Date of Award

Summer 2017

Document Type

Dissertation

Degree Name

Doctor of Philosophy in Mechanical Engineering - (Ph.D.)

Department

Mechanical and Industrial Engineering

First Advisor

Zhiming Ji

Second Advisor

I. Joga Rao

Third Advisor

Pushpendra Singh

Fourth Advisor

Bernard Koplik

Fifth Advisor

Cong Wang

Abstract

High Intensity Focused Ultrasound (HIFU) is becoming a widely accepted modality for extracorporeal non-invasive hyperthermia and surgical procedures. Since ultrasonic transducers need to operate in various challenging body locations, the arrangement of their array elements can be optimized to improve the capability of controlling focus intensity. In the first part of this dissertation, patterns of pressure field variations with several selected design variables (kerf, transducer element’s number and element’s width-height) are studied. These patterns indicate that there is a more suitable shape and arrangement of transducer elements in a specified area to achieve highest possible pressure. In order to obtain this arrangement, a Genetic Algorithm (GA) based evolutionary global search method is used to optimize the design shape and the distribution of ultrasonic transducer elements that can deliver maximum pressure at the focus zone.

This dissertation also presents a fast estimation model of focus ultrasound simulation from phased array transducer. Many simulation models have been developed to provide important information on the interactions between ultrasound beam and biological tissues as well as predictions of focused beam pattern. One of the commonly investigated issues in HIFU simulation is the calculation speed and most of the numerical models require considerable amount of time (minutes to hours) to finish one set of simulation in biological media. In the development of a fast estimation model of pressure-temperature response to support HIFU treatment planning, a numerical simulation model, known as Rayleigh-Sommerfeld method, is used. As the Rayleigh-Sommerfeld method is applicable only with homogenous media, a modified computation method that can deal with scattering and refractions from multiple tissue layers is developed to simulate the pressure field at different focus distances. A profile for prediction of maximum output pressure, power and temperature rise is then generated by using a standard Gaussian function and a Genetic Algorithm. The optimized form of prediction model function is adopted as estimation models for different tissue layers and geometric arrangements.

The average percentages of error found in homogeneous (liver) media for maximum pressure, power deposition and temperature with the fast estimation model are 0.10%, 0.20% and 0.25%, respectively, when compared with the Rayleigh-Sommerfeld method. When compared with the Angular Spectrum method, these errors are 0.50%, 1.00% and 0.77%, respectively. For heterogeneous viscera, kidney and pancreas tissues, average percentages of error in pressure estimation compared to Rayleigh-Sommerfeld method are 0.10%, 0.05% and 0.14%, respectively, and compared to Angular Spectrum method these errors are 1.83%, 1.72% and 0.76%, respectively. Average model error for maximum power deposition and temperature rise are also found to be within 1% in heterogeneous media. The methodology of this estimation model can significantly reduce the calculation time for numerical simulations. A graphical user interface program is integrated with the model to provide interactive visualization of the pressure-temperature responses at focus zone and hot spot locations.

Share

COinS