Spatial transcriptomics enables studying cellular interactions by measuring geneexpression in situ while preserving tissue context. Within tissues, distinct cellularniches define micro-environments that influence cell states and function. A fundamental task in spatial transcriptomics is identifying differentially expressed geneswithin a specific cell type across different niches to quantify context-dependentcell state variation. Despite advances in cell segmentation algorithms, the persisting problem of the wrong assignment of molecules to cells can obscure the analysis by introducing spurious differentially expressed genes that originate fromneighboring cells rather than the group of interest. Here, we introduce 2DE, aprobabilistic framework designed to refine spatial differential expression analysesby filtering out genes that are over-expressed due to local contamination ratherthan true cell-intrinsic expression. 2DE operates downstream of any differentialexpression method, filtering irrelevant genes by considering gene over-expressionrelative to the expression in the neighborhood and returning marker confidencescores. In a study of human breast cancer, we demonstrate that 2DE improves theprecision of the discoveries.