Abstract

Advances in multiplexed in situ imaging are revealing important insights in spatial biology. However, cell type identification remains a major challenge in imaging analysis, with most existing methods involving substantial manual assessment and subjective decisions for thousands of cells. We propose a novel machine learning algorithm, CELESTA, which uses both cells protein expression and spatial information to identify cell type of individual cells. We demonstrate the performance of CELESTA on multiplexed immunofluorescence in situ images of colorectal cancer and head and neck cancer. Using the cell types identified by CELESTA, we identify tissue architecture associated with lymph node metastasis in HNSCC, which we validate in an independent cohort. By coupling our in situ spatial analysis with single-cell RNA-sequencing data on proximal sections of the same tissue specimens, we identify and validate cell-cell crosstalk associated with lymph node metastasis, demonstrating the power of spatial biology to reveal clinically-relevant cellular interactions.