A complex z-transform technique for speckle reduction

The ubiquitous speckle artifact manifests itself as a fine noise-like structure superimposed on an ultrasound image. Conventional speckle reduction techniques rely on a stochastic description of the artifact, but there are difficulties associated with such a definition. It is shown that a deterministic definition of a speckle corrected signal is feasible and can lead to a new method of removing the speckle artifact. The algorithm implementing this is presented and consists of two stages. Stage 1 deterministically identifies a local center of destructive interference, and stage 2 provides an additive local correction to (only a subset of) the RF-Aline data. Processing operates on a one-dimensional data sequence via complex z-transform techniques. No resolution tradeoff exists, and the scheme is shown to be robust to noise and ultimately applicable on a real time-basis. The efficacy of the technique is assessed via B-mode images obtained from a tissue-equivalent phantom and suppression of partially developed speckle is demonstrated. Although the algorithm is presented in the medical ultrasonic context, it is applicable to all imaging modalities that are corrupted by speckle and implemented via a coherent detection system