Date of Award

Winter 12-15-2017

Author's School

School of Engineering & Applied Science

Author's Department

Biomedical Engineering

Degree Name

Doctor of Philosophy (PhD)

Degree Type

Dissertation

Abstract

Ultrasound computed tomography (USCT) and photoacoustic computed tomography (PACT) are two emerging imaging modalities that have a wide range of potential applications from pre-clinical small animal imaging to cancer screening in human subjects. USCT is typically employed to measure acoustic contrasts, including the speed of sound (SOS) distribution, while PACT typically measures optical contrasts or some related quantity such as the initial pressure distribution. Their complementary contrasts and similar implementations make USCT and PACT a natural fit for a hybrid imaging system. Still, much work remains to realize this promise. First, USCT image reconstruction methods based on the acoustic wave equation, known as waveform inversion methods, are computationally burdensome, limiting their widespread use. Instead, image reconstruction methods based on geometric acoustics are often employed. These methods do not model higher-order diffraction effects and consequentially have poor resolution. In this dissertation, use of a novel stochastic optimization method, which overcomes much of the computational burden of waveform inversion, is proposed. Second, most traditional PACT image reconstruction algorithms assume a constant SOS distribution. For many biological applications, this is a poor assumption that can result in reduced resolution, reduced contrast, and an increase in the number of imaging artifacts. More recent image reconstruction algorithms can compensate for a known heterogeneous SOS distribution; however, in practice, the SOS distribution is not known. Further, in general, the joint reconstruction (JR) of the SOS and initial pressure distributions from PACT measurements is unstable. Two methods are proposed to overcome this problem. In the first, a parameterized JR method is employed. Under this approach, the SOS distribution is assumed to have a known low-dimensional representation. By constraining the form of the SOS distribution, the JR problem can be made more stable. In the second method, few-view USCT measurements are added to the PACT data, and the initial pressure and SOS distributions are jointly estimated from the combined measurements. This approach effectively exploits acoustic information present in the PACT data, allowing both the initial pressure and SOS distributions to be more accurately reconstructed.

Language

English (en)

Chair

Mark A. Anastasio

Committee Members

Joseph P. Culver, Joseph A. O'Sullivan, Lihong V. Wang, Quing Zhu,

Comments

Permanent URL: https://doi.org/10.7936/K74T6HSJ

Share

COinS