{
"title": "A New Hybrid RMN Image Segmentation Algorithm",
"authors": "Abdelouahab Moussaoui, Nabila Ferahta, Victor Chen",
"country": null,
"institution": null,
"volume": "12",
"journal": "International Journal of Medical, Health, Biomedical, Bioengineering and Pharmaceutical Engineering",
"pagesStart": 648,
"pagesEnd": 656,
"ISSN": "1307-6892",
"URL": "http:\/\/waset.org\/publications\/5907",
"abstract": "The development of aid's systems for the medical\r\ndiagnosis is not easy thing because of presence of inhomogeneities in\r\nthe MRI, the variability of the data from a sequence to the other as\r\nwell as of other different source distortions that accentuate this\r\ndifficulty. A new automatic, contextual, adaptive and robust\r\nsegmentation procedure by MRI brain tissue classification is\r\ndescribed in this article. A first phase consists in estimating the\r\ndensity of probability of the data by the Parzen-Rozenblatt method.\r\nThe classification procedure is completely automatic and doesn't\r\nmake any assumptions nor on the clusters number nor on the\r\nprototypes of these clusters since these last are detected in an\r\nautomatic manner by an operator of mathematical morphology called\r\nskeleton by influence zones detection (SKIZ). The problem of\r\ninitialization of the prototypes as well as their number is transformed\r\nin an optimization problem; in more the procedure is adaptive since it\r\ntakes in consideration the contextual information presents in every\r\nvoxel by an adaptive and robust non parametric model by the\r\nMarkov fields (MF). The number of bad classifications is reduced by\r\nthe use of the criteria of MPM minimization (Maximum Posterior\r\nMarginal).",
"references": null,
"publisher": "World Academy of Science, Engineering and Technology",
"index": "International Science Index 12, 2007"
}