Show simple item record

Novel design and data analysis concepts for MINFLUX nanoscopy

dc.contributor.advisorHell, Stefan W. Prof. Dr.
dc.contributor.authorBarthel, Hannah Eva
dc.titleNovel design and data analysis concepts for MINFLUX nanoscopyde
dc.contributor.refereeHell, Stefan W. Prof. Dr.
dc.description.abstractengMINFLUX is a novel concept for super-resolution fluorescence microscopy that achieves unprecedented image resolution despite the limited photon budget of fluorescent molecules. The goal of this thesis is to improve MINFLUX image quality using new optical and analytical approaches. To this end, we aim to increase the information contained in the emitted photons by optimizing the excitation light pattern. Furthermore, by integrating new concepts into the post-processing workflow, we utilize this information in an optimal manner. In the first part, I examine the suitability of low-order Lissajous figures for 2D MINFLUX illumination schemes that are compatible with alternative, simpler scanning systems based on galvanometric beam deflectors. The width and aspect ratio of these figures are optimized as a function of different imaging parameters in order to maximize the theoretically achievable localization precision. The optimized Lissajous geometries exhibit a substantially more isotropic precision than the previously published 2D excitation pattern. In the second part, I implement a fast, parameter-free 3D position estimator for MINFLUX nanoscopy based on deep learning. After an optimization of the architecture and the training data, the neural network yields a lower localization error than the currently used maximum-likelihood estimator, especially in the low-photon and high-background regimes. Moreover, the amount of time required for estimating the emitter positions is significantly reduced. To conclude, this thesis provides new methods for increasing the information content of MINFLUX images by effectively making best use of the available photons. By allowing simpler instrumentation and reducing the number of analysis parameters, the approaches developed herein will reduce the complexity and improve the performance of future MINFLUX
dc.contributor.coRefereeRizzoli, Silvio O. Prof. Dr.
dc.subject.engOptical nanoscopyde
dc.subject.engSuper-resolution fluorescence microscopyde
dc.subject.engSingle-molecule localizationde
dc.subject.engDeep learningde
dc.subject.engNeural networksde
dc.subject.engStatistical modelingde
dc.affiliation.instituteGöttinger Graduiertenschule für Neurowissenschaften, Biophysik und molekulare Biowissenschaften (GGNB)de
dc.subject.gokfullBiologie (PPN619462639)de

Files in this item


This item appears in the following Collection(s)

Show simple item record