Rapid light sheet fluorescence microscopy for dynamic imaging of living organisms

Abstract

The primary goal of this thesis is to develop a light sheet based microscopy system that provides non-invasive images having high spatial and temporal resolution. The fluorescence microscopy has become an indispensable tool for biologists to understand the underlying mechanisms of various biological processes. The phenomenon of fluorescence offers non-invasive imaging with high speci ficity and single molecule sensitivity which is of interest in the life sciences research. The confocal microscopy has emerged as a potential technique which enables optical sectioning by employing a pinhole in the detector side to eliminate out of focus light. The confocal microscopy has been combined with other techniques such as multi-photon excitation microscopy to tackle scattering and to improve the penetration depth. The super resolution 4Pi microscopy is also combined with confocal detection to improve axial resolution. Past decade has seen the evolution of a large number of super resolution techniques such as STED, fPALM, PALM, STORM, SSIM, RESOLFT and GSDIM. This was a major breakthrough in fluorescence microscopy enabling single molecule resolution. Imaging of dynamical processes such as embryo development is an important and challenging problem in developmental biology. Long period monitoring is required to capture these processes which demand the sample to be kept in its natural environment with minimal interference from the probing light. The photobleaching and photodamage are the real issues when the sample is exposed to probing light for a long time. Imaging at high spatial and temporal resolution with minimal photodamage is a real challenge even for confocal and two photon excitation microscopy. Light sheet microscopy offers multitude of possibilities to the photobleaching problem and has found applications in various domains during the last decade. This thesis introduces an improvised light sheet microscopy technique where the number of images needed to construct 3D volume is greatly reduced by choosing an alternate acquisition strategy. The statistical image reconstruction techniques such as maximum likelihood (ML) approach is used for post-processing and has been shown to improve the image quality for the applications presented in the thesis. The introduction of light sheet illumination on a micro fluidic platform is also studied to enable 3D imaging of uni-cellular and multi-cellular organisms. A brief summary of the work is given below. Chapter 1 gives a brief outline of the developments in the field of fluorescence microscopy. The quest for improved resolution, contrast, penetration depth, speed and minimal damage to the sample has motivated researchers to come up with different microscopy designs which can tackle the aforementioned aspects. The emergence of light sheet microscopy is explained in detail as a promising tool in scenarios where the conventional techniques like confocal and multi-photon excitation microscopy are inadequate to overcome the challenges in various life science research areas. Chapter 2 is dedicated to explain the fundamentals of fluorescence. A brief outline is given to introduce various contrasts existing in optical microscopy, highlighting the advantages of fluorescence. Resolution of an imaging system is discussed in detail and the idea of the point spread function is introduced as it determines the performance of an optical system in terms of resolution. Some of the widely used fluorescence microscopy techniques are explained to provide an overview of developments happening in the fi eld of fluorescence microscopy. Image degradations due to blurring, noise and aberrations are inescapable in an imaging system. In general most of the microscopic samples are 3D objects. But what we acquire is a 2D image and can have ambiguities due to its 3D nature, like presence of out of focus features in the image. Hence the 2D image is a false representation of the 3D object. In order to tackle these issues image deconvolution techniques are usually employed. In chapter 3 we discuss about the state-of-the-art statistical image reconstruction technique maximum likelihood (ML) approach for image deconvolution. Due to its iterative nature ML algorithm is inherently slow. In order to process and view images in real-time, ML algorithm has to be accelerated. A step in this direction has been the incorporation of Biggs-Andrews algorithm into ML framework to accelerate ML. BA approach is based on vector extrapolation method which is a simple method without any derivative calculation and inherits automatic acceleration. We have tested the performance of the algorithm on microscopy images obtained from three techniques wide field, confocal and super-resolution 4Pi microscopy. The convergence is improved by a factor of two for all the tested images. Developmental biology is one of the promising areas which demands fast and high resolution imaging to understand various biological processes during embryo development. Light sheet florescence microscopy was introduced to tackle this highly challenging problem because of its inherent optical sectioning capability which helps in eliminating out of focus illumination. This reduces photobleaching and phototoxicity. We have developed limited view light sheet microscopy (LVLSM) in which the volume of zebra sh embryo is constructed from a very few angular views. Chapter 4 deals with LVLSM. The rotation and translation involved in multi-view microscopy is time consuming and hundreds of images are acquired to construct a 3D volume. In this work we have developed a scheme that uses only rotation to acquire the data. In addition, we have used ML algorithm to improve the contrast of the image and to remove the noise. We have reconstructed a ve day old zebra sh embryo using 18 views which is an order of magnitude less compared to the number of images used in multi-view light sheet microscopy. We study the effect of limited number of views in the 3D image reconstruction in the next chapter in the light of speeding up the imaging process as well as for reducing of photobleaching further. We have done the time-lapse imaging of a ve day old zebra sh embryo to study the effect of photobleaching. The uorescence decay curve is fitted with mono-exponential decay and the parameters are obtained. We have constructed 3D volume from 18, 9 and 6 angular views at 10o; 20o and 30o angular separation respectively and checked the performance in terms of contrast. The image quality with 18 and 9 views were found to be almost the same. However, there is a reduction in contrast for reconstruction with only 6 views. But even with 6 views, it is found that the structural details are retained in volume reconstruction. Intensity line plots are employed to quantitatively check the reconstruction with 18, 9 and 6 views and it is found that there is a good agreement between reconstructions with 18 and 9 views. In chapter 6 we have explored the light sheet illumination for the 3D imaging of organisms during ow on micro fluidic platform. Light sheet illumination can provide optical sectioning and hence it is possible to get 2D cross-sections of the sample during the fllow. The micro fluidic channel is kept at an angle with respect to the illumination axis and the illumination and detection arms are orthogonal. Optimization of the light sheet dimensions, ow parameters and camera settings facilitates 3D imaging without any translation of the sample. The performance of the system is checked by imaging samples of two different sizes, HeLa cells and C: elegans worms. In order to tackle the problem of motion blur, maximum likelihood algorithm is employed in which we have used experimentally measured PSF to deblur the image. PSF is estimated by owing nanobeads through the channel at the same ow rate as that of the sample. The reconstructed images show better contrast and less noise compared to raw images. The proposed system is promising for 3D imaging ow cytometry as well as for 3D imaging of live model organisms for high throughput screening. The conclusion for the thesis is given in chapter 7. Some of the prospects of the work is given as future scope.