Fig. 4.

Exploring the spatial heterogeneity of immune cells in human breast cancer tissue. A The spatial composition of immune cells in human breast cancer was predicted by integrating spatial data with single-cell data covering all cell types. The results from CellDART were considered a reference for comparison with spSeudoMap. B The spatial composition of the immune cell types was predicted by spSeudoMap using CD45+ sorted single-cell data and visualized on the tissue. Macrophages (sum of macrophage_1, macrophage_2, and macrophage_3), memory CD4 T cells, and B cells showed similar spatial distributions, while CD8 NKT-like cells presented different patterns. C The scatter plots show the correlation between the cellular proportion predicted by spSeudoMap with sorted single-cell data and that estimated by CellDART with an unsorted single-cell dataset covering all cell types (reference). Spearman’s correlation coefficients and statistical significance (p-value) were calculated and are presented in the top-left corner of each plot. The correlation between CellDART and spSeudoMap was the highest in macrophages (the sum of macrophage_1, macrophage_2, and macrophage_3). Other cell types also showed weak but positive correlations except for CD8+ NKT-like cells, which showed a negative correlation