[Imageworld] Post-doc position at INRIA on Sparse representation
for Brain image analysis
Christian.Barillot at irisa.fr
Fri Feb 19 17:58:32 CET 2010
Post-doctoral Position at INRIA/IRISA Rennes, France in collaboration between the Unit/Project VISAGES U746 (INSERM/INRIA/CNRS/university of Rennes I) and the project-team METISS (INRIA/CNRS). During her/his post-doc, the researcher will work more precisely with P.Maurel and C.Barillot for VISAGES and R.Gribonval for METISS. This work will benefit from a new research 3T MRI system on which new research protocols will be set up.
title: SPARSE REPRESENTATION OF LARGE MEDICAL DATASETS FOR MRI BRAIN IMAGE ANALYSIS
Application should be conducted through this web site (send application before March 30th):
The objective of this post-doc position is to elaborate a new dimensionality reduction framework based on sparse representations which will take into account the redundancy of information embedded in MRI multidimensional and longitudinal images. The goal is to achieve a more suitable representation of the data for image analysis and detection of features specific to a form or a grade of pathology or specific to a population of subjects (e.g. patients vs normal control). A first idea could be to apply multimodal joint sparse decomposition techniques based, for example, on a wavelet dictionary. Further developments could include a dictionary-learning step, which adapts the dictionary to a specific class of images or to a specific class or form of pathology.
Concerning the experimental part, this project can address a very large scale of application in the domain of neurological, neurodegenerative, psychiatric and neurodevelopmental brain diseases (Strokes, Dementia, Alzheimer, Parkinson Diseases, Depression, Dysphasia…). More particularly, we will first focus our application objectives toward multiple sclerosis and dysphasia.
Keywords: Image Analysis, Medical Imaging, Sparse representation, Statistical Learning, Dimensionality reduction,
Competence and Profile:
This position requires background in applied mathematics, numerical analysis, and statistics as well as in image processing. A good practice in computer sciences, especially in Matlab or in object oriented programming (C++), and an experience in medical imaging will be appreciated.
We refer the candidates to the following websites for more information:
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Imageworld