cell segmentation tool for cryo soft x-ray tomography

Contour, a new semi-automated segmentation and quantitation tool for Cryo soft X-ray tomography

Cryo-soft X-ray tomography (cryo-SXT) is an emerging imaging technology that solves the demand for imaging the whole ultrastructure of relatively thick cell samples without staining or chemical treatment. Cryo-SXT can image the subcellular organization of whole hydrated cells, resolution of 25-40 nm, in their native state, with very little preparation.

Cryo-SXT is becoming more popular in biological research for studying the shape of cellular compartments and how they change in response to various stimuli, such as viral infections. The segmentation of these compartments is constrained by time-consuming manual tools or machine learning algorithms that need a significant amount of time and effort to learn.

Contour is a new, simple-to-use, automation segmentation tool that allows for the rapid segmentation of tomograms to distinguish various cellular compartments. Using Contour, cellular structures may be segmented based on their projection intensity and geometrical width by applying a threshold range to the picture and removing noise smaller in width than the cellular compartments of interest.

The approach is less laborious and less prone to human judgment mistakes than existing technologies that need features to be manually traced, and it does not require training datasets.

The paper referred to below has shown that Contour can easily segment high contrast compartments such as mitochondria, and lipid droplets. Contour can extract geometric metrics from 3D segmented volumes, giving users a new way to quantify cryo-SXT data.

The growing power of new segmentation tools and techniques, as well as the availability of a laboratory Soft X-ray Microscope (SiriusXT’s SXT-100), is making cryo-SXT a fast, convenient and effective imaging modality for imaging cellular substructure.

Contour, semi-automated segmentation and quantitation tool for cryo-soft-X-ray tomography

Kamal L. Nahas, Stephen Graham, Maria Harkiolaki, et.al.