Texture feature analysis with fuzzy possibilistic c-means for brain MR image segmentation
Date Issued
2015
Date
2015
Author(s)
Hsieh, Chih-Chung
Abstract
Segmentation of brain tissue from non-brain tissue, also known as skull stripping, has been challenging due to the complexity of human brain structures and variable parameters of MR scanners. It is one of the most important preprocessing steps in medical image analysis. Skull stripping is often performed using a sequence of mathematical morphological operations following an initial separation of the brain from other tissues of the head. We propose a new brain segmentation algorithm that is based on a texture feature analysis, fuzzy possibilistic c-means and morphological operations. Tamura texture feature consist of six features. Gray Level Run-Length Matrices method is a comparably simple and straightforward texture analysis approach, and So does gray level co-occurrence matrix. Three methods are well-known and representative. After computation of textures, we apply fuzzy possibilistic c-means(FPCM) for voxel clustering, which provides a labeled image for the following morphological operations. The last step, we then apply sequence morphological operations followed by FPCM to find out the brain region. Our method starts from middle image to side because of the high accuracy in middle. We compare our methods with two famous methods, with internet brain segmentation repository data sets. Experimental results indicated that the proposed algorithm is effectively and potential application in a wide variety of brain image segmentation.
Subjects
Skull-stripping
texture-feature
image segmentation
brain
Tamura texture feature
GLCM
GLRLM
fuzzy possibilistic c-means
Type
thesis
File(s)![Thumbnail Image]()
Loading...
Name
ntu-104-R02525088-1.pdf
Size
23.54 KB
Format
Adobe PDF
Checksum
(MD5):a32ecb1d027eda24586e0b7ce523faea