certified Venkatakrishnan et al., IEEE GlobalSIP 2013
7 Wiener Deconv 0.458 23.0 0.650 ✓ Certified Zinin et al., J. Appl. Phys. 1997
8 SAFT 0.408 21.5 0.600 ✓ Certified Schickert et al., NDT&E Int. 2003

Dataset: PWM Benchmark (8 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
🥇 AcousticFormer + gradient 0.704
0.792
32.62 dB / 0.949
0.708
28.64 dB / 0.894
0.612
24.62 dB / 0.790
✓ Certified Zhu et al., Ultrasonics 138:107212, 2024
🥈 PINN-SAM + gradient 0.676
0.750
30.7 dB / 0.927
0.662
26.25 dB / 0.839
0.615
24.58 dB / 0.789
✓ Certified Guo et al., IEEE UFFC 71:340, 2024
🥉 DiffusionSAM + gradient 0.670
0.782
32.23 dB / 0.945
0.655
25.91 dB / 0.829
0.574
22.21 dB / 0.699
✓ Certified Score-based diffusion for SAM reconstruction, 2024
4 SAM-Net + gradient 0.593
0.688
26.64 dB / 0.849
0.591
22.95 dB / 0.729
0.501
20.34 dB / 0.615
✓ Certified Guo et al., Ultrasonics 122:106679, 2022
5 PnP-ADMM + gradient 0.587
0.629
24.15 dB / 0.774
0.590
23.08 dB / 0.734
0.541
21.01 dB / 0.646
✓ Certified Venkatakrishnan et al., IEEE GlobalSIP 2013
6 Self-Sup Deconv + gradient 0.567
0.725
29.18 dB / 0.903
0.549
21.38 dB / 0.663
0.426
17.87 dB / 0.494
✓ Certified He et al., IEEE Trans. Instrum. Meas. 73, 2024
7 Wiener Deconv + gradient 0.493
0.525
20.12 dB / 0.604
0.504
20.09 dB / 0.603
0.451
18.7 dB / 0.535
✓ Certified Zinin et al., J. Appl. Phys. 1997
8 SAFT + gradient 0.488
0.501
19.64 dB / 0.581
0.511
20.14 dB / 0.605
0.451
18.7 dB / 0.535
✓ Certified Schickert et al., NDT&E Int. 36:339, 2003

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 AcousticFormer + gradient 0.792 32.62 0.949
2 DiffusionSAM + gradient 0.782 32.23 0.945
3 PINN-SAM + gradient 0.750 30.7 0.927
4 Self-Sup Deconv + gradient 0.725 29.18 0.903
5 SAM-Net + gradient 0.688 26.64 0.849
6 PnP-ADMM + gradient 0.629 24.15 0.774
7 Wiener Deconv + gradient 0.525 20.12 0.604
8 SAFT + gradient 0.501 19.64 0.581
Spec Ranges (4 parameters)
Parameter Min Max Unit
coupling_medium_speed 1466.0 1508.0 m/s
focus_depth_error -4.0 8.0 um
lens_aberration -0.04 0.08 waves
gate_position_error -1.0 2.0 -
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 AcousticFormer + gradient 0.708 28.64 0.894
2 PINN-SAM + gradient 0.662 26.25 0.839
3 DiffusionSAM + gradient 0.655 25.91 0.829
4 SAM-Net + gradient 0.591 22.95 0.729
5 PnP-ADMM + gradient 0.590 23.08 0.734
6 Self-Sup Deconv + gradient 0.549 21.38 0.663
7 SAFT + gradient 0.511 20.14 0.605
8 Wiener Deconv + gradient 0.504 20.09 0.603
Spec Ranges (4 parameters)
Parameter Min Max Unit
coupling_medium_speed 1463.2 1505.2 m/s
focus_depth_error -4.8 7.2 um
lens_aberration -0.048 0.072 waves
gate_position_error -1.2 1.8 -
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 PINN-SAM + gradient 0.615 24.58 0.789
2 AcousticFormer + gradient 0.612 24.62 0.79
3 DiffusionSAM + gradient 0.574 22.21 0.699
4 PnP-ADMM + gradient 0.541 21.01 0.646
5 SAM-Net + gradient 0.501 20.34 0.615
6 Wiener Deconv + gradient 0.451 18.7 0.535
7 SAFT + gradient 0.451 18.7 0.535
8 Self-Sup Deconv + gradient 0.426 17.87 0.494
Spec Ranges (4 parameters)
Parameter Min Max Unit
coupling_medium_speed 1470.2 1512.2 m/s
focus_depth_error -2.8 9.2 um
lens_aberration -0.028 0.092 waves
gate_position_error -0.7 2.3 -
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_m coupling_medium_speed Coupling medium speed (m/s) 1480.0 1494.0
f_d focus_depth_error Focus depth error (um) 0.0 4.0
l_a lens_aberration Lens aberration (waves) 0.0 0.04
g_p gate_position_error Gate position error (-) 0.0 1.0

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.