Automated super-resolution microscopy for high-throughput imaging
Abstract: Fluorescence microscopes enable the visualization of biological samples with high contrast by labeling specific structures with fluorescent molecules. However, the spatial resolution of widely used microscopy techniques, such as widefield and confocal microscopy, is limited by the size of a focused spot of light, which roughly corresponds to half the wavelength of the illumination. To overcome this limitation, optical fluorescence nanoscopy techniques were developed, which achieve a higher spatial resolution by switching the fluorescent molecules within the sample between bright and dark states. Optical fluorescence nanoscopy techniques can be divided into two categories. The first, namely coordinate-targeted nanoscopy, switches the fluorescent molecules in a spatially annotated way, where it is known where and when the switching is induced. Instead, in stochastic approaches, the emitting molecules appear randomly in the sample and their location can be annotated with high spatial precision. This thesis focuses on reversible saturable optical fluorescence transitions (RESOLFT), a coordinate-targeted nanoscopy technique that initially relied on a beam of light that is moved across the sample (i.e., point scanning). Beams of different shapes and wavelengths are synchronized in time to generate super-resolution images. However, this approach creates a trade-off between imaging speed and the field of view. While it can acquire small fields of view at a fast speed, imaging larger areas can take up to several minutes. To overcome this limitation, the molecular nanoscale live imaging with sectioning ability (MoNaLISA) microscope employs patterns of light to parallelize RESOLFT imaging, collecting the fluorescence at different points simultaneously.Throughput in microscopy is characterized as the number of cells per unit of time and area that a microscope can image. Achieving high throughput enables capturing fast cell dynamics and understanding how they correlate over large fields of view, providing insights into biological processes. Therefore, in this thesis I developed strategies to increase the throughput of coordinate-targeted nanoscopy methods. Firstly, I was involved in the mathematical formulation of fluorophore switching and its relationship to image resolution, in order to provide a framework to relate different parameters to image quality (Paper I). Secondly, I developed ImSwitch, an open-source software for microscope control. It implements a software architecture that enables flexibility and adaptability between different microscopy modalities (Paper II). Thirdly, I built a setup that increases the field of view by more than four times than previous implementations of MoNaLISA (Paper III). Finally, I combined MoNaLISA and ImSwitch to provide a framework to parallelize image acquisition, reconstruction, and visualization using multiple computational units (Paper IV).
CLICK HERE TO DOWNLOAD THE WHOLE DISSERTATION. (in PDF format)