Spectre Home About Citation Get started Protocols Tutorials Spatial Report Issue Q&A Github

Spatial - image visualisation and cell segmentation

Image visualisation

Image visualisation

Here we provide links for common software options for image visualisation. These are not required for cell segmentation, but are helpful for exploration of the raw imaging data.


Cell segmentation options

Here we provide protocols for image visualisation and cell segmentation. Data required: MCD files generated by the Hyperion imaging mass cytometer (IMC).

Basic IMC cell segmentation using nuclear pixel expansion
Modified Bodenmiller segmentation protocol with Ilastik
Multi-cut boundary segmentation for dense IMC images

This is the simplest form of cellular segmentation. The nuclear signal is identified, and the boundary is expanded outwards by a certain number of pixels. This then becomes the boundary of the cell mask. Although a very simplistic approach, only nuclear signal is required. In this approach, the user trains a classifier to identify and predict 'nuclear', 'cytoplasmic', or 'background' pixels. This can be used to create more comprehensive cell masks than can be achieved using simple pixel expansion methods. This is based on the Bodenmiller lab workflow described here. Some IMC images contain extremely dense collections of cells, where cytoplasmic (and sometimes) nuclear signal from one cell is difficult to distinguish from another. In this protocol we describe boundary-based segmentation using the 'multi-cut' workflow in Ilastik.


Code for the table structure adapted from https://satijalab.org/seurat