Identifying and eliminating image artefacts in live cell localisation microscopy

Identifying and eliminating image artefacts in live cell localisation microscopy

Speaker: Dr Richard Marsh, Randall Centre for Cell & Molecular Biophysics, KCL

Host: Susan Cox

Abstract: The super-resolution imaging techniques based on localisation of individual fluorophores (PALM & STORM) have become commonplace in biological imaging. However, the imaging speed required for practical live cell experiments necessitates having to image data in which fluorophores overlap each other. Many specialised algorithms designed to cope with the spatial overlap of individual flurophore emission patterns (or 'spots') exist but all tend to produce artefacts in the reconstructed image under certain conditions. As the nature of these artefacts depends on the local structure and can resemble true biological features they can be very difficult to detect. The presence of false structure in a reconstruction also makes the assessment of the resolution of the image very difficult, often with misleading results.

Here I will talk about our new and simple analysis tool, Haar Wavelet Kernel Analysis (HAWK), that does not produce artefacts, and demonstrate its effectiveness on synthetic, fixed and live cell data. I will also show that it is compatible with 3D imaging. Lastly, I will discuss the problems of assessing resolution in an artefactual image, the origin of these artefacts and our initial attempts at a generalised method for detecting sub-diffraction scale artefacts in a super-resolution microscopy image.