Model-Based Ultrasound Imaging for Challenging Acoustic Clutter Suppression
Ultrasound is one of the most widely used imaging modalities in medicine, with an excellent safety record. Ultrasound is attractive because it is real-time, accessible and affordable compared to other modalities. Based on these characteristics, ultrasound has become a common tool for both diagnosis and therapeutic guidance in a clinical setting. However, image artifacts frequently encountered in clinical ultrasound are problematic and impair its usefulness, especially when scanning patients who are obese. Obesity is a global epidemic affecting more than half a billion adults worldwide in 2008, an increase of over 100% since 1980. In the United States, 34.2% of adults were obese in 2012. Obese patients possess a large amount of fat in the subcutaneous layer of skin, which is where aberration and reverberation occur. Recent studies have identified aberration and reverberation as primary sources of ultrasound image artifacts. These artifacts severely degrade ultrasound image quality and may lead clinicians to misinterpret an image and obscure diagnosis and therapeutic guidance. To mitigate such artifacts, numerous ultrasound beamforming methods have been developed, including our model-based beamformer called aperture domain model image reconstruction (ADMIRE). The basis of ADMIRE is a decomposition of received aperture domain signals using a physical model followed by reconstruction using only model signals from the region of interest. A major aim of this dissertation is not only to investigate ADMIRE’s ability but to also identify its limitations in suppressing ultrasound acoustic clutter sources, including reverberation, off-axis scattering, wavefront aberration and gross sound speed mismatch. ADMIRE has demonstrated an ability to suppress acoustic clutter and significantly improve ultrasound image quality in challenging high clutter environments. However, we identified some limitations to ADMIRE’s performance and efficiency. This dissertation also reports proposed solutions to address the identified limitations.