Cell Spatial Analysis in Crohn's Disease: Unveiling Local Cell Arrangement Pattern with Graph-based Signatures

Area of Science:

• Gastroenterology
• Computational Pathology
• Bioinformatics

Background:

• Crohn's disease (CD) is a chronic inflammatory condition impacting the gastrointestinal tract.
• Assessing CD activity typically relies on histological analysis, focusing on neutrophil density in Hematoxylin and Eosin (H&E) stained tissues.
• Current methods face challenges in capturing comprehensive morphometric and spatial cellular arrangement data.

Purpose of the Study:

• To develop a novel computational approach for characterizing local cell spatial signatures in CD.
• To analyze differences in cellular environments and spatial patterns between active CD patients and controls.
• To investigate the impact of data heterogeneity and multi-institutional data on cell spatial arrangements.

Main Methods:

• Characterization of six distinct cell types from H&E images.
• Development of a 10-cell neighborhood matrix to represent local cell arrangements.
• Application of t-SNE for non-linear spatial projection and Kernel Density Estimation for contour plotting.
• Comparative analysis of spatial patterns between active CD and control groups.

Main Results:

• Identification of heterogeneous nearest-neighbor patterns in cellular arrangements.
• Observed distinct cell clustering tendencies, particularly in the rectal region.
• Demonstrated variations in spatial distribution linked to data heterogeneity and research site.
• Quantified differences in cellular environments associated with CD activity.

Conclusions:

• Novel computational methods reveal distinct spatial cell arrangements in Crohn's disease.
• Cellular microenvironment heterogeneity significantly influences spatial patterns in CD.
• Inter-institutional data disparities necessitate careful consideration in multi-center studies.
• The developed analysis pipeline offers a tool for deeper insights into CD pathogenesis.