A Spatially Adaptive Filter Reducing Arc Stripe Noise for Sector Scan Medical Ultrasound Imaging

Abstract

Sector scan medical ultrasound images usually have arc stripes, which do not represent the physical structure and thus are a kind of noise. This paper analyzes the source and characteristics of the arc stripes, and then proposes an adaptive filter based on the geometrical properties of these arc stripes. The proposed filter is the weighted summation of radially adaptive filter and common Gaussian filter. The radially adaptive filter aims to reduce the arc stripe noise. The common Gaussian filter is used to counteract the radial stripe artifact produced by the radial filter and suppress the randomly directional noise as well. The weights of the radially adaptive filter and common Gaussian filter are adapted to the proportion between the arc stripe noise and non-directional noise. The results show that the combined filter obviously enhances the image quality and is superior to common Gaussian filter. © Springer-Verlag 2004.