Fundamental performance metrics and optimal image processing strategies for ultrasound systems

Fundamental limits on imaging system performance are developed using Bayesian signal detection theory. The analysis expands upon the wellknown theory of Smith and Wagner. Envelope-detected signals are shown to be sub-optimal for detection tasks. Two image processing strategies are presented that may improve upon current B-mode processing: deconvolution and wavefront curvature matched filtering. The later technique takes advantage of remarkable spatial-frequency bandwidths available in curved wavefronts to obtain high quality images. Both techniques attempt to approach the fundamental limits of performance by mimicking the strategy of the ideal observer.