Cell Segmentation (Master)

The workflow aims to segment cell nuclei from cytoplasm. As a first step the images need to be preprocessed by splitting the channels into separate columns, nuclei and cytoplasm. To correct for background staining and reduce noise, we are performing a background subtraction of the nuclei channel and also apply Gaussian smoothing.

For segmentation, we first use the Otsu threshold method to distinguish between foreground and background. Afterwards we identify and label connected components (=nuclei) using the Connected Component Analysis node. Subsequently, we remove cell clumps and too small nuclei. As a last step, we perform a Voronoi based segmentation that identifies and labels the corresponding cytoplasm to the already labeled nuclei.