HAL will be down for maintenance from Friday, June 10 at 4pm through Monday, June 13 at 9am. More information
Skip to Main content Skip to Navigation
Conference papers

Mélanges de lois de Student à Échelles Multiples pour la caractérisation de tumeurs par IRM multiparamétrique

Alexis Arnaud 1 Florence Forbes 1 Benjamin Lemasson 2 Emmanuel Barbier 2 Nicolas Coquery 2
1 MISTIS - Modelling and Inference of Complex and Structured Stochastic Systems
Inria Grenoble - Rhône-Alpes, LJK - Laboratoire Jean Kuntzmann, Grenoble INP - Institut polytechnique de Grenoble - Grenoble Institute of Technology
2 NeuroImagerie Fonctionnelle et Perfusion Cérébrale
GIN - [GIN] Grenoble Institut des Neurosciences
Abstract : In this study we develop a statistical method for the classification of multiparametric MRI, which allows data quality control (atypical observation detection), and provides a tumour signatures dictionary. A previous study has been based on a Gaussian mixture model in which observations are grouped into classes corresponding to Gaussian distributions. This model is known for its sensitivity to outliers which can degrade the relevance of the obtained groups. As an alternative, we propose to use generalized Student distributions which extend the standard multivariate Student distribution by allowing different weights on the different dimensions of each observation.
Document type :
Conference papers
Complete list of metadata

Contributor : Alexis Arnaud Connect in order to contact the contributor
Submitted on : Monday, January 11, 2016 - 9:43:21 AM
Last modification on : Thursday, January 20, 2022 - 5:30:20 PM
Long-term archiving on: : Tuesday, April 12, 2016 - 11:06:17 AM


Files produced by the author(s)


  • HAL Id : hal-01253588, version 1



Alexis Arnaud, Florence Forbes, Benjamin Lemasson, Emmanuel Barbier, Nicolas Coquery. Mélanges de lois de Student à Échelles Multiples pour la caractérisation de tumeurs par IRM multiparamétrique. 47èmes Journées de Statistique de la SFdS, Jun 2015, Lille, France. ⟨hal-01253588⟩



Record views


Files downloads