Type of presentation: Poster

IT-3-P-3035 Drift Correction Strategies for Single Molecule Localisation Microscopy

Hagmann M.^{1 2 6}, Prakash K.^{3 4 6}, Best G.^{1 2}, Kaufmann R.⁵, Birk U.^{1 4}, Cremer C.^{1 3 4}

¹Kirchhoff-Institute for Physics (KIP), University of Heidelberg, Heidelberg, Germany, ²University Hospital Heidelberg, University of Heidelberg, Heidelberg, Germany, ³Institute for Pharmacy and Molecular Biotechnology (IPMB), University of Heidelberg, Heidelberg, Germany, ⁴Institute of Molecular Biology (IMB), Mainz, Germany, ⁵Division of Structural Biology, Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford, UK, ⁶equal contributions

K.Prakash@imb-mainz.de

The correct position determination of fluorescent molecules is crucial for the interpretation of localisation microscopy data, e.g. of the biological structure investigated. The relative position of fluorophores with respect to the detector is highly sensitive to environmental disturbances (e.g. acoustic vibrations) and to mechanical instabilities of the microscope hardware (e.g. thermal expansion or mechanical relaxation). These disturbances cause distortion in the recorded image which pose a new natural limit to the localisation accuracy, especially for new acquisition protocols that allow acquisition times in the order of hours.

Here, we present two drift correction strategies based solely on data already acquired without any fiducial markers. We found that in some flavours of SMLM, many of the biological samples exhibit enough permanent (photostable) structure to reveal information about the sample location, or if this is not the case, reconstructions of a subset of the complete localisation data stack can be used at several time points to gain information about the sample drift. In both cases, the drift is found by determining the peak of the computed 2D-autocorrelation function. A polynomial or a set of Fourier functions is fitted through the data, based on which the dislocation of every localised fluorophore in a given frame of the acquired image stack is subtracted.

Using this approach, we successfully corrected localisation microscopy data down to a final drift less than 5 nm, which is comparable with fiducial markers based strategies. We demonstrate that with this procedure the resolution of the final reconstructions is substantially enhanced and the theoretical limit of localisation accuracy is almost restored.

We gratefully acknowledge the colleagues at IMB who supported us with reagents. In particular, we would like to thank Aleksander Szczurek and Hyun-Keun Lee for samples, reagents and many interesting discussions. This work is supported by the Boehringer Ingelheim Foundation. The support of University Hospital Heidelberg (Prof. S. Dithmar) to G.B. and M.H. is also gratefully acknowledged.