Fig. 1: Overview of the RAMCES segmentation pipeline.
From: Membrane marker selection for segmenting single cell spatial proteomics data
First, CODEX data goes through a series of pre-processing steps, including rescaling, normalization, and the discrete wavelet transform (DWT). Next, a pre-trained CNN is applied to the data to rank proteins/markers. Top markers are combined to create weighted images, which are then used for segmenting cells by a membrane-based segmentation pipeline, such as Cytokit10. The resulting segmented data can then be used for downstream analyses and for cell type assignment.
