Finite Element simulation of methods for improving tissue ultrasound backscatter coefficient estimation

Abstract

This thesis will describe the efforts to generate Finite Element simulations of methods to improve the accuracy and reliability of backscatter coefficient methods (BSC). BSC measurements are a promising diagnostic tool in tissue characterisation, but are limited by corrections for attenuation and diffraction. To investigate this, simulation models were developed and analysed to investigate the sources of variability in BSC estimation,to make it a more clinically applicable tool for the investigation of tissue state. Computational models mimicking attenuation of ultrasound by soft tissue mimicking materials were shown to accurately reproduce the frequency dependent attenuation coefficients of the materials, providing a tool with which attenuation corrections can be generated in silico. In addition, the development of mathematical and simulation methods were shown to generate reliable and accurate simulations of BSC estimation. Results of the subsequent analysis revealed how the diffraction correction affected the quality of BSC estimates under different conditions. In addition, an algorithm was developed to segment backscattered echoes based on their spatial wave coherence. This algorithm was shown to be capable of segmenting coherence outliers embedded in incoherent scattering media, improving the resulting BSC estimate through omitting regions of a simulated tissue mimicking material that did not align with the conditions required for accurate evaluation of the BSC.