Automated multi-scale computational pathotyping (AMSCP) of inflamed synovial tissue

Area of Science:

• Immunology
• Computational Pathology
• Rheumatology

Background:

• Rheumatoid arthritis (RA) is a complex immune-mediated inflammatory disease characterized by inflammatory-erosive arthritis.
• Synovial tissue heterogeneity in RA reveals distinct cellular phenotypes (pauci-immune, diffuse, lymphoid), indicating varied etiologies.
• The need for cost-effective phenotyping tools is crucial for developing targeted therapies in RA research.

Purpose of the Study:

• To develop and validate an automated multi-scale computational pathotyping (AMSCP) pipeline for analyzing human and mouse synovial tissue.
• To enable cost-effective phenotyping of RA for preclinical and clinical applications.
• To demonstrate the pipeline's efficacy, efficiency, and robustness in discovering novel RA phenotypes.

Main Methods:

• Development of an automated multi-scale computational pathotyping (AMSCP) pipeline.
• The pipeline includes two components: tissue type segmentation and cell type classification within tissue compartments.
• Application of the pipeline to both human and mouse synovial tissue samples.

Main Results:

• Demonstrated the efficacy, efficiency, and robustness of the AMSCP pipeline.
• Successfully characterized tissue-level changes and assessed cellular changes across disease states.
• Showcased the pipeline's capability to discover novel RA phenotypes.

Conclusions:

• The AMSCP pipeline provides a valuable, cost-effective method for RA research.
• Facilitates detailed analysis of synovial tissue heterogeneity.
• Supports the development of targeted therapies by enabling precise phenotyping in both preclinical and clinical settings.