Accurate High-Resolution Tissue Characterization for Breast Cancer Screening Using Transmission and Reflection Ultrasound Tomography

PROJECT SUMMARY/ABSTRACT
Breast cancer is the most widely diagnosed cancer in women in the United States and the most commonly
occurring cancer worldwide according to the World Health Organization. The primary screening method for non-
palpable breast cancers is X-ray mammography, which uses ionizing radiation and often has low sensitivity and
specificity in dense breast tissue. Ultrasound tomography (UST) offers a radiation- and compression-free
alternative to X-ray mammography that can image the breast based on its acoustic properties and enable early
cancer detection and diagnosis with greater sensitivity and specificity, especially in dense breast tissue.
Currently, UST imaging of the breast relies on the transmission of ultrasound through the tissue to reconstruct
underlying tissue properties such as sound speed and attenuation. However, the full-waveform inversion (FWI)
algorithm used to reconstruct these tissue properties is susceptible to false local minima because of a
nonconvexity in FWI known as cycle skipping. Additionally, the slicewise imaging approach currently used in
UST limits the ability to correct for 3D out-of-plane scattering. Finally, because FWI reconstruction in UST is
primarily based on the transmission of ultrasound through tissue, little attention has been given to exploiting
reflected signals during FWI. Each of these shortcomings prevents FWI from adequately modeling high-
frequency signals, ultimately limiting the currently achievable imaging resolution of UST. Therefore, this work
proposes to enable high-resolution UST imaging by (1) mitigating cycle skipping effects, (2) capturing
the 3D insonification of the tissue, and (3) incorporating reflected signals into FWI.
In order to expand the capabilities of UST imaging, I aim to: overcome cycle-skipping effects that limit
the usability of high-frequency signals in UST by reformulating FWI (Aim 1); design and build a UST imaging
prototype that can optimally capture and utilize the 3D insonification of the tissue to reconstruct sound speed
and attenuation with greater accuracy and resolution (Aim 2); and incorporate reflected signals into FWI to
recover sharp boundaries in the sound speed image caused by impedance changes in the tissue (Aim 3). These
aims will improve the scientific understanding of acoustic models applicable to ultrasound signals in both
transmission and reflection, as well as how these models may be inverted to reconstruct accurate and spatially
resolved images of tissue properties. Expanding the tissue characterization capabilities of ultrasound
tomography will enable a multi-parametric approach for the detection and characterization of breast cancer using
an imaging modality without ionizing radiation. Improving the robustness of UST in acoustically challenging cases
will also enable new imaging applications such as small-animal and human-transcranial imaging. The proposed
work will also broadly benefit the field of ultrasound imaging by improving tissue modeling and characterization
capabilities and enabling the early detection of disease based on acoustic tissue properties.

Grant Number: 1K99EB037080-01
NIH Institute/Center: NIH
Principal Investigator: Rehman Ali