期刊名称：International Journal of Hybrid Information Technology
印刷版ISSN：1738-9968
出版年度：2016
卷号：9
期号：12
页码：47-54
出版社：SERSC
摘要：Accurate segmentation of critical tissues in Magnetic Resonance (MR) Images is severely affected due to imperfections in the RF coil of the acquisition device. The problem becomes more prominent when there does not exist clear boundaries between various substructures. Hippocampus delineation from given MR image thus requires numerous preprocessing steps for correcting the intensity inhomogeneity resulting due to imperfections. Most of the time, extracted region will be partially containing the surrounded tissues. In this current work, an effective algorithm is developed aimed at segmenting the Hippocampus from the human brain MR image after performing the pre required bias correction. Before the given image is segmented, the image contrast is adjusted and then bias field is corrected. These steps are followed by clustering as well as region growing algorithms to finally yield effective results.
关键词：Bias field; Intensity in Homogeneity; level set approach; Gaussian Fit; ;Clustering
