SpaceANOVA: Spatial co-occurrence analysis of cell types in multiplex imaging data using point process and functional...

Area of Science:

• Computational pathology
• Spatial biology
• Statistical modeling

Background:

• Multiplex imaging reveals cell organization in tumor microenvironments (TME).
• Understanding cell co-occurrence across diseases offers pathological insights.
• Existing methods lack generalizability and rely on strict assumptions.

Approach:

• Developed a novel statistical method based on Poisson point process (PPP) and functional analysis of variance (FANOVA).
• Method accommodates multiple images per subject and handles missing tissue regions.
• Demonstrated superior statistical power and robustness via simulations.

Key Points:

• Analyzes differential spatial co-occurrence of cell types across multiple tissue/disease groups.
• Addresses challenges of missing data and multiple images per subject.
• Reveals novel spatial insights in colorectal cancer precursor lesions.

Conclusions:

• The new method provides a powerful and generalizable approach for spatial cell co-occurrence analysis.
• Enhances understanding of tissue microenvironments and disease pathology.
• Facilitates the development of targeted intervention strategies.