Computational methods for contrast-enhanced intravascular ultrasound sequence analysis

Intravascular ultrasound (IVUS) is an invasive imaging modality capable of providing high-resolution, cross-sectional images of the interior of human blood vessels. Although its history of use has so far spanned over two decades, it still remains the most commonly-available and widely-used technology for obtaining detailed, in vivo imagery of normal and pathological vessels. While IVUS excels at structural imaging, its uses for functional imaging remain less advanced. However, modern ultrasound contrast agents present the potential to improve IVUS in this respect. These agents consist of solutions of suspended microbubbles or other particles designed to be efficient reflectors of acoustic energy. Due to the small scale of these particles (i.e., approximating the size of a red blood cell), they may be used to track blood perfusion under ultrasound, providing useful indicators of flow and vascular anatomy. They may also be designed to have an affinity to specific molecular signatures, "illuminating" diseased areas of tissue. Unfortunately, a wide variety of confounding factors with respect to microbubble detection have so far prevented IVUS imaging from taking full advantage of these applications. These problems include eliminating the severe motion artifacts caused by the beating heart and sensor ego-motion, isolating the contrast effect from irrelevant background, and reliably tracking the elastic tissues imaged by IVUS. This dissertation presents computational methods designed to systematically address these problems. To demonstrate clinical feasibility, also presented are the first published in vivo IVUS analyses of plaque perfusion in human patients and swine. The potential for imaging the vasa vasorum microvasculature is also discussed.