A NEW RESTORATION METHOD AND ITS APPLICATION TO SPECKLE IMAGES Ivana Duskunovic, Gjenna Stippel, Aleksandra Pizurica, Wilfried Philips, Ignace Lemahieu Dept. TELIN and ELIS, University of Gent, Sint-Pietersnieuwstraat 41, B-9000 Gent-Belgium Tel.: +32 9 264 34 15 id@telin.rug.ac.be ABSTRACT The visual interpretation of ultrasound brain images is a proven method to detect the White Matter Damage at an early stage. A problem, common to all medical ultrasound images, is the presence of speckle noise, which not only complicates the visual interpretation of images, but also quan- titative measurements. This paper proposes a new filter that removes a significant amount of speckle noise from ultra- sound images, while preserving details very well. The filter is based on a new cleaning technique that operates on the detail images of a wavelet decomposition. The papers il- lustrates that the proposed technique has some advantages over other popular techniques, i.e., the ones proposed by Lee, Frost, Malfait and Roose, and Sattar et al. 1. INTRODUCTION In this article we address the problem of noise suppression. Although image acquisition techniques yield an ever im- proving quality, there is always a need for post-processing methods to remove noise from obtained images. This is es- pecially difficult in the case of image dependant noise, e.g., speckle noise. The problem is not just to remove noise but to preserve valuable image data as well. Reduction of speckle noise is a very important post-processing step in medical imaging, including ultrasound imaging for purposes of seg- mentation and visual interpretation. Ultrasound imaging is the perfect means for morpho- logical investigation of the neonatal brain [1]. With real time ultrasound waves, most of the diseases can be discov- ered and followed, and this with a harmless non-invasive technique. In particular post-asphyctic damage, white mat- ter damage and matrix-bleedings, injuries that determine the future of many matured and pre-terminal neonaties, can be investigated well with ultrasound. Many techniques have been proposed for noise reduction in general and speckle This work was financially supported by the FWO (Fund for Scientific Research - Flanders), through a mandate of Research Assistant (Dimitri Van De Ville) and through the IWT-project (Institute for the Advancement of Scientific-Technological Research in Industry - Flanders) IWT 980155 and GOA-project (University of Ghent) 12.0513.98. Fig. 1. The neighboorhood mask. noise in particular. Two classes of techniques are especially successful at reducing speckle noise: 1. Techniques that are applied directly in the original image domain like the Lee [2] and Frost [3] filters. 2. Techniques that are applied in the wavelet domain like the technique developed by Malfait and Roose [4], which is a universal filter, and the technique proposed by Sattar, Floreby, Salomonsson and L¨ ovstr¨ om [5], designed especially for speckle noise reduction. In this paper we propose a new method for noise reduction in the wavelet domain. Originally, we developed the method for suppression of Gaussian noise. However, the results in this paper indicate that this method also removes speckle noise efficiently. The purpose of this paper is to introduce the new tech- nique and to demonstrate that it performs well on speckle noise of ultrasound images of the neonatal brain. The results show that the new technique performs favorably compared to some state-of-the-art techniques. 2. THE NEW METHOD The method proposed in this paper is based on the analysis of the detail images, obtained from the wavelet decompo- sition of the original image. The goal of the analysis is to determine the position of real edges in the image so that false edges and noise can be removed. The wavelet decomposition in our method is based on a discrete redundant wavelet decomposition with spline wavelets [6].