Mean Scatterer Spacing Estimation Using Cepstrum-Based Continuous Wavelet Transform

The goal of this study was to develop an ultrasound (US) scatterer spacing estimation method using an enhanced cepstral analysis based on continuous wavelet transforms (CWTs). Simulations of backscattering media containing periodic and quasi-periodic scatterers were carried out to test the developed algorithm. Experimental data from HT-29 pellets and in vivo PC3 tumors were then used to estimate the mean scatterer spacing. For simulated media containing quasi-periodic scatterers at 1-mm and 100-μm spacing with 5% positional variation, the developed algorithm yielded a spacing estimation error of ~1% for 25and 55-MHz US pulses. The mean scatterer spacing of HT-29 cell pellets (31.97 μm) was within 3% of the spacing obtained from histology and agreed with the predicted spacing from simulations based on the same pellets for both frequencies. The agreement extended to in vivo PC3 tumors estimation of the spacing with a variance of 1.68% between the spacing derived from the tumor histology and the application of the CWT to the experimental results. The developed technique outperformed the traditional cepstral methods as it can detect nonprominent peaks from quasi-random scatterer configurations. This work can be potentially used to detect morphological tissue changes during normal development or disease treatment.

I. Introduction

The backscattered ultrasound (US) signals encode information on scatterer structure with applications in diagnosis of different conditions affecting soft tissues [1], imaging of blood flow [2], and the detection of pathophysiological insults that alter normal tissue structure. This is the premise behind the quantitative US (QUS) [3], which can probe structural tissue changes during therapies such as microbubble-enhanced radiation [4] with the goal of clinically characterizing tumor response to chemoradiotherapy [5], [6].