Speckle noise removal in ultrasound medical imaging using envelope based time domain deconvolution

Abstract

The quality of the ultrasound medical images is usually degraded by two important problems - the transducer influence and the speckle noise. To remove the multiplicative speckle noise, in this paper it is presented a method based on a blind, time-domain, deconvolution algorithm. The proposed approach uses the envelope of the acquired radio-frequency signals for the numerical implementation. It is a two steps blind deconvolution, where firstly, the point spread function is automatically estimated from the measured data, and secondly, the data are reconstructed in a non-blind way using proposed algorithm. It has a nonlinear implementation which at each step intends to eliminate the most important part of speckle noise and blur as possible. The results on real image tests are compared with Tikhonov regularization. Our method shows better results for scatters detection and speckle noise removal.