Department of Electrical and Computer Engineering Ph.D. Public Defense

Shear Wave Elastographic Imaging of Pancreatic Cancer Murine Models

Rifat Ahmed

Supervised by Professor Marvin Doyley

Tuesday, March 17, 2020
11:30 a.m.–12:30 p.m.

Goergen 109

The mechanical environment of a tumor directly impacts its growth, malignancy, and drug uptake. Imaging techniques that can visualize the mechanical properties of a tumor (stiffness, pressure, density etc.) can facilitate the diagnosis, treatment planning, and development of new therapies. Unfortunately, most parameters of tumor biomechanics cannot be estimated noninvasively, limiting their use in clinic. Although stiffness of a tumor can be quantitatively mapped in a noninvasive manner, it is not an established imaging biomarker. To validate stiffness as a useful imaging biomarker, naturally- occurring or therapy-induced changes in stiffness need to be established in different types and stages of tumors. Since performing these studies in the clinic are time consuming and expensive, evaluation of tumor stiffness in pre- clinical small animal models can greatly improve our understanding of the tumor mechanical environment. However, shear wave elastography (SWE), the imaging technique that is used to map stiffness, often has inadequate spatial resolution in the preclinical domain.

SWE uses transverse acoustical waves (shear waves) generated from acoustic radiation force to map the shear wave speed (SWS), which is directly related to the shear modulus of tissue. The propagation of the shear waves are tracked using conventional ultrasound imaging. However, speckle in the ultrasound images introduces a positional uncertainty in the wave-tracking process, which corrupts the SWS maps with noise. This noise is severely exacerbated at small observation scales, which is typical for preclinical imaging. Consequently, larger observation scales (often > 1 mm) need to be used, which degrade the spatial resolution of SWE. Single-tracking-location SWE (STL-SWE) is a recently proposed technique that overcame this fundamental challenge in speckle tracking by using a fixed tracking location. While STL-SWE is a suitable modality for preclinical imaging, there are challenges that need to be overcome. STL-SWE is not immune to incoherent sources of noise under in vivo imaging conditions. Furthermore, the longer acquisition duration of STL-SWE makes it susceptible to physiological motion artifacts.

The objective of this thesis is to develop a preclinical SWE technique that can be used to study the therapy response and natural progression of the tumor mechanical environment in small animal models. To achieve this objective, we developed beamforming approaches to reduce the estimation variance of SWE. We developed an approach to trans-mit multiple encoded-apertures simultaneously, which enhanced  the transmit focusing quality of ultrasound beams and the signal-to-noise ratio. Next, we improved the noise robustness of STL-SWE by using coherently compounded plane wave imaging, a technique we named pSTL-SWE. We, then, extended pSTL-SWE to focused transmit-based parallel beamforming techniques that are applicable to commercially available clinical ultrasound scanners. We also developed a real-time automated respiration gating scheme that enabled us to use pSTL-SWE in preclinical mouse models of cancer. Using this technique, we studied the progression of liver stiffness longitudinally in a murine model of pancreatic ductal adenocarcinoma (PDAC) liver metastasis. We observed, for the first time, a direct correlation between elevated tumor stiff- ness and survival of untreated and chemotherapy-treated mice. We also used the pSTL-SWE technique to validate a recently developed immunotherapy treatment protocol for PDAC. The results presented in this thesis demonstrate the effectiveness of pSTL-SWE in imaging preclinical cancer models and represent a critical first-step toward establishing stiffness as a clinical imaging biomarker for cancer.