Cell segmentation, that is, the automated detection of cells in an image, is the first step for analysis on the Enable Platform. Since Opal-stained images follow the general form of most multiplex fluorescent images (i.e., one channel per biomarker stain and one additional channel for the nuclear counterstain), DeepCell Segmentation is usually the best algorithm for accurately detecting cell nuclei and, if a general membrane marker is used, cell borders.
While DeepCell performs quite well, it is not perfect and will sometimes detect cells incorrectly. For example, it may incorrectly identify cells in acellular structures or imaging artifacts, or over- or under-segment overlapping cells. To improve the segmentation results, it is often useful to perform Cell QC to find these segmentation artifacts or false cells and take them out of downstream analysis.
The end result of performing these steps is the generation of a table of cells that have passed QC. Each cell within the table will have spatial x-/y-coordinates, and values for the average biomarker signal detected within the cell border for each biomarker channel. This table of cells is used for all downstream analysis on the platform.