Multiresolution imaging in elastography

摘要

The range of strains that can be imaged by any practicalnelastographic imaging system is inherently limited, and a performancenmeasure is valuable to evaluate these systems from the signal and noisenproperties of their output images. Such a measure was previouslynformulated for systems employing cross-correlation based time-delaynestimators through the strain filter. While the strain filter predictsnthe signal-to-noise ratio (SNRe) for each tissue strain innthe elastogram and provides valuable insights into the nature of imagennoise, it understated the effects of image resolution (axial resolution,nas determined by the cross-correlation window length) on the noise. Innthis work, the strain filter is modified to study the strain noise atnmultiple resolutions. The effects of finite window length on signalndecorrelation and on the variance of the strain estimator areninvestigated. Long-duration windows are preferred for improvednsensitivity, dynamic range, and SNRe. However, in this limitnthe elastogram is degraded due to poor resolution. The results indicatenthat for nonzero strain, a window length exists at which the variance ofnstrain estimator attains its minima, and consequently the elastographicnsensitivity, dynamic range and SNRe are strongly affected bynthe selected window length. Simulation results corroborate thentheoretical results, illustrating the presence of a window length wherenthe strain estimation variance is minimized for a given strain value.nMultiresolution elastography, where the strain estimate with the highestnSNRe obtained by processing the pre- and post-compressionnwaveforms at different window lengths is used to generate a compositenelastogram and is proposed to improve elastograms. All the objectivenelastogram parameters (namely: SNRe, dynamic range,nsensitivity and the average elastographic resolution-defined as thencross-correlation window length) are improved with multiresolutionnelastography when compared to the traditional method of utilizing ansingle window length to generate the elastogram. Experimental resultsnusing a phantom with a hard inclusion illustrates the improvement innelastogram obtained using multiresolution analysis