Noise reduction of ultrasound prostate images using multi-scale products of the wavelet transform

Abstract

A noise reduction scheme based on multi-scale products of the dyadic discrete wavelet transform is proposed and a new automatic threshold finding strategy is defined, which assumes no a priori knowledge about the image structure or noise. The preliminary results of applying this scheme to prostate ultrasound images are promising. The important features of the original image are preserved while most of the speckle noise is removed.