Enhancement of Proton Density Weighted Magnetic Resonance Images using Singular Value Decomposition in Wavelet Domain

K. Kannan

Abstract

Image enhancement techniques for low contrast medical images using Singular Value Decomposition (SVD) in Discrete Wavelet Transform domain (SVD-DWT) are proposed in the literature. However, shift sensitivity, poor directionality and a lack of phase information are the primary drawbacks of the discrete wavelet transform. This work introduces a novel method of image enhancement (SVD-SWT) using SVD in Stationary Wavelet Transform (SWT) to improve the contrast of Proton Density weighted Magnetic Resonance Images (PD weighted MRI) while preserving the brightness. The low frequency sub bands of PD weighted MRI and the low frequency sub bands of General Histogram Equalized PD weighted MRI are used in the proposed method as a weighted sum of singular value matrices to enhance the contrast of PD weighted MRI. The proposed method is compared with several histogram equalisation methods and improvement techniques employing Singular Value Decomposition and Discrete Wavelet Transform (SVD-DWT). Discrete Entropy (DE), Average Brightness (AB), Pixel Distance (PD), and Standard Deviation (SD) all showed that the proposed method is the best.

Abstract

Keywords