img-logoSerpicoSoutenancePhilippeRoudot img-logoInstitutCurieSoutenancePhilippeRoudot

Image Processing Methods for Dynamical Intracellular Processes Analysis in Quantitative Fluorescence Microscopy

img-soutenanceThesePhilippeRoudot Philippe Roudot

img-logoEquipeProjetSerpico (Equipe-projet SERPICO)

Soutenance de thèse - 22 mai 2014

L'exposé [53:45 mn]

Retour index-soutenances Thèses


We will present two projects involved in the quantification of dynamical intracellular processes in fluorescence microscopy.

We will first introduce a study of the instrumentation required for the quantification in frequency domain fluorescence lifetime imaging microscopy (FD FLIM). A FD FLIM measurement is defined as a series of images with sinusoidal intensity variations. The fluorescence lifetime is defined as the nanosecond-scale delay between excitation and emission of fluorescence. We propose two main contributions in this area: a modeling of the image process and noise introduced by the acquisition system (ICCD sensor); a robust statistical method for lifetime estimation on moving structures and intracellular vesicles.

The second part presents a contribution to the tracking of multiple particles presenting heterogeneous transports in dense conditions. We focus here on the switching between confined diffusion in the cytosol and motor-mediated active transport in random directions. We show that current multiple model filtering and gating strategies fail at estimating unpredictable transitions between Brownian and directed displacements. We propose a new algorithm, based on the u-track algorithm [Jaqaman et al., 2008], based on a set of Kalman filters adapted to several motion types, for each tracked object. The algorithm has been evaluated on simulated and real data (vimentin, virus) data.

We show that our method outperforms competing methods in the targeted scenario, but also on more homogeneous types of dynamics challenged by density.