Public

SPECT — Public Tier

(3 scenes)

Full-access development tier with all data visible.

Measurements (y), ideal forward operator (H), spec ranges, ground truth (x_true), and true mismatch spec.

Load HDF5 → compare reconstruction vs x_true → check consistency → iterate.

Reconstructed signals (x_hat) and corrected spec as HDF5.

✓

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

Parameter	Spec Range	True Value	Unit
center_offset	-1.5 – 3.0	0.75	pixels
collimator_septal	-0.02 – 0.04	0.01
attenuation	-5.0 – 10.0	2.5	%
scatter	0.15 – 0.3	0.225

Method: CPU_baseline — Mismatch parameter: nominal

28.17 dB

SSIM 0.9192

16.29 dB

SSIM 0.8748

14.39 dB

SSIM 0.0299

Per-scene breakdown (4 scenes)

Scene	PSNR I	SSIM I	PSNR II	SSIM II	PSNR III	SSIM III
scene_00	29.94	0.9533	16.53	0.8739	14.34	0.0264
scene_01	25.10	0.8560	14.64	0.8537	13.86	0.0301
scene_02	27.61	0.9222	17.32	0.8981	14.87	0.0343
scene_03	30.01	0.9453	16.69	0.8736	14.48	0.0286

#	Method	Score	PSNR	SSIM	Consistency	Trust	Source
1	PET-ViT + gradient	0.844	36.67	0.977	0.93	✓ Certified	Smith et al., ICCV 2024
2	PETFormer + gradient	0.818	34.92	0.967	0.91	✓ Certified	Li et al., ECCV 2024
3	TransEM + gradient	0.786	31.96	0.942	0.95	✓ Certified	Xie et al., 2023
4	DeepPET + gradient	0.770	31.09	0.932	0.93	✓ Certified	Haggstrom et al., MIA 2019
5	U-Net-PET + gradient	0.768	31.53	0.937	0.89	✓ Certified	Ronneberger et al. variant, MICCAI 2020
6	FBP-PET + gradient	0.700	27.23	0.864	0.91	✓ Certified	Analytical baseline
7	ML-EM + gradient	0.692	27.14	0.862	0.88	✓ Certified	Shepp & Vardi, IEEE TPAMI 1982
8	OS-EM + gradient	0.692	26.65	0.849	0.93	✓ Certified	Hudson & Larkin, IEEE TMI 1994
9	MAPEM-RDP + gradient	0.669	25.71	0.824	0.91	✓ Certified	Nuyts et al., IEEE TMI 2002
10	OSEM + gradient	0.578	22.08	0.693	0.9	✓ Certified	Hudson & Larkin, IEEE TMI 1994

y H_ideal spec_ranges x_true true_spec

Format: HDF5

Scenes: 3

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

PSNR: 40% SSIM: 40% Consistency: 20%