DeepCLEM: automated registration for correlative light and electron microscopy using deep learning

Rick Seifert; Sebastian M Markert; Sebastian Britz; Veronika Perschin; Christoph Erbacher; Christian Stigloher; Philip Kollmannsberger

doi:10.12688/f1000research.27158.2

DeepCLEM: automated registration for correlative light and electron microscopy using deep learning

F1000Res. 2023 Dec 28:9:1275. doi: 10.12688/f1000research.27158.2. eCollection 2020.

Authors

Rick Seifert^{1

2}, Sebastian M Markert², Sebastian Britz², Veronika Perschin², Christoph Erbacher³, Christian Stigloher², Philip Kollmannsberger¹

Affiliations

¹ Center for Computational and Theoretical Biology, University of Würzburg, Würzburg, 97074, Germany.
² Imaging Core Facility, Biocenter, University of Würzburg, Würzburg, 97074, Germany.
³ Department of Neurology, University of Würzburg, Würzburg, 97074, Germany.

Abstract

In correlative light and electron microscopy (CLEM), the fluorescent images must be registered to the EM images with high precision. Due to the different contrast of EM and fluorescence images, automated correlation-based alignment is not directly possible, and registration is often done by hand using a fluorescent stain, or semi-automatically with fiducial markers. We introduce "DeepCLEM", a fully automated CLEM registration workflow. A convolutional neural network predicts the fluorescent signal from the EM images, which is then automatically registered to the experimentally measured chromatin signal from the sample using correlation-based alignment. The complete workflow is available as a Fiji plugin and could in principle be adapted for other imaging modalities as well as for 3D stacks.

Keywords: Correlative Microscopy; Deep Learning; Image Registration; In-silico labeling.

Grants and funding

SMM was supported by a PhD grant from the Studienstiftung des Deutschen Volkes (German Academic Scholarship Foundation). VP and CS are funded by Deutsche Forschungsgemeinschaft GRK2581 SPHINGOINF and CS by STI 700/1-1. Computational work was performed using the High-Performance Computing Cloud of Würzburg University (DFG project 327497565).This publication was supported by COST Action NEUBIAS (CA15124), funded by COST (European Cooperation in Science and Technology). The JEOL JSM-7500F field emission scanning electron microscope (SEM) was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – 218894895 (INST 93/761-1 FUGG) and the structured illumination microscope (SIM) was funded by DFG - 261184502 (INST 93/823-1 FUGG). VP and CS were funded by DFG within RTG 2581 SphingoINF (project 6).