Abstract

Spatial biology with multiplexed fluorescence imaging allows proteomic characterization of tissues at subcellular resolution. Multi-sample analyses promise large-scale spatially informed tissue atlases, but are held back by variations due to experimental and technical confounders. To address this, we introduce Iterative Gamma Normalization, an unsupervised algorithm that fits a gamma distribution to each marker’s image background and iteratively refines that estimate by excluding likely signal pixels. Each pixel is then scored, on statistically interpretable background-relative positivity, which is comparable across samples. We compared Gamma Normalization to inverse hyperbolic sine followed by minmax scaling and Nimbus on 17 tonsil samples acquired on the MACSima imaging platform. After Gamma normalization, variance between samples was separated along expected and biologically interpretable dimensions, such as B-cell, T-cell, and epithelial markers. Improved biological interpretability after Gamma Normalization was confirmed by Leiden clustering, where clusters matched known cell types and consistently showed contributions from multiple samples. Iterative Gamma Normalization is an unsupervised pre-processing step that achieves cross-sample integration of multiplexed spatial proteomics data while remaining compatible with dedicated downstream batch correction. Code used is here: https://github.com/bbanc/Iterative-Gamma-Normalization.

Links to Paper and Supplementary Materials

Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/MISO_006.pdf

SharedIt Link: Not yet available

SpringerLink (DOI): Not yet available

Supplementary Material: Not Submitted

Link to Open Review

Open Review Page: https://openreview.net/forum?id=ZOPtoQVOou

BibTex

@InProceedings{BanBen_Iterative_MICCAISAT2026,
        author = { Bancher, Benjamin AND Onay, Yagmur AND Praveen, Paurush AND Staubach, Fabian AND Kollet, Jutta},
        title = { { Iterative Gamma Normalization for Inter-Sample Integration of Multiplexed Tissue Images } },
        booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
        year = {2026},
        publisher = {Springer Nature Switzerland},
        volume = {LNCS 17252},
        month = {pending},
        page = {pending}
}


back to top