High-plex immunofluorescence (IF) workflows typically rely on unsupervised clustering, followed by cell type annotation at a cluster level for cell type assignment. Most of these methods use marker expression averages that lack a statistical evaluation of cell type annotations, which can result in misclassification. Here, we propose a strategy through an end-to-end pipeline using a semi-supervised, random forests approach to predict cell type annotations. Our method includes cluster-based sampling for training data, cell type prediction, and downstream visualization for interpretability of cell annotation that ultimately improves classification results. We show that our workflow can annotate cells more accurately with a training set < 5 % of the total number of cells tested. In addition, our pipeline outputs cell type annotation probabilities and model performance metrics for users to decide if it could boost their existing clustering-based workflow results for complex IF data.

Selected samples from anal precancers and cancers were used to create a tumor microarray (TMA) for spatial phenotyping analysis using the Akoya Phenocycler Fusion (formerly known as CODEX). CODEX data were generated at The Bursky Center for Human Immunology and Immunotherapy Programs, Washington University School of Medicine.

This approach uses tissue-based cyclic immunofluorescence for highly multiplexed immunofluorescence imagining on FFPE specimens from glass slides. Data shown within this manuscript were taken from 2, 2mm core biopsies sourced from the University of Wisconsin-Madison using a custom panel. Final stitched images for the cores, segmentation results from StarDist, and expression summaries are used within the workflow for processing and cell annotation.

30 maker panel: each image slice corresponds to a marker in the order shown:

DAPI, Ki67, CD31, FOXP3 , CD56

CD34, CD4, CD20, CD45, CD163

HLA-A, LAG3, CD8, SMA, PDL1

CD21, PanCK, IDO1, bCat1, CD14

PD-1, CD44, CD3e, CD45RO, CD68 

GZMB, HLA-DR, ICOS, HIF1A, CK17