Markov Random Field Segmentation of Brain MR Images
Karsten Held, Elena Rota Kops, Bernd J. Krause, William M. Wells, III, Ron Kikinis, Hans-Wilhelm Müller-Gärtner · 1997
We describe a fully-automatic 3Dsegmentation technique for brain MR images. By means of Markov random fields the segmentation algorithm captures three features that are of special importance for MR images: nonparametric distributions of tissue intensities, neighborhood correlations and signal inhomogeneities. Detailed simulations and real MR images demonstrate the performance of the segmentation algorithm. In particular the impact of noise, inhomogeneity, smoothing and structure thickness is analyzed quantitatively. Even singleecho MR images are well classified into grey matter, white matter, cerebrospinal fluid, scalpbone, and background. A simulated annealing and an iterated conditional modes implementation are presented. Index Terms Magnetic Resonance Imaging, Segmentation, Markov Random Fields I. INTRODUCTION Excellent soft-tissue contrast and high spatial resolution make magnetic resonance imaging the method for anatomical imaging in brain research. Segmentation of the MR imag...