Quadratic B-mode and pulse inversion imaging of perfusion defects in vivo

Abstract

In this paper, a comparison of quadratic B-mode (QB-mode) and pulse inversion (PI) imaging for characterization of perfusion defects in heat treated tumor model in vivo is presented. SCK mammary carcinomas grown in A/J mice were treated with 125 or 250 μg/kg CYT-6091 (formerly known as PT-cAu-TNF) followed by local heating at 42.5°C using a waterbath for 60 min, 4 hrs after nanoparticle injection. Ultrasound imaging was performed on days 1 and 5 after treatment using a modified Technos MPX system from ESAOTE S.p.A. (Genoa, Italy). A linear array probe (LA522E, 2-cycle transmit pulse centered at 5.5 MHz) was used to collect data in PI mode at 26 fps. The contrast agent, BR14 (Bracco Research S.A., Geneva, Switzerland) which is a new experimental agent consisting of high molecular weight perfluorobutane gas bubbles (2 μm diameter) encapsulated by a flexible phospholipids shell and suspended in saline was administered by the tail vein at a concentration of 0.0025 mL/kg of body weight. QB-mode images were obtained from the quadratic components from an adaptive second order Volterra filter (SoVF). A recursive least square (RLS) algorithm was used to estimate the quadratic kernel of the SoVF. PI images were formed in the usual manner by summing the echoes from the positive and negative transmit pulses. The RF echo data from the positive transmit pulse was used as input to the SoVF to extract the quadratic component. Results: PI images (typically 40 dB dynamic range) showed heterogeneous perfusion profile in the treated tumor 1 and 5 days after the treatment. This was quantified by computing a temporal perfusion index (TPI) for each image pixel from a sequence of 126 frames. QB-mode images (typically 70 dB dynamic range) showed similar heterogeneity in the perfusion pattern. However, the results illustrate an important potential advantage of this imaging mode. Specifically, QB-mode images have high dynamic range with reduced speckle with high sensitivity and specificity to nonlinear oscillations from UCA. Conclusion: QB-mode images have shown the same level of sensitivity to perfusion defects as PI images. QB-mode images do not require multiple pulse transmission thus they are not vulnerable to motion artifacts. In addition, they achieve higher specificity due to speckle reduction and noise suppression.