Public

OCTA — Public Tier

(3 scenes)

Full-access development tier with all data visible.

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.

Parameter Specifications

✓

True spec visible — use these exact values for Scenario III oracle reconstruction.

Parameter Spec Range True Value Unit
inter_bscan_time -0.5 – 1.0 0.25 ms
bulk_motion -0.2 – 0.4 0.1 mm/s
decorrelation_threshold 0.45 – 0.6 0.525

InverseNet Baseline Scores

Method: CPU_baseline — Mismatch parameter: nominal

Scenario I (Ideal)

21.43 dB

SSIM 0.2887

Scenario II (Mismatch)

16.84 dB

SSIM 0.0705

Scenario III (Oracle)

18.70 dB

SSIM 0.1535

Per-scene breakdown (4 scenes)
Scene PSNR I SSIM I PSNR II SSIM II PSNR III SSIM III
scene_00 21.77 0.2816 17.22 0.0679 18.28 0.1283
scene_01 20.94 0.2832 16.98 0.0767 19.75 0.1809
scene_02 21.61 0.3160 16.92 0.0784 18.77 0.1851
scene_03 21.40 0.2739 16.23 0.0591 17.99 0.1197

Public Tier Leaderboard

# Method Score PSNR SSIM Consistency Trust Source
1 DiffusionOCT + gradient 0.838 36.2 0.974 0.93 ✓ Certified Zhang et al., NeurIPS 2024
2 SpeckleFormer + gradient 0.831 35.81 0.972 0.92 ✓ Certified Devalla et al., ECCV 2024
3 ScoreOCT + gradient 0.820 35.19 0.969 0.9 ✓ Certified Wei et al., ECCV 2025
4 OCTA-Net + gradient 0.803 33.6 0.958 0.92 ✓ Certified Hybrid U-Net+Transformer, 2023
5 RetinalFormer + gradient 0.802 33.8 0.959 0.9 ✓ Certified Chen et al., ICCV 2024
6 OCT-ViT + gradient 0.799 33.73 0.959 0.89 ✓ Certified Tian et al., ICCV 2024
7 U-Net-OCT + gradient 0.789 32.45 0.947 0.93 ✓ Certified Ronneberger et al., MICCAI 2015 (OCT variant)
8 Speckle-DenoiseNet + gradient 0.758 31.23 0.934 0.86 ✓ Certified Devalla et al., BOE 2019
9 BM4D + gradient 0.715 27.61 0.872 0.95 ✓ Certified Maggioni et al., IEEE TIP 2013
10 NLM-OCT + gradient 0.708 28.11 0.883 0.87 ✓ Certified Buades et al., Multiscale Model. Simul. 2005
11 TV-Denoising + gradient 0.673 26.22 0.838 0.88 ✓ Certified Rudin et al., Phys. A 1992
12 Speckle-Lee + gradient 0.654 25.12 0.806 0.9 ✓ Certified Lee, IEEE TGRS 1980
13 FFT-OCT + gradient 0.612 23.63 0.755 0.87 ✓ Certified Analytical baseline

Visible Data Fields

y H_ideal spec_ranges x_true true_spec

Dataset

Format: HDF5
Scenes: 3

Scoring Formula

0.4 × PSNR_norm + 0.4 × SSIM + 0.2 × (1 − ‖y − Ĥx̂‖/‖y‖)

PSNR: 40% SSIM: 40% Consistency: 20%
Back to OCTA