Description
Date depot: 9 avril 2018
Titre: Next-generation Structured Illumination Microscopy for biological imaging
Directeur de thèse:
Jean-Christophe OLIVO-MARIN (Analyse d'images biologiques)
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Non defini
Resumé:
Fluorescence
microscopy is one of the most used tools in modern experimental biology. During
the last few decades, several super-resolution techniques have been proposed to
overcome the theoretical limit related related to diffraction. The goal of this project is to take
advantage of several mathematical tools and frameworks developed at the BioImage Analysis
Unit (BIA) to improve one
of these optical super-resolution techniques, namely Structured Illumination
Microscopy (SIM)[1]. SIM is a wide field technique that increases the lateral
spatial resolution up to a factor 2. Although the resolution enhancement
obtained with SIM cannot rival that of pointillistic approaches one of its main
advantages is the ability to provide enhanced resolution from a modest number
of images and low illuminance. It is thus the technique of choice for imaging
dynamic processes in live samples. Super resolution is achieved by projecting
periodic illumination patterns onto the sample. This structured illumination is
generated by the interference of two or more coherent light beams. The
acquisition of a sequence of images captured with different pattern orientations
and phases allows computing the final super resolved image through a dedicated
algorithm.
Over
the last few years in collaboration with teams at ESPCI-Paristech and
Centrale-Supélec/Université Paris-Saclay, Olivo-Marin’s team has developed an
alternative approach for image reconstruction in SIM [2] that requires a
reduced number of raw images to build a super resolved optically sectioned
image. The method is based on an inverse problem approach [3] where the
solution is inferred via a joint myopic criterion for image and modulation (or
acquisition) parameters. The estimate is chosen as the minimizer of a nonlinear
criterion, numerically calculated by means of a block coordinate optimization
algorithm. The efficiency of this approach has been demonstrated on fixed cells
samples as well as on living cells: with only 4 acquired images it has been
possible to obtain a final image similar to the one obtained with 9 raw images.
This novel, fast and unique SIM microscope therefore allowed dynamic
super-resolution imaging [4].
Notwithstanding
these advances, the reconstruction algorithm is not fully robust and the
resulting super-resolution image is strongly dependent on the signal-to-noise
ratio of each sample. Moreover, this technique has to be scaled to several
excitation wavelengths at the same time with a fast acquisition rate, demanding
the processing of an increased amount of data. We propose to address these
technical challenges through the introduction of new algorithms as well as their
applications to different optical setups (3D SIM, blindSIM, …).
The
project will consider the design and implementation of dedicated dynamic
acquisition algorithms for SIM as well as the application to relevant
biological questions, like the orchestration of the molecular mechanisms of pathogen
entry in cells. One
challenging part of this project will be to develop new algorithms to improve
image quality and acquisition time, and further improve the capability of the
structured illumination microscope and in particular to design and validate
fast Compressive Sensing-based dynamic protocols in SIM microscopy via specific
acquisition devices.
The
resulting advances in SIM should lead to massively increased image acquisition
rates with guaranteed resolution and quality for a given acquisition rate and
should enable the design of optimized dynamic imaging protocols, dedicated to
specific biological paradigms and experimental conditions.
Doctorant.e: Chalumeau Robin