Biological Image Super-resolution Enhanced with Tensor

About the Project

The resolution of "conventional" microscopes does not allow to answer a certain number of biological questions that need to better locate the observed phenomena. For example, the confocal microscopes have a resolution of 200 nm in the lateral (plane (x, y) orthogonal to the axis optical microscope) and 400nm in depth (z axis in the direction of the optical axis) while the study of the composition and stoichiometry of ion channels have, so far, been difficult to evaluate. The SimPull technique, developed at iBV by Guillaume Sandoz's team, can answer these questions but only in vitro and without precise location [Levitz 2016].

Recently, a TIRF-MA (Total Internal Reflection Fluorescence-Multi Angle) microscopy system has been prototyped at iBV to obtain 3D images of a deep resolution 30 nm (in z) over a thickness of 600nm above the lamella, thanks to an algorithm of
reconstruction developed at I3S. This makes it possible to visualize the exchanges of the cell with its outside. However, the lateral resolution (x, y) of this system remains limited by diffraction.

The objective of BIZET  project is to develop an acquisition protocol and an algorithm of reconstruction to obtain a super-resolved 3D image in the 3 dimensions, in combining TIRF-MA reconstruction with two super-resolution techniques in lateral (x, y). The two approaches considered use blinking fluorophores, the time variable becomes a full variable of the 3D reconstruction problem. To exploit these series images, extracting correlations, separating sources (neighboring fluorophores) and to reconstruct the volume of 3D image BISET project team proposes to use tensorial analysis.

The joint exploitation of several diversities for the (blind) separation of sources via the Tensor decomposition is a widely used approach in signal processing, and well adapted to the data that will be acquired in this project. The ultimate goal is to achieve a sufficient 3D resolution to be able to study dynamically the ion channels

Principal Investigator
Project's partner(s)
  • Guillaume SANSOZ, iBV
July 2018 - December 2019
Related Documentation

BISET presentation (DATA SCIENCE MEETUP, December 2017)