UC Irvine

Elastography is of great interest in the fields of solid mechanics and biomechanics due to its nondestructive capability of mapping elasticity of materials and tissues. The elastography framework relies on external excitations which stimulate deformation inside an object. The internal response is then acquired and analyzed to map the distribution of elastic moduli. The first method developed in this dissertation is that, with no need of measuring any internal responses, an elastography method integrated with tomography is proposed, only requiring the transmitted responses of applied sound waves. During the process, the tomography image (e.g., CT or MRI) and the applied waves are integrated into a computational model. Following the principle of factorial design, elastic distribution of all phases in the object is reconstructed when the computational transmission of waves matches with the measured transmission. As an improved algorithm to the integration method, in the dissertation, deep convolutional neural networks (CNNs) are studied for mapping elastography with much less computation time. A CNNs architecture is developed, considering the contribution of raw features.

In the dissertation, another elastography method is developed by untangling the complex wave-induced strain field into the one due to pure compressional or shear disturbance. The proposed untangling method is realized according to the fact that the volumetric strain is caused by compressional waves. By transforming both the volumetric strain tensor and the general tangled strain tensor to the shear wave direction, the transient vibration velocity and strain generated by compression wave can be separated from their initial coupled fields and used to reconstruct elastography.

Nondestructive ultrasound-based methods have been applied to evaluate the elastic properties of composite materials. While the wave modulus of elasticity is frequently reported higher than the static counterpart, the microstructural and physical mechanisms are not well understood. In the dissertation, a computational micromechanics is conducted to investigate the effects of inclusions on both the effective wave modulus of elasticity and static modulus of elasticity. Taking concrete as an example, based on concrete microstructures resolved with X-ray micro-tomography. It is demonstrated that the existence of void defects plays a significant role on the elastic properties of concrete when compared with the particles that are also called aggregate. It is shown that the higher wave modulus of elasticity of concrete than the static one is caused by the existence of crack-like voids with small aspect ratios.