Signal and Image Restoration |
|
Members
Project description
One of the major limitation on US image quality is reduced resolution due to both the effect of limited effective aperture size and noise.
Moreover, since the US transducer pulse ntroduces an unwanted spectral shaping of the backscattered echo signal, econvolution is used to
eliminate this effect and to obtain the pure tissue esponse.
We implemented deconvolution method based on efficient equalization
techniques usually applied in Digital Communications: the ultrasonic RF signal is considered as a sequence of discrete values (symbols)
affected by channel intersymbol interference (ISI), and processed with a reduced-complexity Viterbi algorithm which is an optimum solution
for the ML estimation of sequences in presence of ISI and white gaussian noise. Due to the electromechanical properties of the
crystals, the geometry of the system, the focusing used in the imaging process, and the frequency dependant attenuation opposed by the
tissue, the transduceres response, i.e. the channel, is highly non-stationary and needs to be estimated.In our research group we
implemented algorithms for minimum and non-minimum phase, mono and bidimensional channel estimation, performing the non trivial
task of phase unwrapping within a bayesian framework. Spatial variations of the channel are tracked by coupling the Viterbi
algorithm with adaptive-filtering techniques. Contact: Nicola Testoni >
This e-mail address is being protected from spam bots, you need JavaScript enabled to view it
Collaborations
|