Toronto Metropolitan University
Browse
d7632b01aa9498eab7db776b967f4103.pdf (3.16 MB)

Decorrelated Compounding in Ultrasound Images

Download (3.16 MB)
thesis
posted on 2024-06-18, 19:29 authored by Na Zhao

This dissertation introduces the decorrelated compounding methods in synthetic transmit aperture (STA) ultrasound imaging and the spatial frequency domain of beamformed ultrasound images. They improve the detectability of low-contrast lesions in terms of lesion signal-to-noise ratio (lSNR), and visual detection. First, a decorrelation procedure was applied to traditional spatial and frequency compounding in STA to improve the lSNR in this dissertation. The decorrelated compounding method shown a better performance of speckle reduction than the conventional incoherent compounding methods at the cost of spatial resolution loss. The overall effect in terms of lSNR, which considered both speckle reduction and spatial resolution loss, indicated that the DC in STA outperformed the Delay-and-Sum (DAS) method. Then, we proposed to apply a two dimensional low-pass filter in the aperture domain to suppress the artifacts caused by the off-axis signals of DC in STA. In clinical applications, strong off-axis signals can be encountered such as irregular borders, calcification or development of vascularity. Both simulation and experiment images demonstrate the effectiveness of the filter. Lastly, the principle of decorrelated compounding was extended beyond STA to the any radiofrequency ultrasound images. The spatial frequency spectrum of beamformed ultrasound images was divided into overlapped sub-domains to generate sub-images for decorrelation and compounding. This method improved lSNR over the DAS method. In addition, the computational complexity was reduced by a factor of 16 compared to DC in STA. This dissertation investigate these decorrelated compounding based methods with the goal of improving the detectability of low-contrast lesions.

History

Language

eng

Degree

  • Doctor of Philosophy

Program

  • Biomedical Physics

Granting Institution

Ryerson University

LAC Thesis Type

  • Dissertation

Thesis Advisor

Yuan Xu

Year

2022

Usage metrics

    Biomedical Physics (Theses)

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC