Defining Keypoints to Align H&E Images and Xenium DAPI-Stained Images Automatically

Abstract

10X Xenium is an in situ spatial transcriptomics platform that enables single-cell and subcellular-level gene expression analysis. In Xenium data analysis, defining matched keypoints to align H&E and spatial transcriptomic images is critical for cross-referencing sequencing and histology. Currently, it is labor-intensive for domain experts to manually place keypoints to perform image registration in the Xenium Explorer software. We present Xenium-Align, a keypoint identification method that automatically generates keypoint files for image registration in Xenium Explorer. We validated our proposed method on 14 human kidney samples and one human skin Xenium sample representing healthy and diseased states, with expert manually marked results. These results show that Xenium-Align could generate accurate keypoints for automatically implementing image alignment in the Xenium Explorer software for spatial transcriptomics studies. Our future research aims to optimize the method's runtime efficiency and usability for image alignment applications.