Evaluating metrics used in deep learning for microscopy (infocus #73 March 2024)

Fluorescence microscopy is an essential tool in cell biology as it allows scientists to look at what is inside of a cell and easily see the organisation of organelle by labelling different organelles selectively.

DOI: 10.22443/rms.inf.1.264

One of this technique’s major limitations is that the light needed to image a live sample can also damage it, if applied at too high an intensity or for too long. This leaves the researcher needing to choose between a limited duration of imaging or image artefacts introduced by photo-damage. A possible solution is imaging at a low laser intensity and later computationally restoring acquired images (Weigert
et al., 2018). My project focused on investigating two state-of-the art deep learning methods used for image restoration in microscopy.

Return to listing