UniFORM: Towards Universal ImmunoFluorescence Normalization for Multiplex Tissue Imaging.

Area of Science:

• Computational Biology
• Biotechnology
• Bioinformatics

Background:

• Multiplexed tissue imaging (MTI) provides high-dimensional spatial analysis of tumor microenvironments.
• Technical variability in staining intensities poses challenges for MTI data analysis.
• Existing normalization methods struggle with heterogeneous, right-skewed MTI data distributions.

Purpose of the Study:

• To develop a robust and accurate normalization pipeline for multiplexed tissue imaging (MTI) data.
• To address limitations of current methods in handling technical variability and skewed data distributions.
• To improve signal alignment and downstream analysis of MTI data.

Main Methods:

• Introduction of UniFORM, a non-parametric, Python-based normalization pipeline for MTI data.
• UniFORM preserves marker intensity distribution shapes and positive population proportions.
• Automated normalization with an optional guided fine-tuning for complex datasets.

Main Results:

• UniFORM demonstrated superior performance in mitigating batch effects across different MTI platforms and datasets.
• Improved marker distribution alignment and enhanced kBET scores were observed.
• UniFORM enhanced downstream analyses, including UMAP visualizations and Leiden clustering.

Conclusions:

• UniFORM offers a scalable and robust solution for MTI data normalization, outperforming existing methods.
• The pipeline enables more accurate and biologically meaningful interpretations of spatial tumor microenvironment data.
• UniFORM is particularly valuable for fluorescence-based MTI platforms but applicable broadly.