Tissue Characterization and Attenuation Mapping Using Multifrequency Ultrasound Imaging

Attenuation estimation and correction plays an essential role in improving quantitative ultrasound (US) performance, producing stronger tissue characterization and diagnostic accuracy. Common methods estimate one global attenuation coefficient for a defined region-of-interest (ROI), which can reduce the precision of correction. We propose a novel, pixel-based, 2D attenuation mapping technique utilizing multiple frequency-shifted narrowband transmit pulses, termed multifrequency US imaging (MFUI). The Vantage research scanner (Vantage NXT 256, Verasonics Inc) was used to generate 24 transmit frequencies spaced along the bandwidth of the transducer from 4.71 to 10.4 MHz. Received data were filtered using a set of Gaussian filters with center frequencies corresponding to the transmit frequency. Increasing the number of pulse cycles from 1 to 3 had the effect of narrowing the frequency bandwidth of the return signal, which provided improved frequency sensitivity and reduced overlap between each transmission. The filtered radiofrequency data was converted to B-scan US images using the Hilbert transformation along with logarithmic scaling. This produced 24 B-scan US images from which the 2D attenuation map is generated pixel-by-pixel using the linear relationship between logarithmic image intensity and frequency. Estimations were performed using a multi-purpose US phantom (Model 040GSE, CIRS Inc, Norfolk, VA) that had two regions with known attenuation coefficients of -0.70 or -0.95 dB/cm/MHz. An ROI was defined for each. Overall, MFUI was used for US attenuation coefficient estimation and results were accurate and within 0.06 dB/cm/MHz of known values.