Public

DESI Mass Spectrometry Imaging — Public Tier

(5 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
spray_angle_error	-1.0 – 2.0	0.5	deg
solvent_flow_variation	-3.0 – 6.0	1.5	-
ion_suppression_(matrix_effect)	-10.0 – 20.0	5.0	-
spatial_resolution_degradation	-10.0 – 20.0	5.0	-

Method: CPU_baseline — Mismatch parameter: nominal

20.86 dB

SSIM 0.5556

17.94 dB

SSIM 0.2739

21.09 dB

SSIM 0.4477

Per-scene breakdown (4 scenes)

Scene	PSNR I	SSIM I	PSNR II	SSIM II	PSNR III	SSIM III
scene_00	22.20	0.5293	20.35	0.2280	21.93	0.3469
scene_01	22.95	0.6386	19.17	0.3063	21.54	0.4791
scene_02	17.62	0.5116	14.75	0.2820	20.16	0.5023
scene_03	20.65	0.5429	17.50	0.2793	20.72	0.4626

#	Method	Score	PSNR	SSIM	Consistency	Trust	Source
1	DiffusionMSI + gradient	0.824	35.52	0.971	0.9	✓ Certified	Palmer et al., Nat. Methods 2024
2	MSIFormer + gradient	0.799	34.04	0.961	0.87	✓ Certified	Kalinichenko et al., Nat. Methods 2023
3	SpaMSI-Net + gradient	0.777	31.84	0.941	0.91	✓ Certified	Rappez et al., Nat. Methods 2021
4	DeepMSI + gradient	0.769	31.06	0.932	0.93	✓ Certified	Gruber et al., Anal. Chem. 2021
5	MSI-GAN + gradient	0.764	31.39	0.936	0.88	✓ Certified	Yang et al., Anal. Chem. 2021
6	MSI-TV + gradient	0.709	27.51	0.87	0.93	✓ Certified	Fonville et al., Bioinformatics 2012
7	MSI-NMF + gradient	0.659	25.09	0.805	0.93	✓ Certified	Blanco et al., Anal. Chem. 2013
8	MSI-PCA + gradient	0.585	22.48	0.71	0.88	✓ Certified	Alexandrov et al., J. Bioinform. Comput. Biol. 2010
9	MSI-Hotelling + gradient	0.542	20.44	0.62	0.95	✓ Certified	Deininger et al., Proteomics 2011

y H_ideal spec_ranges x_true true_spec

Format: HDF5

Scenes: 5

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

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