Noise reduction and image enhancement of MRI using adaptive multiscale data condensation

MR images often suffer from different types of noises, artifacts and inhomogeneities due to diverse reasons which hinders the image understanding process. Here we have proposed a method of MR image denoising and image enhancement using adaptive multiscale data condensation (MDC) strategy. The strategy is based on the selection of representative neighbourhood on the basis of their information content with respect to the entire filter mask. The information content is computed according to their feature and relative location within a search window. Each pixel is evaluated and subsequently modified, if required, in terms of its feature content and weightage with respect to its neighbour. Parameter calculated in one mask window continues to the adjacent mask to get adaptively adjusted according to the local requirement. The strategy is studied over MR brain images with Rician noise and the performance is compared with noise removal using Wiener filter and wavelet transformation based noise reduction and reconstruction tools.