Multi-resolution approach to strain imaging
Abstract
An essential goal of a strain imaging experiment is to balance the desire to apply large deformations, and thereby maximize contrast, with the need to maintain waveform coherence between echoes recorded before and after deformation, and thereby minimize decorrelation noise. The tools available to achieve an optimal balance include control of boundary conditions and application of reconstruction algorithms that use all the available waveform information. This paper describes our work with a multi-resolution strain algorithm employing the maximum-likelihood strategy of filtering and waveform warping at several spatial scales. We found that echo waveform filtering improves the strain CNR when the echo SNR>30 dB and the applied strain >5%. We found that envelope-detected echoes provide more robust estimates than low-pass filtered rf echoes under most practical conditions. The benefits of echo-signal filtering for minimizing decorrelation noise are modest.