A multitaper approach to speckle reduction for medical ultrasound imaging

A typical approach to reduce unwanted speckle patterns in coherent imaging systems is to average images with different speckle realizations. Traditionally, these images have been formed by observing the target region from slightly different angles (spatial compounding) or by varying the involved frequencies (frequency compounding). In this paper we investigate a third option based on Thomson's multitaper approach to power spectrum estimation applied in the spatial domain. Our investigations, based on both real, measured ultrasound data and data from simulated scenes of point scatterers, verify that the multitaper approach can be used for speckle reduction at a rate compared to that of the more traditional method of spatial compounding. Due to the spectral concentration of the tapers, an added benefit is reduced sidelobe levels, which can yield sharper edges and more defined cyst-like structures.