Medical Image Denoising using X-lets

Abstract

Wavelets are proved to be well adapted for 1-D signal but can only capture limited directional information in 2D due to its poor orientation selectivity. Transforms like curvelets and contourlets have very high degree of directional specificity which is necessary for medical images. These transforms are based on certain anisotropic scaling principle which is quite different from the isotropic scaling of wavelets. Simulation test carried out on medical images like ultrasound images, magnetic resonance images and computerized tomography scan images, show that better denoising results were obtained by curvelets and contourlets, than with wavelets, in terms of mean square error, signal to noise ratio and visual evaluation