SPRM: spatial process and relationship modeling for multiplexed images

Area of Science:

• Spatial biology
• Computational pathology
• Bioinformatics

Background:

• Growing datasets from projects like HuBMAP and Human Cell Atlas present analysis challenges.
• Current spatial proteomics analysis often focuses solely on marker intensity per cell, overlooking other valuable image data.
• There is a need for advanced tools to extract comprehensive information from spatial proteomics data.

Purpose of the Study:

• To introduce SPRM (Spatial Proteomics Research Module), a novel tool for spatial proteomics image analysis.
• To enable richer cell characterization by extracting diverse cell features beyond simple marker intensities.
• To facilitate the analysis of cell spatial distributions and subtypes within tissue images.

Main Methods:

• SPRM calculates image quality metrics, including cell segmentation quality.
• It extracts a wide array of cell features for detailed cell characterization.
• The tool clusters cells into potential cell types and subtypes, comparing them with expert annotations.
• SPRM constructs a cell adjacency matrix to analyze spatial relationships.

Main Results:

• SPRM provides a more comprehensive characterization of cells in spatial proteomics datasets.
• The tool enables the identification and analysis of distinct cell types and subtypes based on extracted features.
• SPRM effectively characterizes cell spatial distributions through an adjacency matrix.
• Example analyses demonstrate the utility of SPRM in exploring complex tissue architectures.

Conclusions:

• SPRM significantly enhances the analysis of spatial proteomics data by providing deeper insights into cell characteristics and spatial organization.
• The open-source availability of SPRM promotes its adoption in research pipelines, including HuBMAP.
• SPRM is a valuable tool for researchers working with large-scale spatial proteomics datasets, enabling advanced exploration of tissue biology.