Essay On Brain Tissue

1285 Words6 Pages
Brain tissue can be classified into three tissue types; grey matter (GM), white matter (WM) and cerebro-spinal fluid (CSF). Initially, all MR images are segmented into three images, each represents one tissue type. The segmented images are then registered iteratively with their average to estimate the deformations that best align the images together. These deformations are used to generate spatially normalized and smoothed gray matter images. The image values after seg- mentation represent the probability of belonging to the tissue (GM or WM) of every voxel. These features generate one part of the feature vector In a parallel step, image texture features are extracted deriving the other part of the feature vector. Five texture features are…show more content…
producing a measure of the local balance between the count of gray matter and non-gray matter voxels). Second, the smoothing helps ameliorate the effects of misalignment of structures when the registration is imperfect. Third, it can increase sensitivity if the extent of smoothing matches the size of an effect of interest. Fourth, smoothing renders the data more Gaussian distributed, improving the validity of the commonly used Gaussian random field (GRF) theory thresholding approach. Typically between 4 and 16 mm full- width half maximum (FWHM) smoothing (with a Gaussian linear filter) is applied. Smoothing is done to correct noise and small variations. Finally, spatially normalized images are generated. every voxel of the image. These features generate one part of the feature vector 2) Texture features: Texture features reflect the regular changes of gray values in images. These changes in the values are correlated statistically and spatially. Textural feature vector is constructed from the gray level co-occurrence matrix. The co-occurrence matrix and texture features were initially used for for image classification by Haralick [24]. GLCM estimate image properties related to second-order statistics. It accounts for the spatial inter-dependency of two pixels at specific neighboring positions. GLCM is created from the high resolution grayscale NIFTI images. It is then normalized and used to calculate five different textural features; Contrast(1), Correlation(2),
Open Document