Deep cell phenotyping and spatial analysis of multiplexed imaging with TRACERx-PHLEX

Area of Science:

• Computational Biology
• Bioinformatics
• Digital Pathology

Background:

• Multiplexed imaging generates large, high-dimensional datasets.
• Existing computational pipelines can be complex and require specialized expertise.
• There is a need for reproducible, user-friendly tools for analyzing highly multiplexed data.

Purpose of the Study:

• To develop and validate TRACERx-PHLEX, an automated computational pipeline for multiplexed imaging data analysis.
• To provide an end-to-end solution for cell segmentation, cell-type annotation, and spatial analysis.
• To enable clinically relevant insights from complex imaging datasets.

Main Methods:

• TRACERx-PHLEX integrates deep learning-based cell segmentation (deep-imcyto), automated cell-type annotation (TYPEx), and interpretable spatial analysis (Spatial-PHLEX).
• The pipeline is implemented as an automated and containerised Nextflow workflow.
• Validation was performed using Imaging Mass Cytometry (IMC), Co-detection by indexing (CODEX), and orthogonal data across diverse experimental conditions.

Main Results:

• PHLEX generates single-cell identities, cell densities, marker positivity, and spatial metrics.
• The pipeline demonstrated robust performance across various tissue types, fixation conditions, image sizes, and antibody panels.
• Benchmarking confirmed its competitiveness against state-of-the-art methods.

Conclusions:

• TRACERx-PHLEX offers a comprehensive, automated, and interoperable solution for analyzing highly multiplexed imaging data.
• The pipeline simplifies complex analyses, making them accessible without manual assessment or pathology expertise.
• PHLEX facilitates the generation of clinically relevant insights from large-scale spatial biology studies.