Development of open source microscopy including optical autofocus for automated imaging and single molecule localisation microscopy
File(s)
Author(s)
Lightley, Jonathan
Type
Thesis or dissertation
Abstract
This PhD thesis presents the development of a cost-effective and open (hardware and software) microscopy platform able to perform automated single molecule localisation microscopy (SMLM) for super-resolved high-content analysis (HCA). The work is motivated by the desire to increase access to advanced microscopy techniques like SMLM and HCA, including in lower resourced communities. For SMLM I have utilised and refined easySTORM, a low-cost and robust implementation of dSTORM. For automated microscopy I have developed a novel optical autofocus module that can maintain focus lock during extended easySTORM acquisitions and maintain focus as the microscope images multiple fields of view (FOV) across a multiwell plate. Initially I worked with commercial microscope frames and then extended automated microscopy and easySTORM to a cost-effective, modular, open microscopy platform, “openFrame” that can be adapted to a wide range of microscopy modalities. For this, I combined the openFrame with a low-cost multimode diode laser bank and novel low-cost, cooled CMOS sensor cameras.
I developed two different hardware-based autofocus systems focussing near infrared laser beams from fibre-coupled diode laser sources onto the microscope coverslip with the back reflection being imaged at a secondary autofocus camera. The first autofocus technique presented uses two cylindrical lenses to create a collimated elliptical beam profile with differing confocal parameters (and therefore different range/precision) in orthogonal directions. The second autofocus technique presented utilises a slit to form the elliptical beam profile and then uses a convolutional neural network (CNN) to determine the microscope defocus from the autofocus camera images. The CNN autofocus was then able to operate robustly for months over a range of ±100μm and achieve an accuracy of greater than 200nm. I integrated this autofocus system with an automated multiwell plate easySTORM microscope applied to image focal adhesion structures within melanoma cells and bacteria undergoing phagocytosis in THP1 cells.
I developed two different hardware-based autofocus systems focussing near infrared laser beams from fibre-coupled diode laser sources onto the microscope coverslip with the back reflection being imaged at a secondary autofocus camera. The first autofocus technique presented uses two cylindrical lenses to create a collimated elliptical beam profile with differing confocal parameters (and therefore different range/precision) in orthogonal directions. The second autofocus technique presented utilises a slit to form the elliptical beam profile and then uses a convolutional neural network (CNN) to determine the microscope defocus from the autofocus camera images. The CNN autofocus was then able to operate robustly for months over a range of ±100μm and achieve an accuracy of greater than 200nm. I integrated this autofocus system with an automated multiwell plate easySTORM microscope applied to image focal adhesion structures within melanoma cells and bacteria undergoing phagocytosis in THP1 cells.
Version
Open Access
Date Issued
2023-03
Date Awarded
2024-01
Copyright Statement
Creative Commons Attribution NonCommercial Licence
Advisor
French, Paul
Dunsby, Christopher
Sponsor
Engineering and Physical Sciences Research Council
Publisher Department
Physics
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)