Segmentation of brain magnetic resonance images through morphological operators and geodesic distancea

PASTORE, J.; MOLER, E.; BALLARIN, V.

DIGITAL SIGNAL PROCESSING

ACADEMIC PRESS INC ELSEVIER SCIENCE

Año: 2005 vol. 15 p. 153 - 160

1051-2004

When segmenting Magnetic Resonance (MR) images, a wide range of useless information arises, which has to be discarded as a step prior to classifying the different cerebral cortex areas. To obtain effective results during the classification process, it is necessary to work with images solely containing the brain and eliminate the cranium and surrounding meninges. This work introduces an automatic method for the detection of said structures based on the application of Morphology Alternating Sequential Filters by Reconstruction with structuring elements of growing size. Apart from enhancing and filtering in this way, this method captures the interior of a closed simple curve employing geodesic distance. Said curve represents the external brain boundary.