Frequency equalized compounding for effective speckle reduction in medical ultrasound imaging

Frequency compounding (FC) is commonly used to reduce the speckle variance in order to enhance contrast resolution by averaging two or more uncorrelated sub-band images. However, due to the frequency dependent attenuation, the contrast resolution cannot be enhanced to the theoretical limit when imaging deep-lying tissue. In this paper, we propose the frequency equalized compounding (FEC) method to achieve contrast enhancement in the area of imaging as a whole. In this proposed method, a sub-band signal is divided into several zones along the imaging depth (or time), and the center frequencies and weighting factors for each zone are estimated; the estimated values are used in dynamic quadrature demodulation (DQDM) and image compounding respectively. The performance of the proposed method was evaluated through simulations and experiments. During the evaluation, the contrast resolution was quantified by speckle's signal-to-noise ratio (SSNR) in speckle regions and contrast-to-noise ratio (CNR) in hyper- and hypoechoic regions. Theoretical values of the SSNR and the CNR by the FC were computed by multiplying the SSNR and CNR values measured from the original image by N, where N is the number of sub-bands used in the compounding. From in vitro phantom experiments, it was learned that the SSNR and CNR values from the proposed method were similar to the theoretical values; the maximum and minimum errors from the theoretical value were 9% and 1% while those of the conventional FC (CFC) method were 25% and 7%. Similar results were obtained from the in vivo experiments with RF data acquired from the liver and the kidney. In addition, signal-to-noise ratio (SNR) improvement was measured. The SNR also improved due to the DQDM; maximum improvements for the in vitro and the in vivo experiments were 2.3 dB and 4.8 dB higher the results from the CFC method. These results demonstrate that the proposed FEC method can improve the contrast resolution up to a theoretically achievable value and may be useful in imaging technically difficult patients.