Quantitative Ultrasound Imaging of Thermal Ablation Therapy in the Liver.

摘要

The purpose of this dissertation is to investigate methods for delineating the extent of the thermally coagulated region following thermal ablation therapy of malignant tumors in the liver. In this dissertation, we examine quasistatic elasticity imaging for delineating the extent of the treated region. Next, we characterize scattering in normal, untreated liver and thermally coagulated liver. Finally, we examine attenuation estimation with reference phantom methods in the liver.;It has been well established that the Young's modulus of thermally coagulated tissue is much larger than in untreated liver tissue, making elastographic imaging an attractive candidate for delineating the extent of thermal coagulation. In quasistatic elastographic imaging of superficial organs such as the breast, external compression may be performed with a compressor plate or the transducer itself. However, in abdominal imaging, external compression is not effective. In response, our laboratory has developed electrode displacement for strain and modulus imaging. In electrode displacement strain imaging, deformation is applied by displacing the end of the RF electrode or microwave antenna by a small amount following treatment. We present electrode displacement strain images of thermally coagulated regions in in vivo porcine animal models during open surgery. We show that cross-sectional areas of treated regions delineated on electrode displacement strain images correlate well with coagulated areas determined on optical photographs of ablated regions of the excised livers. Additionally, we present electrode displacement strain images of VX2 tumors in rabbit animal models prior to and following thermal ablation.;Next, we develop mean scatterer spacing (MSS) estimation in the liver. In the liver, it has been hypothesized that periodic scattering results from a unique arrangement of microvasculature. Indeed, the liver is a unique organ because of its dual vascular supply from the hepatic artery and portal vein. We first show that utilizing a multi-taper spectral calculation results in a lower mean square error in MSS estimates using simulations of periodic tissue. Next, we show that average spectral coherence decreases in thermally coagulated liver compared with untreated liver in a small number of ex vivo RF ablations. Finally, we demonstrate that thermally coagulated regions are best modeled by aperiodic collections of scatterers, while normal liver exhibits periodicity.;In the final section of the dissertation we consider attenuation estimation. This is an attractive area of research because the attenuation coefficient is approximately doubles in the liver following thermal ablation. In past works, authors have developed attenuation algorithms with the assumption that the tissue being imaged may be modeled by a collection of a large number of randomly distributed scatterers. We show that in ex vivo bovine liver the scatterer number density is small. We demonstrate that the effect this has on attenuation estimation is to significantly increase the variance in attenuation estimation relative to tissue that may be modeled with a large scatterer number density. Using computer simulation and phantom experiments we show that when the scatterer number density is sufficiently large it is possible to achieve low variance, high resolution attenuation estimates. However, as the scatterer number density decreases either spatial resolution in the attenuation estimate must be sacrificed or attenuation estimate variance will increase. We find a linear attenuation coefficient in thermally coagulated liver that is approximately double the attenuation coefficient of normal, untreated liver. However, we also find a low scatterer number density in these tissues and a correspondingly high attenuation estimation variance.