A number of diseases result in reduced perfusion of the vital organs of the body. Measures of the blood flow in the small vessels (the perfusion) of these organs would contribute significantly to a clinical diagnosis. Myocardial contrast echocardiography is a rapidly developing technique for non-invasive ultrasound imaging of perfusion. The work presented in this thesis is related to the preprocessing of the ultrasound data in order to facilitate the segmentation of the heart walls in which the perfusion is to be estimated.

In an ultrasound image artifacts can occur for a number of reasons. These lead to a more complicated interpretation procedure as important information may be obscured or lost. One of the most frequently occuring artifacts in diagnostic ultrasound imaging is reverberation. This type of artifact is caused by the incorrect assumption that the ultrasound pulse's propagation path is always straight to a backscattering target and straight back to the transducer. Reverberation artifacts may be seen as multiple ghost structures in the ultrasound image.

A method to detect these reverberation artifacts has been developed and tested on three different ultrasound image sequences. The method uses ratios of lognormal quadrature filter responses to estimate the local bandwidth of the ultrasound signal. For the in-vivo sequences the result is a "map" indicating the artifact locations.

In order to reduce the reverberation artifacts an interactive procedure for estimating two­ dimensional and three-dimensional Wiener filters has been developed. The design procedure used is a weighted least squares approach, in which both a frequency and a spatiotemporal weighting function are used to optimize the filter. In the results after filtering with the optimized filters the magnitude of the artifact is significantly reduced. For the in-vivo sequences the 3D approach display a better selection capability compared to the 2D approach.

The results from these studies show that the proposed methods constitute a promising basis for a useful clinical tool.