Reduction of noise in tomographic images using multiwavelet techniques

Abstract

Research in the field of tomography focuses on the techniques for efficient and accurate reconstruction of an image from the noisy degraded image. The problem of reconstructing an unknown image from sparse and noisy measurements is an ill-posed inverse problem. Inverse problems estimate some parameter or data from a set of indirect noisy observations. Due to insufficient information, solution to reconstruction problems are typically non-unique. To tackle the inherent ambiguity of reconstructing the problem solution, we need to incorporate estimation of apriori information, about the structure of desired solution set. Reduction of noise in medical imaging is not an easy task. Noise should be reduced in such away that the features of an image must be retained. Medical images have large variations, for which it should be dealt case wise. This motivated us to reconstruct new techniques that preserves the important features of medical images. We have chosen medical tomographic imaging area, to denoise anatomical structures such as human abdomen, thorax, head phantom etc. The denoising process should not destroy anatomical details. Medical imaging is used for clinical diagnosis and computer aided surgery. A number of researchers have published image denoising literature from past many years but excellent noise reduction is still a challenge in improving the image quality. In view of this, we have developed several algorithms using multiwavelet techniques and are outlined as given below. Our research is focused on improvements in tomographic images.