Annotation and automated segmentation of single-molecule localisation microscopy data

Oliver Umney,Joanna Leng,Gianluca Canettieri,Natalia A Riobo-Del Galdo,Hayley Slaney,Philip Quirke,Michelle Peckham,Alistair Curd

DOI: https://doi.org/10.1111/jmi.13349

2024-08-02

Abstract:Single Molecule Localisation Microscopy (SMLM) is becoming a widely used technique in cell biology. After processing the images, the molecular localisations are typically stored in a table as xy (or xyz) coordinates, with additional information, such as number of photons, etc. This set of coordinates can be used to generate an image to visualise the molecular distribution, for example, a 2D or 3D histogram of localisations. Many different methods have been devised to analyse SMLM data, among which cluster analysis of the localisations is popular. However, it can be useful to first segment the data, to extract the localisations in a specific region of a cell or in individual cells, prior to downstream analysis. Here we describe a pipeline for annotating localisations in an SMLM dataset in which we compared membrane segmentation approaches, including Otsu thresholding and machine learning models, and subsequent cell segmentation. We used an SMLM dataset derived from dSTORM images of sectioned cell pellets, stained for the membrane proteins EGFR (epidermal growth factor receptor) and EREG (epiregulin) as a test dataset. We found that a Cellpose model retrained on our data performed the best in the membrane segmentation task, allowing us to perform downstream cluster analysis of membrane versus cell interior localisations. We anticipate this will be generally useful for SMLM analysis.