Multiplexed immunohistochemistry image analysis using sparse coding

Multiplexed immunohistochemical (IHC) methods have been developed to evaluate multiple protein biomarkers in a single formalin-fixed paraffin-embedded (FFPE) tissue section. Since distinct populations of resident and recruited immune cells in tissues (and tumors) not only regulate progression of malignant disease, these also represent targets for novel immune-based therapies; thus, improved tissue biomarker assessment evaluating immune responses in situ are needed. To objectively identify distinct cell subsets in tissues and tumors, we adopted sparse coding approaches enabling modeling of data vectors as sparse linear combinations of basis elements, to audit cellular presence and phenotypes using image cytometry datasets with unbiased assessments. By doing comparative analyses between manual gating (ground truth) and sparse coding, we report that results are comparable as obtained by manual gating strategies, and demonstrate robustness and objectivity of this novel bioinformatics approach.