Thierry Pécot home page
Graduated from the Ecole Nationale d'Ingénieurs
de Brest in 2005 in electronics, I obtained a research master in image
analysis and statistics at the university of Rennes in 2006. Since october
2006, I am in PhD between the VISTA
team at the INRIA
Rennes - Bretagne Atlantique centre and the Pict-IBISA platform at the Institut Curie, under the supervision of Charles
Kervrann and Jean Salamero.
Research Topics
Biological context
In biology, the wide acquisition of multidimensional data (MRI, confocal microscopy, ...) make the image analysis more and more necessary. My research activities focus on intracellular traffic analysis. More specifically, I am interested in estimating global traffic characteristics for the protein of interest Rab6.
The image sequences of interest show Rab6 proteins under 3 possible states: they are either anchored in the Golgi, or free (diffusion) in the cytosol, or embedded into vesicles and trafficking on the microtubule network from the Golgi to Endoplasmic Reticulum entries presumably located at the periphery of the cell. The goal is to analyse Rab6 proteins that are trafficking, i.e. Rab6 embedded into vesicles. Consequently, separating the vesicles from the other components will make easier the traffic analysis.
Image taken from a 3D+T fluorescence spinning disk confocal image sequence on
a micropatterned cell (Institut Curie).
Objects/background separation
We developped a Conditional Random Field-based method for detecting objects of interest in fluorescence time-lapse microscopy. In this modeling, an energy functional defined by a discriminative term and spatial and temporal regularization terms is minimized thanks to a min-cut/max-flow algorithm. The non local discriminative term a patch-based distance that enlights the significative intensity peaks in contrast to the slowly-varying background. Furthermore, an improvement of the algorithm is provided by the joint estimation of the background and the vesicle detection.
Network tomography
Traffic estimation in image analysis is usually achieved by tracking algorithms (connexionnist approach, particular filtering, ...). These methods perform well and enable to track independently each object during the whole sequence. But they have limitations when a lot of objects that can fuse or merge are simultaneously moving, particularly if these objects can show a significant displacement between two consecutive images. That is why we want to develop a new approach for traffic analysis able to extract global characteristics. Network tomography is a method that estimates the origin-destination flows on a graph using link counts. This method was first developped for car trafficking and was repopularized in 1996 by Vardi to estimate the source-destination traffic intensities in communication networks like the Internet. It is a typical ill-posed problem. We have proposed to use this approach in order to simulate and analyse the intracellular traffic. Indeed, we can compare Rab6 trafficking and car trafficking by considering Rab6 proteins as people, vesicles as cars, and the microtubule network as a road network.
We exploit the network tomography to model vesicle trafficking. The specifity of this approach is that it does not intend to simulate the complex mechanisms responsible for traffic but rather provides image sequences sharing the same statistical properties than the real ones. Additional properties (biophysical characteristics, vesicle behaviors...) are added to better reproduce the observed trafficking. Furthermore, the biological assumption of closed stock of fluorescence (i.e. the cycle Golgi -> vesicles -> cytosol -> Golgi...) is respected.
Traffic analysis using network tomography requires several preliminary steps to be applied. First, it necessitates to design a cell partition (based on the origin-destination detection for example) to define a graph. Then, the fluorescence exchanges between neighbor regions or the number of vesicles that pass from one region to another one have to be estimated. Afterwards, a routing matrix (binary or probabilistic) defining the different paths from an origin to a destination is estimated. Finally, the inverse problem must be optimized to extract the different origin-destination pairs. Though developped for Rab6 protein trafficking estimation, this method should provide a general framework for the analysis of different proteins involved in the intracellular trafic.
Publications
Book chapter
J. Boulanger, Th. Pécot, P. Bouthemy, J. Salamero, J.-B. Sibarita, Ch. Kervrann. Simulation and estimation of intra-cellular dynamics and trafficking. Microscopic Image Analysis for Life Science Applications, S. Wong, R. Machiraju, J. Rittscher (eds.), chapter 7, pp. 169-190, Artech Publishing House, 2008.
International conferences
Th. Pécot, A. Chessel, S. Bardin, J. Salamero, P. Bouthemy, Ch. Kervrann. Conditional random fields for object and background estimation in fluorescence video-microscopy. Proc. IEEE Int. Symp. on Biomedical Imaging: from nano to macro (ISBI'09), Boston, USA, June 2009.
A. Chessel, Th. Pécot, S. Bardin, Ch. Kervrann, J. Salamero. Evaluation of image sequences additive decomposition algorithms for membrane analysis in fluorescence video-microscopy. Proc. IEEE Int. Symp. on Biomedical Imaging: from nano to macro (ISBI'09), Boston, USA, June 2009.
Th. Pécot, Ch. Kervrann, S. Bardin, B. Goud, J. Salamero. Patch-based Markov models for event detection in fluorescence bioimaging. Int. Conf. on Medical Image Computing and Computer Assisted Intervention (MICCAI'08), pp. 95-103 (2), New York City, USA, September 2008.
Th. Pécot, Ch. Kervrann. Patch-based Markov models for change detection in image sequence analysis. Proc. Int. Workshop on Local and Non-Local Approximation in Image Processing (LNLA'08), Lausanne, Switzerland, August 2008.
Th. Pécot, Ch. Kervrann, P. Bouthemy. Minimal paths and probabilistic models for origin-destination traffic estimation in live cell imaging. Proc. IEEE Int. Symp. on Biomedical Imaging: from nano to macro (ISBI'08), pp. 843-846, Paris, France, May 2008.
Th. Pécot, Ch. Kervrann, P. Bouthemy, A. Chessel, J.-B. Sibarita, J. Salamero, J. Boulanger. Network tomography and minimal paths for traffic flow estimation in time-lapse fluorescence microscopy. European Light Microscopy Initiative meeting (ELMI'08), Davos, Switzerland, May 2008.
Th. Pécot, Ch. Kervrann, P. Bouthemy. Network tomography-based tracking for intracellular traffic analysis in fluorescence microscopy imaging. Proc. Int. Conf. on Bio-inspired Systems and Signal Processing (BIOSIGNALS'08), pp. 154-161 (1), Funchal, Portugal, January 2008.
Th. Pécot, J. Boulanger, Ch. Kervrann, P. Bouthemy. Network tomography for trafficking simulation and analysis in fluorescence microscopy imaging. Proc. IEEE Int. Symp. on Biomedical Imaging: from nano to macro (ISBI'07), pp. 268-271, Arlington, USA, April 2007. (pdf)
National conference
Th. Pécot, J. Boulanger, C. Kervrann, P. Bouthemy. Tomographie de réseau appliquée à la simulation et à l'analyse de trafic dans des séquences d'images de microscopie à fluorescence. Proc. Colloque GRETSI, pp. 1-4, Troyes, France, September 2007.