Bayesian segmentation of magnetic resonance images using the α-stable distribution

Authors:
Diego Salas-Gonzalez;Matthias Schlögl;Juan M. Görriz;Javier Ramírez;Elmar Lang
Affiliations:
Dpt. Signal Theory, Networking and Communications, ETSIIT-UGR, University of Granada, Granada, Spain;Dpt. Signal Theory, Networking and Communications, ETSIIT-UGR, University of Granada, Granada, Spain;Dpt. Signal Theory, Networking and Communications, ETSIIT-UGR, University of Granada, Granada, Spain;Dpt. Signal Theory, Networking and Communications, ETSIIT-UGR, University of Granada, Granada, Spain;Institute of Biophysics, University of Regensburg, Regensburg, Germany
Venue:
HAIS'11 Proceedings of the 6th international conference on Hybrid artificial intelligent systems - Volume Part I
Year:
2011

Citing 8
Cited 1

A parametric gradient descent MRI intensity inhomogeneity correction algorithm

Pattern Recognition Letters
Finite mixture of α-stable distributions

Digital Signal Processing
Bayesian mixture models of variable dimension for image segmentation

Computer Methods and Programs in Biomedicine
Editorial: Hybrid learning machines

Neurocomputing
A First Study on the Use of Coevolutionary Algorithms for Instance and Feature Selection

HAIS '09 Proceedings of the 4th International Conference on Hybrid Artificial Intelligence Systems
An adaptive field rule for non-parametric MRI intensity inhomogeneity estimation algorithm

Neurocomputing
Modelling with mixture of symmetric stable distributions using Gibbs sampling

Signal Processing
Editorial: Hybrid intelligent algorithms and applications

Information Sciences: an International Journal

Two fully-unsupervised methods for MR brain image segmentation using SOM-based strategies

Applied Soft Computing

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this work, a segmentation method of Magnetic Resonance images (MRI) is presented. On the one hand, the distribution of the grey (GM) and white matter (WM) are modelled using a mixture of astable distributions. A Bayesian a-stable mixture model for histogram data is used and the unknown parameters are sampled using the Metropolis-Hastings algorithm, therefore, voxel intensity information is included in the model via a parameterized mixture of a-stable distribution which allows us to calculate the likelihood. On the other hand, spatial information is also included: the images are registered to a common template and a prior probability is given to each intensity value using a normalized segmented tissue probability map. Both informations, likelihood and prior values, are combined using the Bayes' Rule. Performance of the segmentation approaches using spatial prior information, intensity values via the likelihood and combining both using the Bayes' Rule are compared. Better segmentation results are obtained when the latter is used.