Open AccessStatistical characterization of ischemic stroke lesions from MRI using discrete wavelet transformations
Author(s) -
R. Karthik,
R. Menaka
Publication year - 2016
Publication title -
ecti transactions on electrical eng. / electronics and communications
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.148
H-Index - 7
ISSN - 1685-9545
DOI - 10.37936/ecti-eec.2016142.171142
Subject(s) - wavelet , segmentation , magnetic resonance imaging , artificial intelligence , lesion , pattern recognition (psychology) , texture (cosmology) , transformation (genetics) , computer science , range (aeronautics) , wavelet transform , radiology , medicine , image (mathematics) , pathology , materials science , biology , biochemistry , composite material , gene
The segmentation and characterization of lesion structures from brain Magnetic Resonance Imaging (MRI) slices serves to recognize the degree of the influenced tissues for effective diagnosis and planning in the treatment of ischemic stroke. The different portions of the affected tissues might exhibit different properties in the different imaging modalities. Hence, developing a fully-automatic approach for segmentation of these abnormal structures is considered to be a challenging research issue in medical image processing. This research applies the discrete wavelet transformation of different types for characterizing the properties of the lesion structures from MRI images. The wavelet co-efficients were determined for different levels and the statistical parameters were extracted from it for characterizing the texture properties of the brain tissues. The experimental results were presented for both normal and abnormal MRI datasets. Observations indicate that there was a clear demarcation between the range of values in the statistical features obtained for normal and abnormal images.