A task-based analytical framework for ultrasonic beamformer comparison

Nghia Q. Nguyen, Richard W. Prager, Michael F. Insana
2016 Journal of the Acoustical Society of America  
A task-based approach is employed to develop an analytical framework for ultrasound beamformer design and evaluation. In this approach, a Bayesian ideal-observer provides an idealized starting point and a way to measure information loss in practical beamformer designs. Different approximations of this ideal strategy are shown to lead to popular beamformers in the literature, including the matched filter, minimum variance (MV), and Wiener filter (WF) beamformers. Analysis of the approximations
more » ... dicates that the WF beamformer should outperform the MV approach, especially in low echo signal-to-noise conditions. The beamformers are applied to five typical tasks from the BIRADS lexicon. Their performance is evaluated based on ability to discriminate idealized malignant and benign features. The numerical results show the advantages of the WF over the MV technique in general; although performance varies predictably in some contrast-limited tasks because of the model modifications required for the MV algorithm to avoid ill-conditioning.
doi:10.1121/1.4960607 pmid:27586736 fatcat:ldskuj66l5hctftdv6ykmooqp4