Backscatter coefficient estimation using array transducers

M.F. Insana, T.J. Hall, L.T. Cook
1994 IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control  
This paper describes an extension of our broadband method for estimating backscatter coefficients from random media to include data from array transducers. Four different transducer designs have now been considered: one-and twodimensional linear arrays, annular arrays, and single-element focused pistons commonly used in mechanical sector scanners. The analysis shows that if the backscatter echo spectrum is properly normalized, the shape of the piezoelectric elements affects only the magnitude
more » ... nly the magnitude and not the frequency dependence of the backscatter coefficient estimates. Experimental data were acquired using laboratory and clinical imaging instrumentation to verify the analysis. We compared backscatter coefficients measured as a function of frequency from well-defined scattering media that were obtained using a l -D linear array and focused piston transducers. Instrument-independent results were found that matched theoretical predictions within the measurement error between 2 and 12 MHz. We conclude from this study that accurate backscatter coefficient estimates may be easily obtained using current clinical imaging instrumentation. Applying the Huygens-Fresnel principle ([ 101, section 4. l ) , we consider the radiating surface of the transducer as a collection of point sources that oscillate sinusoidally. The net pressure field pu(r, t ) at the vector position (bold typeface) r: time t. and temporal frequency ul = 2~f , may be determined by summing the pressure fields from the oscillating sources at position r' (Fig. l) . If the sources lie entirely within the X', y' plane, the radiating surface is defined by the aperture distribution function a ( d , y'). The standard Fresnel approximations are applied to restrict the geometry of the experiment and thereby greatly simplify the following analysis ([lo], section 4.1). First, the distance between the radiator and observation planes z is much greater than the maximum linear dimension of the radiating area, i.e., z >> Max{d, y'}. Second, only a small region about the beam axis is of concern in the analysis, i.e., the paraxial 0885-3010/94$04.00 0 1994 IEEE Authorized licensed use limited to: University of Illinois. Downloaded on November 11, 2008 at 16:11 from IEEE Xplore. Restrictions apply.
doi:10.1109/58.308508 pmid:18263260 fatcat:4faimpdhmbb7taiga2b4ndrmua