A Simulation Study of Echographic Imaging of Diffuse and Structurally Scattering Media

Abstract

Realistic simulations of echographic image formation were performed. The simulations were based on a focussed single element transducer of 3.5 MHz and linear scanning. The tissue model scanned was composed of a homogeneous, nonattenuating, medium containing point-like scatterers. The scatterers were distributed in space in two different arrangements: randomly with a fixed number density of 7500 per cubic centimeter and regularly in a cubic matrix with a characteristic distance of 1 mm. These two populations of scatterers yielded the diffuse and the structural scattering component, respectively. The B-mode texture was assessed by first and second order statistical parameters. The effects of various modifications of the matrix, i.e., the relative scattering strength, the orientation and the position uncertainty of the scatterers, on the statistical parameters were systematically investigated. Increasing the relative scattering strength yields a monotonic increase of the mean grey level, the ratio of structural over diffuse scattering intensity and a decrease of the signal-to-noise ratio and the autocorrelation lengths. Increasing the position uncertainty to 20 percent reduces these effects practically to zero, except for the lateral ACF's. Rotation of the matrix with respect to the beam axis induces a periodicity in all statistical parameters, which is symmetric around 45 degrees. Again, the effects become statistically insignificant at a position uncertainty of 20 percent, with the exception of length of the lateral ACF. While assuming that in a single clinical B-mode echogram of the liver many orientations of the matrix will be simultaneously present, the B-mode lines of images obtained for a range of orientations were taken together and the overall statistical parameters calculated. The mean, the signal-to-noise ratio of the echo amplitude, SNRA, and the axial and lateral lengths of the ACF for the multirotation condition are significantly different from both the diffuse and the zero degree orientation. The structural backscattering intensity cannot be estimated any more with a reasonable precision. The structural separation distance, however, can still be assessed at a 10 percent level of the position uncertainty, even in the multirotation case and is therefore a robust clinical parameter.