Detection of subresolution sources of coherent scattering for parametric image formation

This work investigates the performance of speckle statistics parameters to detect coherent scattering with stationary features in echo signals. We focus on the task of parametric image formation for medical applications and the implicit tradeoff between spatial resolution and detection performance. Parameters include the model-free envelope and intensity point-wise SNRs, Nakagami shape parameter m and a Generalized Likelihood Ratio Test T, and the Homodyned-K clustering parameter and the ratio of coherent to incoherent scattering amplitudes. Media with periodic scatterers as a source of coherent scattering embedded in a diffuse scattering background were simulated and mimicked in phantoms. Model-free parameters offered better detection in terms of the parameters' contrast-to-noise ratio with respect to a medium with diffuse scattering conditions. This was observed for a wide range of spatial resolutions of parametric images, scatterer spacing values, and regardless of distorting the periodic array or reducing the scattering power of the periodic scatterers. Poorest performance was achieved with the Homodyned-K parameters, although their physical insight motivates their use when compounding methods are available or for the task of bulk parameter estimation. These results are to be merged with the detection of coherent scattering with non-stationary features and corresponding scatterer structure descriptors, as well as detection of false coherence from low scatterer number densities.