Description:
- End-to-end spatial proteomics pipeline integrating image correction, cell segmentation, and cell phenotyping
- Identifies both common and rare cell populations through semi-supervised clustering for deeper tissue characterization
- Generates clinically actionable insights that can predict outcomes such as organ transplant rejection
Abstract
The inventors developed a comprehensive pipeline for processing and analyzing multiplex spatial proteomic image data. The pre-processing includes steps like spillover correction, denoising, and aggregate removal to address multiple types of image artifacts. Following pre-processing, a semi-supervised clustering pipeline analyzes the refined images. This includes a supervised algorithm to identify general immune cell phenotypes and non-immune phenotypes, and an unsupervised algorithm to identify specific subclusters for detailed characterization of cell states. The method improves the accuracy of cell classification and phenotype identification, which can be used to predict clinical outcomes such as organ transplant rejection.
Benefit
- Comprehensive: Integrates image pre-processing, cell segmentation, and cell phenotyping into a single pipeline.
- Detailed phenotyping: Semi-supervised clustering approach provides sensitive and specific identification of cell phenotypes while enabling discovery of rare subpopulations.
- Clinically applicable: Includes algorithms that extract signals from identified cell states to make clinical predictions.
Publications
Spatially resolved immune exhaustion within the alloreactive microenvironment predicts liver transplant rejection, Science Advances, 12 April, 2024.
Other
Benchmarking studies using IMC images from human liver tissue and mouse hepatocellular carcinoma models