Tosaporn Nilmanee and Pornchai Phukpattaranont

pp. 181 - 187

Abstract

We present a novel design of quadratic filters (QFs) in the frequency domain in order to improve the quality of contrastassisted ultrasound images for medical diagnosis. The QF is designed as a 2D linear-phase filter. In addition, the magnitude is
based on the sum of two 2D Gaussian filters. The centers of the Gaussian filters are placed at the locations where the power
strength of signals from ultrasound contrast agent over surrounding tissue is maximal. The design parameters consist of two
centers and a standard deviation (SD) of the Gaussian filters. The coefficients of the QF are obtained using the inverse discrete
Fourier transform. The QFs from the proposed design method are evaluated using in vivo ultrasound data, i.e., the kidney of a
guinea pig. We find that the appropriate SD and two center points of the QF for the in vivo data are at 0.34, (3.30, 3.30) and (-3.30,
-3.30) MHz, respectively. Results show that the images produced from the output signals of the new design are superior to the
original B-mode both in terms of contrast and spatial resolution. The quadratic image provides clear visualization of the
kidney shape and large vascular structures inside the kidney. The contrast-to-tissue ratio value of quadratic image is 24.8 dB
compared to -1.5 dB from the B-mode image. In addition, we can use this new design approach as an efficient tool to further
improve the QF in producing better contrast-assisted ultrasound images for medical diagnostic purposes.