Fig. 8: Application of FUSION for 10x Xenium data.

A, B Glomerulus from a healthy patient with automatically segmented cells and functional tissue units overlaid. B Same view as in (A) with the glomerular overlay turned off to highlight segmented cells. A high proportion of detected cells (76.4%) in this glomerulus were identified as podocytes. C, D Glomerulus from a diabetic patient with segmented cells and functional tissue units overlaid. D Same view as in (C) with the glomerular overlay turned off to highlight segmented cells. Compared to the reference glomerulus, this glomerulus contains a higher proportion of fibroblasts and endothelial cells. Scale bar indicates 50 µm. Color bar indicates the most abundant cell type in each structure. E Spatially variable gene analysis highlighting leukocyte migration (Gene Ontology term GO_0050900) pathway expression aligned with the histology image in a chronic kidney disease sample. Region 1 shows immune infiltration overlying interstitial matrix expansion. Region 2 denotes immune infiltration with upregulation of the pathway in epithelial nuclei. F Visualization of the same leukocyte migration pathway in a chronic kidney disease sample in FUSION with a minimum threshold of 0.35. Region 1 shows upregulation in epithelial nuclei, while Region 2 indicates ontology mapping to immune cells in the interstitium. FUSION ontology mapping helps identify regions of epithelial–immune crosstalk.