Volatility Clustering in Medical Ultrasound Imaging and System Identification Based Deconvolution

We propose a system identification based deconvolution algorithm for medical ultrasound imaging enhancement. Deconvolution is realized using Multivariate Autoregressive (MVAR) models taking into account the heteroscedastic nature of the recorded ultrasonic signals. Based on simulations, phantoms and in vivo data we show the superiority of the proposed algorithm under the presence of volatility in the obtained images.