Resolving phenotyping discordance with SPACEMAP, an integrated machine learning framework.

Area of Science:

• Computational Biology
• Bioinformatics
• Digital Pathology

Background:

• Multiplex imaging generates complex cellular data within tissue microenvironments.
• Analyzing this data requires a unified analytical framework for biological insights.

Purpose of the Study:

• To develop SPACEMAP, a comprehensive platform for multiplex imaging analysis.
• To integrate image registration, segmentation, artifact removal, and phenotyping into a single system.
• To provide high-fidelity spatial cell phenotyping and classification.

Main Methods:

• SPACEMAP is a Python and Qupath-based platform.
• It integrates image registration, segmentation, artifact removal, tissue/zone classification, and spatial feature extraction.
• Two workflows are used: a machine learning model and a consensus classifier.

Main Results:

• SPACEMAP was benchmarked against Leiden clustering, Self-Organizing Maps, and SCIMAP, revealing substantial disagreement among existing methods.
• SPACEMAP's machine learning and consensus classifier workflows demonstrated robust performance.
• Validation was performed on in-house colorectal cancer samples and a public dataset.

Conclusions:

• SPACEMAP provides a robust and unified analytical framework for multiplex imaging data.
• It overcomes limitations of existing methods for spatial cell phenotyping and classification.
• SPACEMAP enhances the extraction of meaningful biological insights from complex tissue microenvironments.