Averkiou, MichalakisHammond, Ryan2020-08-142020-08-142020-08-142020Hammond_washington_0250O_21855.pdfhttp://hdl.handle.net/1773/45841Thesis (Master's)--University of Washington, 2020In ultrasound imaging, an abrupt change in medium, such as the interface between tissue and bone can have a dramatic effect on the image. A bone-tissue interface strongly reflects acoustic signals in comparison to soft tissue. Large amplitude scatterers can lead to artifacts spreading across adjacent regions of the tissue during the spatial filtering of beamformation. The following work investigates three different approaches to minimize the influence of large amplitude off-axis echoes, while minimally impacting regions of smaller amplitude echoes. We have observed thresholding the large amplitude signals before beamformation can prevent side lobe artifacts into lower amplitude echoes. However, the selection of thresholds too low can impact the signal of interest. Ideally, a threshold would be as high as possible to not impact signals of interest, but still adequately remove artifacts. In order to address this limitation of a simple threshold we have investigated the use of a neural net to adaptively select an optimal threshold. Lastly, we have also investigated the use of adaptive selection of aperture apodization to further suppress high amplitude off-axis signals. Specifically, we image spinal cord blood flow and changes in blood flow following spinal cord injury in a rodent model. The lamina or top of the vertebral bones are surgically removed for injury permitting ultrasound imaging. With the spinal cord resting on top of vertebral bone, large amplitude reflections from bone can contaminate surrounding weaker reflecting spinal cord tissue. These artifacts impede visualization of blood flow and further analysis in detecting changes in blood flow resulting from spinal cord injury.application/pdfen-USnoneBiomedical engineeringMedical imagingBioengineeringThesis