On the convergence of EM-like algorithms for image segmentation using Markov random fields

Inference of Markov random field images segmentation models is usually performed using iterative methods which adapt the well-known expectation-maximization (EM) algorithm for independent mixture models. However, some of these adaptations are ad hoc and may turn out numerically unstable. In this paper, we review three EM-like variants for Markov random field segmentation and compare their convergence properties both at the theoretical and practical levels. We specifically advocate a numerical scheme involving asynchronous voxel updating, for which general convergence results can be established. Our experiments on brain tissue classification in magnetic resonance images provide evidence that this algorithm may achieve significantly faster convergence than its competitors while yielding at least as good segmentation results.

[1]  Michael I. Jordan.  Learning in Graphical Models , 1999, NATO ASI Series.

[2]  Koenraad Van Leemput,et al.  A unifying framework for partial volume segmentation of brain MR images , 2003, IEEE Transactions on Medical Imaging.

[3]  Mark W. Woolrich,et al.  Variational bayes inference of spatial mixture models for segmentation , 2006, IEEE Transactions on Medical Imaging.

[4]  Stephen M. Smith,et al.  Segmentation of brain MR images through a hidden Markov random field model and the expectation-maximization algorithm , 2001, IEEE Transactions on Medical Imaging.

[5]  Meritxell Bach Cuadra,et al.  Comparison and validation of tissue modelization and statistical classification methods in T1-weighted MR brain images , 2005, IEEE Transactions on Medical Imaging.

[6]  Jun Zhang.  The mean field theory in EM procedures for Markov random fields , 1992, IEEE Trans. Signal Process..

[7]  Geoffrey E. Hinton,et al.  A View of the Em Algorithm that Justifies Incremental, Sparse, and other Variants , 1998, Learning in Graphical Models.

[8]  D. Louis Collins,et al.  Design and construction of a realistic digital brain phantom , 1998, IEEE Transactions on Medical Imaging.

[9]  Richard Szeliski,et al.  A Comparative Study of Energy Minimization Methods for Markov Random Fields with Smoothness-Based Priors , 2008, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[10]  Karl J. Friston,et al.  Bayesian fMRI time series analysis with spatial priors , 2005, NeuroImage.

[11]  Stan Z. Li,et al.  Markov Random Field Modeling in Computer Vision , 1995, Computer Science Workbench.

[12]  M. Opper,et al.  Comparing the Mean Field Method and Belief Propagation for Approximate Inference in MRFs , 2001 .

[13]  Olga Veksler,et al.  Fast Approximate Energy Minimization via Graph Cuts , 2001, IEEE Trans. Pattern Anal. Mach. Intell..

[14]  D. Rubin,et al.  Maximum likelihood from incomplete data via the EM - algorithm plus discussions on the paper , 1977 .

[15]  Jun Zhang.  The convergence of mean field procedures for MRFs [image processing] , 1996, IEEE Trans. Image Process..

[16]  Zhengrong Liang,et al.  An EM Approach to MAP Solution of Segmenting Tissue Mixtures: A Numerical Analysis , 2009, IEEE Transactions on Medical Imaging.

[17]  J. Besag.  Spatial Interaction and the Statistical Analysis of Lattice Systems , 1974 .

[18]  Srinivas Yarramalle,et al.  Unsupervised Image Segmentation Method based on Finite Generalized Gaussian Distribution with EM & K-Means Algorithm , 2007 .

[19]  Koenraad Van Leemput,et al.  Automated model-based tissue classification of MR images of the brain , 1999, IEEE Transactions on Medical Imaging.

[20]  P. Narayana,et al.  Compensation for surface coil sensitivity variation in magnetic resonance imaging. , 1988, Magnetic resonance imaging.

[21]  William T. Freeman,et al.  On the optimality of solutions of the max-product belief-propagation algorithm in arbitrary graphs , 2001, IEEE Trans. Inf. Theory.

[22]  D. Louis Collins,et al.  A new improved version of the realistic digital brain phantom , 2006, NeuroImage.

[23]  James C. Bezdek,et al.  Some Notes on Alternating Optimization , 2002, AFSS.

[24]  W. Eric L. Grimson,et al.  Adaptive Segmentation of MRI Data , 1995, CVRMed.

[25]  Martin J. Wainwright,et al.  MAP estimation via agreement on trees: message-passing and linear programming , 2005, IEEE Transactions on Information Theory.

[26]  Koenraad Van Leemput,et al.  Automated model-based bias field correction of MR images of the brain , 1999, IEEE Transactions on Medical Imaging.

[27]  Tina Kapur,et al.  Model based three dimensional medical image segmentation , 1999 .

[28]  Thomas P. Minka,et al.  Divergence measures and message passing , 2005 .

[29]  Jeffrey A. Fessler.  Comments on "The convergence of mean field procedures for MRF's" , 1998, IEEE Trans. Image Process..

[30]  Donald Geman,et al.  Stochastic Relaxation, Gibbs Distributions, and the Bayesian Restoration of Images , 1984, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[31]  R. Leahy,et al.  Magnetic Resonance Image Tissue Classification Using a Partial Volume Model , 2001, NeuroImage.

[32]  Yair Be'ery,et al.  Convergence analysis of turbo decoding of product codes , 2001, IEEE Trans. Inf. Theory.

[33]  Rick Best,et al.  Design and Construction , 2007 .

[34]  Alan C. Evans,et al.  MRI Simulation Based Evaluation and Classifications Methods , 1999, IEEE Trans. Medical Imaging.

[35]  Alan C. Evans,et al.  A nonparametric method for automatic correction of intensity nonuniformity in MRI data , 1998, IEEE Transactions on Medical Imaging.

[36]  Edgar Arce Santana,et al.  Hidden Markov Measure Field Models for Image Segmentation , 2003, IEEE Trans. Pattern Anal. Mach. Intell..

[37]  Gersende Fort,et al.  Combining Monte Carlo and Mean-Field-Like Methods for Inference in Hidden Markov Random Fields , 2007, IEEE Transactions on Image Processing.

[38]  Nicholas Ayache,et al.  Maximum Likelihood Estimation of the Bias Field in MR Brain Images: Investigating Different Modelings of the Imaging Process , 2001, MICCAI.