SU‐E‐J‐108: Texture Segmentation in Magnetic Resonance Images Using Discrete Wavelet Transform Combined with Gabor Wavelets

Abstract

Purpose: Edge detection improves image readability and plays an important role in images preprocessing aimed to their segmentation and automatic recognition of their contents. The purpose of this study was to describe methods of edge detection in magnetic resonance images, with the emphasis on the use of discrete wavelet transform (DWT) combined with Gabor wavelets. Methods: Modulus maxima method by Mallat S (A Wavelet Tour of Signal Processing. Academic Press, 1998), provides the method for edge detection using wavelet transform. This method is based on finding local maxima of horizontal and vertical wavelet coefficients in the first level of wavelet decomposition because it is supposed that this level represents edges. This method was tested with various wavelet functions both on simulated and real medical images. A multiresolution approach using undecimated wavelet transform is also employed which allows the low‐low (LL), low‐high (LH), high‐low (HL), and high‐high (HH) sub‐bands to remain at full size. A simple peak finding algorithm is used to determine the peaks out of array of these texture features. Results: Using wavelet transform method, the decomposition was performed up to two levels. Gabor filters are then applied to the wavelet approximations at all levels to obtain the characteristic texture features such as entropy, second to fourth central moments and coefficient of variation. High values of the second central variance and fourth central variance signify images in which regions can be clearly differentiated. The corresponding filter outputs are compared to obtain an image containing minimum pixel values. Conclusion: A complex wavelet function could help to improve results of edge detection in real images. A comparison of basic edge detection methods including simple gradient operators and Gabor wavelets, and their combination with wavelet transform was presented. Mathematical principals were studied, as well as application of these methods.

Differentiating Grade in Breast Invasive Ductal Carcinoma Using Texture Analysis of MRI

“…Firstly, a breast cancer segmentation model based on DWT and the K-means algorithm is proposed. Secondly, TA was performed, and the Gabor wavelet analysis was used to extract the texture feature of MR images [12]. Then, according to the distance relationship between the features, key features are sorted and feature subsets are selected.…”

“…This section presents the Gabor wavelet analysis of the ROIs of a tumor image for extracting the texture features [12]. The Gabor wavelets have a tunable orientation, radial scale bandwidths, and tunable center scales, allowing them to optimally achieve joint resolution in the spatial and frequency domains.…”

Perception-based fuzzy partitions for visual texture modeling

“…The majority of the image analysis techniques in computer vision try to model texture by means of feature vectors (that usually have very large dimensions) which have no direct relationship with the different perceptual properties. Most of these techniques are based on multiresolution analysis and scale-space theory, such as Gabor functions [12,13,14,15,16,17] or Wavelets [18,19,20,21,22,23], that are considered as the golden standard in the literature. In addition, general image classification and feature learning techniques can be also applied in texture analysis, such as techniques based on kernel learning [24,25,26,27,28,29], dictionary learning [30,31,32,33] or genetic programming [34,35,36,37,38,39].…”