Image Processing Methods For Dual-Energy Contrast-Enhanced X-Ray Breast Imaging

Kristen Lau, University of Pennsylvania

Abstract

Imaging of the breast is essential for the early detection and diagnosis of breast cancer. Digital mammography (DM) and digital breast tomosynthesis (DBT) are the current standards of care for breast cancer screening. However, these imaging modalities are insufficient for screening women who have a higher-than-average risk for developing breast cancer. Contrast enhanced (CE) breast imaging improves the detection and diagnosis of breast cancer by combining morphologic and functional information on vascular kinetics in a single examination. The limitations of anatomic imaging led us to look towards dual energy (DE) CE x-ray breast imaging, which uses an iodinated contrast agent to produce images of the breast vasculature. This dissertation explores image processing methods for DE CE x-ray breast imaging. We developed a hybrid image subtraction method together with 2D and 3D registration methods for DE imaging. The hybrid subtraction method was proven to be robust and was successful in processing both clinical and animal images. Emanating from our work on image subtraction, we developed a framework to characterize DE imaging physics and developed a calibration method to determine breast thickness and tissue composition from DE images. The DE weighting factor used in image subtraction is thickness-dependent, and thus the DE signal in an image intrinsically encodes breast thickness information. Currently, CE breast imaging uses iodinated or gadolinium-based contrast agents. The adverse reactions and disadvantages of these contrast agents have provided the motivation for the search and development of improved imaging contrast agents. The development and evaluation of numerous silver based nanoparticles for DE imaging was explored and imaged. Silica-encapsulated silver nanoparticles, gold-silver alloy nanoparticles, an “all-in-one” silver sulfide nanoparticle, and silver telluride nanoparticles were imaged in phantoms and mice using CE digital mammography (CE-DM). These silver-based nanoparticle contrast agents were found to produce a DE signal that was comparable or stronger than that of iodine when imaged under clinical imaging protocols. Silver sulfide nanoparticles were also imaged using photon-counting mammography. It was demonstrated that these particles could be imaged at lower mammographic energies while retaining comparable image quality when compared to conventional mammography.