Non-minimum phase iterative deconvolution of ultrasound images

The strongest limitation to ultrasound image quality is due to the blurring effect on the back-scattered echo produced by the echographic transducer response. This effect alters significantly the received echo, reducing the resolution of the echographic image. As under proper general assumptions a convolution-based model can be used to represent the radio-frequency incoming echo signal, fast and robust deconvolution algorithms can be successfully employed to improve image quality by means of attenuating the unwanted transducer effects. In this work we propose an iterative deconvolution algorithm designed to deal with both non-minimum phase transducer impulse responses and scattering events not aligned with the sampling grid. This is achieved by means of analytically-designed all-pass filtering stages. Sparse solutions of the deconvolution problem as well asBernoulli-Gaussian output sequences are particularly favoured thanks to the adopted approach. Deconvolution performances, evaluated in terms of axial resolution gain, peak signal to noise ratio, quality index and contrast gain over a data-set of phantoms and in-vivo images show that our algorithm features very good results when compared to other in literature.