SEMINARIO Dr. Siewert Hugelier:"Coupling image processing and chemometric resolution"

Docenti, ricercatori, dottorandi, studenti e tutti gli interessati sono invitati al seminario, organizzato nell'ambito delle attività del Corso di Dottorato in Models and Methods for Materials and Environmental Sciences:

Giovedì 2 Marzo 2017, ore 14.30 - 15.30
Aula U1.1 (1st floor), DSCG - Via Campi 103

"Coupling image processing and chemometric resolution"
Dr. Siewert Hugelier - PhD student at Research Group in Photophysique, réactivité et fonctionnalité, Laboratoire de Spectrochimie Infrarouge et Raman, Université de Lille1, France.

Host: Prof. Marina Cocchi

Abstract 
The first part of the presentation involves the development and implementation of image processing methods as constraints to help in the decomposition of raw data into a bilinear chemically meaningful model. In order to apply Multivariate Curve Resolution - Alternating Least Squares (MCR-ALS) to hyperspectral imaging (HSI), the data is transformed from the three-dimensional data cube to a two-way matrix, and the information regarding pixel neighborhood is lost. We have recently shown that an additional transformation of the concentration profiles into the related distribution maps can be added to each least squares step to restore this information and global spatial constraints can be directly applied to improve the image modeling approach. Examples of such constraints are spatial smoothness, spatial segmentation, spatial filtering, etc. The talk will deal with the implementation of several of these approaches and show their influence on simulated and experimental data. The second part of the presentation talks about the use of penalized regression methods for the analysis of single-molecule fluorescence images. The analysis pipeline consists of different steps in which image processing and temporal approaches are combined to reveal the underlying structures of biological samples in diffraction limited data

CV
Siewert Hugelier is doing a PhD program in chemometrics at LASIR of the University of Lille in France under the supervision of Professor Cyril Ruckebusch. The main research lines of the group are: -Fundamental photophysics and photochemistry of photo-functional molecules -Molecular assemblies and Molecular confinement and photo-controlled reactivity in porous media In particular, Dr. Hugelier research focuses mainly on the resolution of ill-posed inverse problems in spectroscopy and chemical imaging for the processing of spectrokinetic or complex optical imaging data in the domains of optical spectroscopies resolved in vibrational time (Raman scattering, IR absorption) and electronic (absorption / diffuse reflection UV-vis, emission).

Recent Publications 

  • Sparse deconvolution of high-density super-resolution images S. Hugelier, J. De Rooi, R. Bernex, S. Duwé, O. Devos, M. Sliwa, P. Dedecker, P.H.C. Eilers, C. Ruckebusch Sci. Rep. 6 (2016) 21413-1 - 21413-10
  • Constraining shape smoothness in multivariate curve resolution-alternating least squares S. Hugelier, O. Devos, C. Ruckebusch J. Chemometrics 29 (2015) 448 - 456